THE DEVELOPMENT OF AN ECOLOGICAL CLASSIFICATION DATA MANAGEMENT AND ANALYSIS SYSTEM FOR BRITISH COLUMBIA G.M. BRITTON, D.V. M E I D I N G E R and A. B A N N E R British Columbia Forest Service, Research Branch, 31 Bastion Square, Victoria, British Columbia, V8W 3E7, Canada
Abstract, Since 1975, the British Columbia Ministry of Forests has been systematically developing an ecosystem classification of the province, an area covering 94 million hectares. This Biogeoclimatic Ecosystem Classification (BEC) system provides a framework for resource conservation and management. To date, approximately 250 person-years have been invested in the collection, analysis and synthesis of over 8000 ecological (vegetation and environmental data) plots, and in the production of ecological field guides. The development of a database and analysis system on the micro-computer platform to support a classification system of this magnitude was a complex procedure that required judicious planning and coordination. We have developed data-processing software that permits a user to select raw data from broad provincial or regional coverage to plot- and species-level summaries, and to export the data to a variety of output formats. This paper addresses key issues for handling ecological field data on the desktop computer with emphasis on standards, operator ease-of-use, and data access.
1. Introduction The ecological land classification system used throughout the province of British Columbia (BC) is called the Biogeoclimatic Ecosystem Classification (BEC) system (Pojar et al., 1987; Meidinger and MacKinnon, 1989; Meidinger and Pojar, 1991). Mah etal. (this volume) provide a recent account of the classification and its use. The BEC system evolved from the early ecological investigations throughout the province by V.J. Krajina (1965) and his students, and is now acknowledged as the provincial ecosystem classification standard. At present, province-wide coverage of biogeoclimatic mapping and detailed BEC field guides are available for use by both government and industry. The system has proven to be a useful framework for resource conservation and management (MacKinnon et al., 1992). The development of British Columbia's BEC was an evolutionary process. The BEC system was initiated by the British Columbia Forest Service (BCFS) in 1975 when it was implemented as a regional classification program with provincial-level support and coordination of methods. Over the years, methods for data gathering, storage and analysis developed to meet changing program needs. These methods are briefly outlined in this paper with a focus on the recent developments in the use of personal computer (PC) based systems for data storage and analysis. Environmental Monitoring and Assessment 39: 365-372, 1996. (~) 1996 Kluwer Academic Publishers. Printed in the Netherlands.
366
G.M. BRITTON ET AL.
TABLE I Evolution of BEC database and data analysis systems. Time period
Database
Data analysis
Early years 1975-1979 Middle years 1980-1992
Storage: Univ. of B.C, mainframe Format: sequential files Storage: BCSC mainframe Format: sequential files with manual index
Recent years 1993-1994
Storage: PC and Sun-UNIX system. Format: relational xBase
VTAB (Klinka and Standish, 1977; Klinka e t al., 1977; Emanuel, 1984) VEGSORT, ENTREE (Meidinger e t al., 1983) CONSYM, CONMENU (Meidinger e t al., 1983) BCSIS (Sondheim, 1983; Sondheim and Suttie, 1983; Sondheim e t al., 1983) TWINSPAN, DECORANA (Hill, 1979a, 1979b) ORDIFLEX (Gauch, 1977) PC-VTAB, WinVTAB (Emanuel, 1990; Kayahara, 1992) ECOBASE (Wrenshall, 1991; Britton, 1994)
2. Data Collection Data collection for the development of BEC in BC consists of detailed site, soil, and vegetation descriptions as well as forest mensuration measurements using the methods described in Luttmerding et al. (1990). The field data are measured by field crews normally consisting of an ecologist, a soil scientist and a mensuration technician. The sampling method is essentially subjective, with minimal bias (as described in Pojar et al., 1987). Early implementation of provincial standards for field sampling of ecosystems (Luttmerding et aL, 1990), utilizing standard field forms and definitions has resulted in good quality, consistent data from which to build the provincial database.
