TITLE: Internet access to ecological information - the US LTER All-Site Bibliography Project AUTHORS: Harvey Chinn, Caroline Bledsoe DATE: 1997 BioScience v47, n1 (Jan, 1997):50 (8 pages) Internet access to ecological information - the US LTER All-Site Bibliography Project Biologists who wish to conduct a comparative or synthetic analysis of a particular topic face a daunting task. Typically, this endeavor starts - and may end - with a review of the literature. If it proceeds further, experimental data must be collected from a variety of sources, generally from published papers. Data must then be converted into comparable units, stored, and analyzed; the last step is to derive and publish one's conclusions. Until recently, this process required trips to the library; letters, e-mail, and phone calls to other scientists; and possibly a significant investment of time in computer software to assist with the integration of the gathered information. Recent developments on the Internet are changing the way this process can work, making it possible to deliver access to bibliographies, literature, data, and other analyses directly to the desktop computers of biologists. Increasing amounts of biological information are now available online using a variety of network software tools of varying sophistication (e.g., gopher, Wide Area Information Systems [WAIS], Mosaic, Netscape, and others). These tools are rapidly becoming more and more "user friendly" and powerful, suggesting that online systems will be even more valuable in the future. Thus, biologists are likely to face an ever-increasing storehouse of electronic information. At present, however, this storehouse is somewhat disorganized, its contents often not arranged in the manner desired by the biologist or in forms that are easy to use, especially for comparative studies. Evans (1994, p. 205) has described the current organizational state of the Internet in terms of information "deserts" and "oceans": "Not knowing where to find useful information: the user 'thirsts' in an 'information desert.' In other cases...no systematic directories exist to guide a user through the thousands of data sources: the user 'drowns' in an 'information ocean.'" There are several general guides to the Internet, such as Gaffin and Heitkotter's (1994) and Krol's (1994). Hayes (1994) presents an overview of the World Wide Web for scientists. Smith (1993) offers an Internet guide specifically for biologists. An online starting point for finding sources of information on the Internet arranged by subject is the World Wide Web Virtual Library (Secret no date). The US Long-Term Ecological Research (LTER) Program (Callahan 1984, Franklin et al. 1990; see box page 51) is actively working in the arena of Internet access to ecological information (LTERnet 1996d) - including bibliographies, directories of scientists, and a variety of data. The Ecological Society of America (ESA) is investigating such Internet access as a supplement to traditional scientific publication (Colwell 1995). This article describes our efforts at supporting the first component of the comparative and synthetic process for the LTER Network - the literature search. From these efforts, we have learned many lessons about how groups can consolidate, restructure, and provide online access to enhanced ecological information for use by others. Our goal was to create a unified, multisite bibliography from the individual bibliographies of the 18 LTER sites that make up the LTER Network (LTERnet no date c) and of one former site. Each site in the network keeps a bibliography of publications on the research done at the site. Although the individual site bibliographies are useful, simultaneous access to all of these bibliographies would greatly increase the value of a search. Bibliographies are inherently simple data sets, having a standardized structure for the citations (e.g., author, date, title, and journal name). Thus, one might naively expect that combining several bibliographies into a coherent "all-site" bibliography would be a simple task. However, we discovered that local site differences make this integration more complicated than expected, especially if there is also the broader goal of joining the combined bibliography into a larger data access system, which might include a personnel directory listing authors and scientists and a catalog of experimental research data. Because these complications exist for simple, well-defined data like bibliographies, the problems will surely be greater with the more complex and subtle data such as collections of ecological experimental observations with extensive metadata. We chose a simple, reasonably well behaved data set as a small-scale test case to make finding solutions more probable and climbing out of the inevitable pitfalls easier. We wanted to learn how current data - and, more important, future data not yet collected - might be structured to ease the task of consolidating collections of other types of information. Thus, the primary value of our