3. Database Development The large, historical collection of data on standardized field forms dictated a logical structure for the initial database model. The ecological classification field forms consist of vegetation and environment (soil and site) data for each plot sampled. The vegetation data consist of the plot number, vegetation layer codes, and the species code, which is associated with three co-variables: percent cover, distribution, and vigor. The environment data consist of the plot number, location, and many site and soil attributes (e.g., slope, aspect, landform, soil moisture and nutrient regime, etc.). Early in the program (e.g., 1975), the ecological data system was housed in a mainframe file storage and analysis system at the University of British Columbia (UBC) (Klinka and Standish, 1977; Klinka et al., 1977) (Table I). Most BEC computing was done on university mainframes due to micro-computer memory
ECOLOGICAL DATA MANAGEMENT AND ANALYSIS SYSTEM
367
and data processing limitations. Ecologists were required to travel to UBC or work through a consultant in Vancouver via mail/courier. The data files were stored on tape spools in sequential-file formats. The database system, as such, was nonrelational and catalogued by hand. In 1980, the data files and analysis programs were moved to the British Columbia Systems Corporation (BCSC) mainframe computer (Table I). This move gave regional and headquarters ecologists access to all data and programs through the provincial computer network. Differences in the mainframe operating systems required rewriting much of the analytical software and the data formats were altered slightly due to the demand to manage more environmental data (Meidinger et al., I983; Meidinger and MacKinnon, 1989). About the same time, the British Columbia Soil Information System (BCSIS) was developed as a mainframe (MARK IV) database to store site and soil data (Sondheim and Suttie, 1983; Sondheim et aL, 1983). Much of the new data being collected were entered into BCSIS and then a subset extracted for the ecology environment data files on the BCSC mainframe. In recent years, the availability of micro-computers, the increase in use and knowledge of the PC environment, and the rapid improvement in PC storage and processing power encouraged the design of a PC-based BEC database. The PC database, now called ECOBASE, is designed to work in a Microsoft Windows TM (Microsoft Corporation, 1990-1992) environment and has been programmed in C++ using code-modules which will allow for the anticipated evolution of both the database system and its corresponding operating environment. The user-interface was designed for flexibility and ease-of-use. It matches the ecologist's procedural approach, and incorporates a set of default logical-input values and extensive error-checking to provide integrity to the extracted data. The system has a variety of user-specified and default data query options to simplify the extraction of customized research data sets. The database itself is written in the commercially standard dBase I V TM (*.dbf) format (Borland International, 1984) which can be easily exported to the format of most common analytical software packages. The easy to use PC-based procedures allow scientists to easily obtain specific data for their projects and then use the data with their preferred analytical software and computational environment. For example, data is commonly extracted in ASCII format for use with VTAB, a vegetation tabular analysis program (Emanuel, 1990; Kayahara, 1992). Currently, the entire provincial database is available on a database server at BC Forest Service headquarters in Victoria and each of the six BCFS Forest Regions receive region-specific database subsets. In addition, a version of the main database is also operating on a networked Sun-UNIX minicomputer. Any additions and refinements to this provincial database will take place exclusively at headquarters. The regional database subsets are updated at least quarterly. As required, additional project-specific data subsets can be obtained on a per-request basis.
368
G.M. BRITTON ET AL. Vegetation data ]
Sorting instructions
plot number ~>
plot numbers
(~] plot number
species code cover/distribution/vigor
BEC classification
location data
full species names Fig. 1.
Environmentdata]
site data soil data
BEC database structure.
The assemblage of vegetation and environment data with site-series sorted unique plot number data (called, sorting instructions) make up the core structure of the relationally-linked database (Figure 1). The sorting instructions file, which is created to group the plot numbers into their corresponding BEC classification unit, adds the relational dimension to the database. Classifying the sample plot data, and its corresponding placement in the sorting instructions file, is determined via several iterations as a result of Braun-Blanquet-style tabular analysis (MuellerDombois and Ellenberg, 1974) in combination with multivariate techniques. The BEC database is indexed by plot number, biogeoclimatic and vegetation/site unit, and forest region, respectively. The development of the desktop computer database system was specifically focused on the client or end-user. The database system had to work adeptly with standard historical programs and analysis procedures. The program had to be comprehensive to address complex querying of the data but also simple and intuitive to use. In this way the data could be accessed without requiring an intermediary to assist in the data access process. Centralized database storage and operational administration was essential to ensure data integrity. The central-server database management system was designed to allow updating and editing of data by specifically authorized personnel only. In this manner, any changes can be traced through database administration records and the data integrity is maintained. In the near-future, the database will be connected to an extended-network in order to offer controlled accessibility for distant clients. Although not presently linked to the plot database, a geographic information systems (GIS) database of biogeoclimatic units in the form of a geographically referenced data matrix, exists for the province. In the future, we plan to have the plot database entered as points in a GIS data layer and useable with the biogeoclimatic and other map layers. This would allow extraction of data by geographic area, querying of plot data using the biogeoclimatic database, modeling of plant distributions, etc. This may involve various statistical applications such as regression modeling, and non-parametric and fractal analytical methodologies.