test case is what we have learned about how to assemble, structure, and store data for online access. Despite the difficulties, our experience shows that the added value of the combined, unified data set more than justifies the efforts needed to achieve the consolidation. We also learned that those who publish online data should make significant efforts to do so in a manner that facilitates, rather than impedes, the later creation of enhanced collections of information. Based on our experiences constructing and maintaining the LTER All-Site Bibliography, we offer several observations on providing online access to ecological information. The project's goal The long-term goal of the All-Site Bibliography is to allow researchers to conduct sophisticated computerized searches of the bibliographic citations for all LTER sites via the Internet. We have achieved a limited implementation of this goal using gopher (no date) and WAIS (1995), the techniques widely available in early 1993. We created a centralized All-Site Bibliography (LTERnet 1996a) located on an Internet-accessible computer at the LTER Network Office (currently at the University of Washington); details are described below. Recent developments in Internet client-server technology permit the full implementation of the project goal. These developments allow information systems, such as World Wide Web (WWW or Web) servers, to query relational database management systems using standardized techniques. They also allow custom solutions to be used with Web servers. Another likely change will be the conversion from the centralized form of the bibliography to a distributed database system in which each site maintains a local version of its own bibliography in a common format. In such a system, the user would still be able to search all the sites' bibliographies in a single operation. The All-Site Bibliography This section covers the design and construction of the All-Site Bibliography and the lessons learned about developing online data access systems. Development. With the assistance of the LTER data managers (LTERnet no date d), we obtained text versions of 19 site bibliographies, including one from a former LTER site (see Table 1 for a list of sites). The individual sites have developed and are maintaining their bibliographies in several different forms (Table 1). At present, this variety of software includes word processors (four sites), custom software (two sites), general purpose database programs (four sites), and full bibliographic database systems (nine sites). Most of these software systems store information in binary files, with the notable exceptions of those used by the Palmer Station (Antarctica) and Sevilleta (New Mexico) LTER sites. Palmer stores bibliographic records in the UNIX-based BibIX bibliographic system, a readable text format (not to be confused with BibTeX). Sevilleta uses "/rdb," a similar system. Neal (1993) and Walker and Choate (1994) review bibliographic software. Nearly all database programs and word processors provide a method for "saving" (writing) the information from a database or a word processing document to another file as plain text. We wrote a set of programs that rearrange these different text versions into the format used for the online system. Each bibliographic citation, or record, comprises many types of information, such as author(s), date, title, journal name, volume number, publisher, and abstract. This information is more useful, at least in a computer sense, when separated into distinct pieces, or fields. The format for the LTER Bibliography explicitly labels each field in the record. The LTER online format. Our initial task was to homogenize, as much as possible, the various formats of the saved text versions into a single, consistent display format, called the "LTER online format." Ideally, this format presents each type of information in a separate field. We used the awk programming language to write programs that filter the bibliographic text files. These filters rearrange the text into a consistent form and add standardized field labels. The awk language (similar to the Perl and C programming languages) is designed to do exactly this kind of text manipulation (Aho et al. 1988). We followed three guidelines in developing the LTER online format, the basic form for which was proposed by Cohen (1992): * The format should be humanly readable and intelligible; there should be no mistaking what each field contains. Success depends on the balance between clarity and conciseness in choosing the field labels. * The format should also be machine readable (i.e., capable of being processed by a computer program). This factor will aid in importing selected citations into other (bibliographic) database systems. * A good machine-readable format will also support the enhancement of the bibliography through the addition of improved, more sophisticated searches as new software is developed. [TABULAR DATA FOR TABLE 1 OMITTED] Construction difficulties. The extraction of each field from a record for any of the various bibliographic database packages used by the LTER sites is easy because these programs encourage the fine subdivision of a citation into its constituent parts. For example, ProCite has approximately 30 fields available for partitioning the information from various types of bibliographic citations such as books, journal articles, computer programs, and maps; BibIX employs a similar number of fields. On the other hand, not all programs or their users divide the information in the same way or to the same degree, creating inconsistencies when site-based systems are combined into a standard "all-site" format. For the site bibliographies that are maintained with word processing software, it is difficult, if not impossible, for a program to identify and extract each type of information because all the information is lumped into a single field (a formatted paragraph). The site bibliographies created in general-purpose database systems are intermediate in structure; their information is partially lumped. If lumped information is always arranged in the same order, then it is usually possible to program the filter to separate it into most of its constituent parts. Unfortunately, we found that the records in a given bibliography, especially those kept in word processors and general-purpose database systems, rarely have a regular, self-consistent structure. There was frequently inconsistency in the way in which the same type of information was entered across multiple citations, making it harder to filter the records. An even more difficult problem is correcting mistakes, such as typographic errors and entry of information in the wrong field. Thus, the All-Site Bibliography is not completely uniform due to the variety of formats, the degree to which some site "lump" data, or inconsistency in the way the information is structured. Lessons learned. This project stimulated LTER sites to improve their site bibliographies. Following the example of the Coweeta site, three others sites (Konza Prairie, Niwot Ridge, and Palmer Station) began to add abstracts. Several sites added references for research work done before initiation of the LTER Program. Some sites upgraded the software used to maintain their bibliographies. The project also promoted the inclusion of the bibliography from a site no longer in the LTER Network (North Inlet, South Carolina), a valuable preservation of information. We have learned that some choices that work well at the local site level cause problems at the network level. The sites have used at least 16 different computer programs to maintain their bibliographies (see Table 1), which led to problems of consistency in creating the All-Site Bibliography. This diversity is inevitable in a network of sites. The principle difference we observed is how finely each citation is subdivided into distinct fields, which ranges from 1 field per citation for word processing software to 5-7 for general purpose database systems to 10-30 for bibliographic software. To combine separate bibliographies - stored in disparate formats - into a single, self-consistent format for the purpose of unified access, someone must devise "filters" (computer programs) to handle the differences. As scientists in the LTER Network increase their use of integrated online computer facilities, individual sites must exercise care to select software and design solutions appropriate for application in a networked organization. In addition, sites should design these systems to accommodate, rather than resist, the changes that inevitably occur in computer technology. In an organizational sense, this project succeeded in large part because, as LTER's first attempt to integrate separate databases, we implemented the All-Site Bibliography as a centralized system, with all the software and data files collected together on a single computer. A small team did all the programming, assembly, and updating. However, our experience in managing the bibliography project over the last three years has made it amply clear that future "all-site" databases should be designed as distributed systems from the start, with each LTER site maintaining its portion of the particular database locally at the site, organized to facilitate integrated Internet access and searching (see the box "Centralized versus distributed database systems," page 54). The LTER site data managers' group supports this concept and has called for the creation of an LTER Network information system (LTERnet 1996c). Increasingly, the appropriate and satisfactory application of computers to the management of biological data requires a team effort. This is particularly true for data that may be used in comparative or synthetic work. We have advanced beyond the time when individual biologists could fully exploit the potential of computer systems solely through their own programming efforts. The experience of LTER and others shows that development teams producing software tools need both biologically literate computer scientists and computer-savvy biologists to achieve the best results. Current status of the All-Site Bibliography The LTER All-Site Bibliography currently contains 12,000 entries and is available for online use on the LTERnet Web server (LTERnet 1996a). It may also be downloaded as a set of text files, one for each site (LTERnet 1996b). Searching the All-Site Bibliography is presently implemented using gopher and WAIS. Unfortunately, the capabilities of the standard WAIS software are limited. For example, an ecologist might query: "Show me any works published since 1970 on plant roots in forests authored jointly by Andrews and Hubbard." Our ecologist is already stymied, because there is no straightforward way to specify the condition "since 1970" using standard WAIS. Let us put that problem aside. A standard WAIS search works by looking at every word in the entire citation. That is, a WAIS document is essentially a text file with no inherent structure (i.e., no fields). This means a gopher-WAIS search for "Andrews Hubbard roots forest" will not only return citations for the two authors but also all citations for the H. J. Andrews Experimental Forest (1200+ records) and all those for the Hubbard Brook Experimental Forest (1300+ records). Furthermore, there are three more sites with the word forest in their names, resulting in an additional 1200 citations. Although this example is exaggerated, it illustrates some of the main drawbacks of standard WAIS. These problems notwithstanding, use of the All-Site Bibliography has steadily increased from its inception at the beginning of May 1993. The number of accesses during the first five months of 1996 averaged 1420 searches per month [ILLUSTRATION FOR FIGURE 1 OMITTED]. By the end of September 1996, the number of searches totaled 34,119. Future directions for data management and online access Advanced techniques now allow a complete implementation of the All-Site Bibliography project. These include distributed searching of relational databases through the Web. However, the use of cutting-edge technology brings new difficulties, including increased expense in implementing these sophisticated data access systems. New techniques. The World Wide Web has a standard method for running arbitrary programs from a Web server that can access a provider's data in sophisticated or novel ways. This technique uses a type of program called a CGI script (Common Gateway Interface; see NCSA no date b), which connects the Web to external programs or collections of data that are not directly available on the Internet. One method for accessing external data is SQL (Structured Query Language), a standardized system that enables a program to ask complex questions of a relational database. Another method of connecting the Web to a provider's organized information is to use custom programs with Web servers to query that information, independent of either WAIS or SQL. Rather than use the formal standards defined for WAIS and SQL, a custom program will use its own techniques to access the data. Examples of custom programs are Chinn and Beardsley (1996) and Niwot (1996). SQL allows complex queries not possible with standard WAIS software. Let us revisit our ecological query: "Show me any works published since 1970 on plant roots in forests authored jointly by Andrews and Hubbard." In contrast to the WAIS version of this search, an SQL search can be constructed to look for the joint occurrence of the names Andrews and Hubbard only in the author field of the database; for root(s) and forest in the title, key words, or abstract fields; and for any date after 1970 in the date-of-publication field. Limiting the search in this fashion is not possible with standard WAIS. SQL can also couple separate databases for joint searching (e.g., all citations by author Smith, plus the personnel directory entry for Smith). However, the databases need to have common structural elements and the data must be organized properly to produce reliable results. For example, if a joint search has returned a citation by author J. Franklin, there must be an unambiguous way to distinguish between the entries for Janet Franklin and Jerry Franklin in the personnel directory. A Web technique that can be used with custom programs, WAIS, or SQL is the creation of graphical user interfaces using HTML fill-in forms (Hypertext Markup Language, see NCSA no date a; for forms, see NCSA no date c). The selections made by the user are sent from the forms interface to the external program via the CGI script, which processes them and returns a result, usually for display by the user's Web browser. Online examples include the Niwot Ridge LTER site bibliography, which uses a custom program to search the bibliography (Niwot 1996); the LTER Network Personnel Directory, which uses SQL to conduct the search (LTERnet no date b); and the ICE MAPS natural resources query system from the Information Center for the Environment, which uses custom software to query a geographic information system (GIS; Chinn and Beardsley 1996). Fortunately, the details of the SQL language (when it is employed) are made largely transparent to the user by these forms-based interfaces (e.g., LTERnet no date b). A distributed database system. Currently, the entire LTER All-Site Bibliography is housed on the gopher server at the LTER Network Office. For several reasons, it is desirable to convert to a distributed system, in which each site maintains a local, online version of its own bibliography in a standard LTER form. All of these individual bibliographies would be searchable via the Internet in a single operation. This system, although existing as separate, distributed pieces, would appear to the user as a single, integrated unit. The box "Centralized versus distributed database systems" (page 54) lists the advantages and disadvantages of these two systems. Converting to a distributed system would also relieve personnel at the LTER Network Office from the extra work involved in maintaining a single, centralized database. One beneficial result would be to streamline the process of updating the All-Site Bibliography, which currently requires action by personnel at both the local site and the Network Office. A distributed system would still allow a search of the remaining sites should one or more sites be temporarily offline. Under the present system, if the LTER Network Office computer system is down, the only LTER bibliographies available are the offline versions maintained locally at each site and those that are independently online at other sites. In a sense, this conversion is already underway. Three-quarters of the sites have posted their bibliographies on the Internet, and four of these are searchable (see Table 1). So far, however, these are all independent: each must be accessed separately. The conversion to a distributed system could be completed through the use of SQL-queriable database systems (SQL servers) because they support distributed searching. Internet tools and the cutting edge. When this project began early in 1993, the Web was new. Gopher and WAIS were well-established Internet tools, in widespread use, with a fair number of gopher servers specializing in biological topics (see Smith 1993 for a list). The ease of installation of gopher-WAIS servers meant they functioned like so-called plug and play software. When we began the project, we did not adopt cutting-edge technology because it is a double-edged sword. Although cutting-edge technology allows advanced, sophisticated presentation and access techniques, it also cuts out a potentially large portion of one's audience. For example, how many biologists had heard of the Web in 1993, let alone had a browser for using it? Our goal has been to provide the broadest possible Internet access to the All-Site Bibliography. The Web is now the presentation system of choice; early in 1996, we added a home page for the All-Site Bibliography to LTER's Web server (LTERnet 1996a). We expect to replace the gopher-WAIS search system with one that uses HTML forms and SQL. Today, Web HTML forms are no longer on the cutting edge. A recent development is Java, which allows highly sophisticated, platform-independent program code ("applets") to be downloaded with the data (Java 1995). This means that the program for using the data resides, temporarily, on the user's computer along with the data. However, not all browsers support Java, again reducing the number of potential users. We have adopted a policy of staying just far enough behind the cutting edge so the techniques we use provide a reasonable level of useful access for a broad audience. A major factor in considering the use of advanced techniques is their cost. Few, if any, satisfactory relational database programs, GIS, or SQL servers are free. As the Internet becomes increasingly commercialized, new versions of some formerly free Internet tools now come with a price tag. Moreover, implementing the cutting-edge techniques requires personnel with various combinations of specialized knowledge and programming skills, such as relational database, GIS, SQL, CGI scripts, HTML, and advanced Web server operation. In addition, the more advanced the technique, the longer it will take to implement the data access system, adding further to the cost. There has been a general progression in Internet techniques from the easily installed to the difficult to implement: * Gopher and WAIS take no programming to install, but they do require a modicum of computer skill for their installation. The basic format dispensed on a gopher server is plain text; plain text files are easy to create. * Web HTML forms and the CGI scripts underlying them can take a fair amount of programming, especially to query SQL databases or geographic information systems. The basic format dispensed on a Web server is HTML; HTML requires a knowledge of its syntax and proper application. For example, the ICE MAPS system took 350 hours of programmer time to design and implement (Chinn and Beardsley 1996). * Java may require even more complicated programming to provide access to data. The promise of the information superhighway. As the technical and legal kinks of providing access to computerized information are worked out, the usefulness and power of the online LTER system should increase dramatically. For instance, abstracts can immediately be added to the site bibliographies. Another step, after the copyright issues are settled, might be to make the entire text of each article available, along with any figures and photographs (a PostScript or Acrobat version of each article is a likely adjunct). For example, ESA may develop online access to the articles from its printed journals (Colwell 1995). Furthermore, ESA is considering the electronic publication of the data associated with each article (Colwell 1995). In the not-so-distant future, developments in Internet information systems are likely to allow complex searches that begin with a query like "Show me what has been published on [a particular topic]." After reading a number of abstracts, you would instruct the system, "Show me articles similar to items 4 and 27." After perusing the new materials, you might then decide to print or save to disk several articles, complete with tables, figures, and data. If you are interested in any work-in-progress that continues the published efforts, you would ask to see the related grant documentation. Should you wish to confer with any of the LTER researchers involved with the work, you would select one or more names. The system will present you with their entries from the LTER personnel directory, complete with e-mail addresses and phone numbers. Selecting an e-mail address will start up your favorite e-mail program. Alternatively, as telephones become integrated into computers, simply selecting a phone number will dial the individual. As mentioned earlier, some sites have their bibliographies online and searchable. In addition, many sites have online personnel directories and most have scientific data available on the Internet. The next step will be to link, in the hypertext sense, the bibliographies, data, and personnel directories. The Network Office plans to link the Network Personnel Directory (LTERnet no date b), the Core Data Set Catalog (or its successor; Ingersoll and Hartman 1995, LTERnet no date a, Michener et al. 1990), and the All-Site Bibliography, which are now available separately. This will be another step toward fulfilling the promise of the information superhighway for biologists. In this article, we have discussed the difficulties inherent in and the lessons learned from the process of unifying separate electronic data sets, and we have presented guidelines for the ecological community based on that experience. We believe that such unification, when done properly, has value that extends well beyond the group that does the work of consolidation. We hope that ecologists, and biologists in general, who publish data electronically will make a significant effort to do so in a manner that aids, not impedes, the subsequent creation of enhanced, third-party collections of biological information. Acknowledgments and considerations National Science Foundation (NSF) grant DEB-93-00679 from the Division of Environmental Biology to Jerry Franklin, University of Washington, with a subcontract to Caroline Bledsoe, supported this work. Any opinions, findings and conclusions, or recommendations expressed in this article are those of the authors and do not necessarily reflect NSF views. Mention of specific commercial software or businesses is neither a recommendation nor an endorsement. We thank the following current or former LTER site data management personnel: Phyllis Adams, Karen Baker, Barbara Benson, James Brunt, Gil Calabria, A. El Haddi, Don Henshaw, Rick Ingersoll, David Jones, Tom Kirchner, Lolita Krievs, Richard Lent, Phyllis Likens, Eda Melendez, Anne Miller, Barbara Nolen, Rudolf Nottrott, John Porter, and Cindy Veen. Ray Bero of the Network Office is the current librarian for the All-Site Bibliography (biblio@lternet.edu). We thank James F. Quinn, Center for Ecological Health Research, University of California, Davis, for use of computer facilities. This work was supported (in part) by the US Environmental Protection Agency (R819658) Center for Ecological Health Research at the University of California, Davis. Although the information in this article has been funded wholly or in part by EPA, it may not necessarily reflect the views of the agency, and no official endorsement should be inferred. We are grateful to James R. Gosz, Stephanie J. Martin, James F. Quinn, and Christine M. Schonewald for reviewing the manuscript. References cited Aho AV, Kernighan BW, Weinberger PJ. 1988. The AWK programming language. Reading (MA): Addison-Wesley Publishing Co. Callahan JT. 1984. Long-term ecological research. BioScience 34: 363-367. Chinn H, Beardsley K. 1996. ICE MAPS: Interactive California Environmental Management, Assessment, and Planning System. In Information Center for the Environment [Online]. Available: http://ice.ucdavis.edu/ice_maps. Cohen D, ed. 1992 July. A format for e-mailing bibliographic records. Network Working Group, Request For Comments: 1357 [Online]. Available: ftp://ds.internic.net/rfc/ rfc 1357.txt. Colwell RK. 1995. The report of the special committee of the Ecological Society of America on "ESA communications in the electronic age." Bulletin of the Ecological Society of America 76: 121-131. Evans J. 1994. Sharing spatial environmental information across agencies, regions and scales: issues and solutions. Pages 203-219 in Michener WK, Brunt JW, Stafford SG, eds. Environmental information management and analysis: ecosystem to global scales. Bristol (PA): Taylor & Francis. Franklin JF, Bledsoe CS, Callahan JT. 1990. The Long-Term Ecological Research Program: a contributor to ecological science. BioScience 40:509-523. Gaffin A, with Heitkotter J. 1994 September. EFF's (extended) guide to the Internet, version 2.3 (formerly Big dummy's guide to the Internet) [Online]. Available: http://www.eff.org/papers/eegtti/eegttitop.html [1996, March]. gopher. no date. Source for gopher Internet software and information about gopher [Online]. Available: gopher://boombox. micro.umn.edu. Hayes B. 1994. The World Wide Web. American Scientist 82(5): 416-420. Ingersoll R, Hartman M. 1995 July 21. LTER site online datasets & methodology. In US Long-Term Ecological Research Program (LTER) [Online]. Available: gopher://lternet.edu/00/AboutLTER/d-site/ online.txt. Java. 1995. Java[TM]: programming for the Internet. In Sun Microsystems, Inc. [Online]. Available: http://java.sun.com/[1996, April]. Krol E. 1994. The whole Internet user's guide and catalog. 2nd ed. Sebastopol (CA): O'Reilly & Associates. See also the publisher's online description at http:// www.ora.com/www/item/twi2.html. LTERnet. 1996a. All-Site Bibliography. In US Long-Term Ecological Research Program (LTER) [Online]. Available: http:// lternet.edu/biblio/. -----. 1996b. The bibliography in a standard, plain text format (All-Site Bibliography). In US Long-Term Ecological Research Program (LTER) [Online]. Available: http:// lternet.edu/biblio/files/. -----. 1996c (January 18). Developing an LTER network information system for the 21st century (DRAFT). In US Long-Term Ecological Research Program (LTER) [Online]. Available: http://lternet.edu/is/. -----. 1996d (February 22). Information management in the Long-Term Ecological Research (LTER) network. In US Long-Term Ecological Research Program (LTER) [Online]. Available: http://lternet.edu/ im.htm. -----. No date a. Gopher menu for the Core Dataset Catalog. In US Long-Term Ecological Research Program (LTER) [Online]. Available: gopher://lternet.edu/11/catalog [1996, March]. -----. No date b. LTERnet personnel database search. In US Long-Term Ecological Research Program (LTER) [Online]. Available: http://lternet.edu/cgi-bin/lterdb1 [1996, March]. -----. No date c. Long-Term Ecological Research sites. In US Long-Term Ecological Research Program (LTER) [Online]. Available: http://lternet.edu/about/sites/ names.htm [1996, March]. -----. No date d. Site data managers. In US Long-Term Ecological Research Program (LTER) [Online]. Available: http:// lternet.edu/about/scientis/site_dm.htm [1996, March]. Michener WK, Miller AB, Nottrott RW, eds. 1990. Long-Term Ecological Research Network core data set catalog. Columbia (SC): Belle W. Baruch Institute for Marine Biology and Coastal Research, University of South Carolina. [NCSA] National Center for Supercomputing Applications. No date a. A beginner's guide to HTML. In NCSA [Online]. Available: http://www.ncsa.uiuc.edu/General/Internet/ WWW/HTML Primer.html [1996, March]. -----. No date b. The common gateway interface. In NCSA [Online]. Available: http:// httpd.ncsa.uiuc.edu/cgi/[1996, September 30]. -----. No date c. Mosaic for X version 2.0 Fill-Out Form Support. In NCSA [Online]. Available: http://www.ncsa.uiuc.edu/SDG/Software/Mosaic/Docs/fill-out-forms/ overview.html [1996, March]. Neal PR. 1993. Personal bibliographic software programs - a comparative review. BioScience 43:44-51. Niwot. 1996 July 22. Niwot Ridge LTER site bibliography. In The Niwot Ridge (Colorado) site, US Long-Term Ecological Research Program (LTER) [Online]. Available: http://c ulter.colorado.edu: 1030/Niwot/ Niwot_Ridge_LTER_bibliography.html [1996, September 30]. Secret A. No date. The WWW virtual library. In The World Wide Web Virtual Library (subject catalogue) [Online]. Available: http://www.w3.org/hypertext/DataSources/bySubject/Overview2.html [1996, March]. Smith UR. 1993 November. A biologist's guide to Internet resources. Version 1.7 [Online]. Available: http://sunsite.unc.edu/pub/academic/biology/ecology+evolution/ bioguide. [WAIS] Wide Area Information Systems. 1995. Frequently Asked Questions (FAQs) for FreeWAIS 0.5. In The Center for Networked Information Discovery and Retrieval (CNIDR) [Online]. Available: http:// cnidr.org/Software/freewais.html [1996, April]. Walker TJ, Choate PM. 1994. Inputting CD-ROM records - a comparison of biblio-graphic software. BioScience 44: 269-271. RELATED ARTICLE: US Long-Term Ecological Research Program (LTER) In 1980, the National Science Foundation established the US Long-Term Ecological Research Program (LTER) to conduct research on long-term ecological phenomena in the United States. The present program consists of a coordinating network office and 18 sites in the United States, Antarctica, and Puerto Rico. These sites represent diverse ecosystems and research emphases. The LTER Network is a collaborative effort among nearly 1400 scientists and students that extends the opportunities and capabilities of the individual sites to promote ecological synthesis and comparative research across sites and ecosystems in cooperation with other, related. domestic and international research programs. LTER encourages the broader ecological research community to use the Network sites for both long-and short-term projects. RELATED ARTICLE: Centralized versus distributed database systems One decision facing managers of networked, computerized database systems is the choice between a centralized system and a decentralized, distributed one. We illustrate this choice using the All-Site Bibliography. Distributed systems. Advantages: * Most of the parts (the individual site bibliographies) will always be online, available for searching (i.e., they will not all be simultaneously offline). * Logically, that portion of the online, searchable bibliography for a particular site belongs at the site that maintains the original source database. * A distributed system reduces or eliminates the duplication of data, depending on the techniques used for database storage and access. * Reduced data duplication lowers the overall cost of disk storage capacity for the Network, a factor that is increasingly important as the size of biological databases grows. * Giving the responsibility to each site for its portion of the All-Site Bibliography may encourage better maintenance of the original for easier conversion, thereby improving the quality of the online version. Disadvantage: * Despite the final advantage, overall quality control (e.g., maintaining a single online format that is consistent across all sites and keeping the All-Site Bibliography up-to-date) is more difficult when responsibility is dispersed among many individuals located across an area the size of the United States. Centralized systems. Advantages: * When maintained by a single individual or a small group at a single location, quality is easier to control. * Centralization relieves busy site data managers of the maintenance task. Disadvantages: * There is less motivation to improve the site's primary bibliography for better and easier conversion to the online form (someone else deals with the conversion difficulties). * It requires an individual or group to handle the entire project. * If the Network Office computer system is offline, the entire All-Site Bibliography is unavailable. * Data is duplicated. An official bibliography is present at the site, and a separate version of it is in the central bibliography. As a result, there is always the concern of how up-to-date the (centralized) All-Site Bibliography is. * As the amount of data grows, it becomes less feasible to store duplicate versions. The cost of the additional disk storage capacity becomes significant. RELATED ARTICLE: Other online bibliographies There are few lists of pointers to online bibliograpies available on the Internet. Three that list bibliographies for the life sciences are: * A Biologist's Guide to Internet Resources by Smith (1993). See section "3.1. Bibliographies" at http://sunsite.unc.ude/^una/bioguide/ * "Environmental Bibliographic Collections" from the US Geological Survey at http://h2o.usgs.gov/public/biblio.html * A list of the Information Center for the Environment, University of California, Davis, links to nearly 90 bibliographies. See it at http://ice.ucdavis.edu/Cyberspace_Jump_Station/bibliographies/bio logy.html Of related interest is BibNet - A Bibliography Network Project. "This initiative is a step toward sharing information electronically." See ftp://ftp.math.utah.edu/pub/bibnet/faq.html. Harvey Chinn (harvey@ice.ucdavis.edu) is a computer specialist with training in ecology, who worked for both the US Long-Term Ecological Research Program (http://lternet.edu/) and the Information Center for the Environment, University of California, Davis, CA 95616 (http://ice.ucdavis.edu/), where he was the founding Webmaster. Caroline S. Bledsoe (csbledsoe@ucdavis.edu) is associate professor of soil ecology, Department of Land, Air, and Water Resources, University of California, Davis, CA 95616, and research coordinator for the US Long-Term Ecological Research Program. Please send inquiries about this article to Bledsoe. -------------------------------------------------------------------------------- COPYRIGHT 1997 American Institute of Biological Sciences Send questions, comments, or suggestions to cdl@www.cdlib.org MelvylŪ is a registered trademark of The Regents of the University of California