ECOLOGICAL DATA MANAGEMENT AND ANALYSIS SYSTEM
369
4. Data Analysis As mentioned previously, until 1992 (see Table I), most of the classification analyses were accomplished with a mainframe analysis package at the University of British Columbia and then later through the BCSC provincial network. Through the provincial computing system, a complete analysis system was available to the ecologists (Meidinger et aL, 1983, 1987). The analysis programs included ORDIFLEX (Gauch, 1977), TWINSPAN and DECORANA (Hill, 1979a, 1979b), CERO (Ceska and Roemer, 1971), and FASTCLUS (SAS Institute Inc., 1982). Data were analyzed with a variety of programs and then presented using standard tabular display programs. The programs for tabular analysis and final presentation of vegetation and environment/mensuration data were, respectively, VEGSORT and ENTREE. The final tables represent a classification of plots using vegetation composition and site factors. In addition, the analysis system had software to test a classification hierarchy sensu Braun-Blanquet (Mueller-Dombois and Ellenberg, 1974). The program, entitled CONMENU, used vegetation constancy tables produced by VEGSORT and allowed the user to produce hierarchical vegetation tables and test diagnostic values for species in associations, alliances, and orders. Due to the 1992 changes in the data edit and access programs at BCSC and the requirement to rewrite the front end of the whole system, most of the programs were transferred to the Sun-UNIX mini-computer at headquarters and we embarked upon the development of a PC-based analysis and tabling system. The resulting software, entitled PC-VTAB (Emanuel, 1990; Kayahara, 1992), performs the analyses that were available on the mainframe VTAB program at UBC and the VEGSORT, ENTREE, and CONMENU programs at BCSC. At present, both DOS (PC-VTAB) and Windows TM (WinVTAB) versions of this software are available. VTAB performs more than just tabular analysis. VTAB's main screen consists of various menus and subprograms. The file menu allows the user to enter and edit the vegetation and environment data, extract specific plot data via sorting instructions files and edit species codes/names files. The report menu produces various data summaries such as vegetation tables, environment tables, diagnostic tables and summary tables. The data analysis menu produces a similarity matrix, spectral histograms, and has a data export function to ORDIFLEX, Cgroup, or CONSYM format to facilitate the use of other analytical software. For assistance, on-line help is provided through the help menu. Some examples of the VTAB reports: long vegetation table, diagnostic table, and a similarity matrix are shown in Figure 2. WinVTAB, the Microsoft Windows 3.1TM version of VTAB, is currently undergoing extensive revision and should be wholly Windows TM capable soon. The Windows TM environment permits the analysis of a larger number of plots, site units and species - limited only by the amount of available dynamic memory.
370
G.M. BRITTON ET AL.
Pseudotsuga Acer(gla) Alliance Long Vegetation Table IDF xhl 08 KAM S• Douglas maple - Dogwood PLOT AVERAGE RC I BR I BR I 85D NUMBER VALUES 60221AI2OI All91 LOB0 ST SPECIES P MC PERCENT COVER A1 Pseudotsuga menziesii 26.7 2.1 8 441 5 441 Pinus ponderosa 20.0 ,8 1 431.5 141 Populus balsamifera 13.3 1.1 6 44110 731 Thuja plicata 6.7 .3 I I A2 Pseudotsuga menziesii 93.3 16.6155 113 741 74140 Betula papyrifera 40.0 5.2 5 2 441 241 s Picea engelmanniiXglauca 33.3 12.0 I I
Thuja p l i c a t a 26.7 Populus balsemifera 26.7 Pinus ponderosa 20.0 Acer glabrum 20,0 Abies lasiocarpa 13,3 Populus tremuloides 13.3 S a l i x sp. 6,7 Alnus t e n u i f o l i a 6.7 Prunus v i r g i n i a n a 6.7 A3 Acer glabrum 26.7 Betula p a p y r i f e r a 26.7 Picea engelmanniiXglauca 20.0 Pseudotsuga menziesii 20.0 Thuja p l i c a t a 6.7 BI Acer glabrum l 60.0 Cornus s t o l o n i f e r a l 26.7 Picea engelmanniiXglauca l 20.0 Betula papyrifera ] 13.3 Rosa nutkana I 13.3
Diagnostic Table for
9.8 4.0 1.7 .65 2.3 1.0 1.0 .0 .0
Number of plots
[
1 23110 73120
.5 ]31 ] 2 441 2 24 I I I [.5 231 I
VTAB 1.6 S i m i l a r i t y Matrix based on Goldstream c o e f f i c i e n t Pseudotsuga Acer(gla) Alliance
.6 14 I .5 241
5,31 5.51
I I I I I
1,51
1.II .51
I I I I I
2142.4 3 I 40.3 34.5 4 I 35.0 24.2 30.0 5 I 31.1 25.8 39.0 33.9 6 I 24.1 22.9 36.9 30.8 61.5 +. . . . . . . . . . . . . . . . . . . . . . . . .
125 741 3 441 I 6 44115 74 I I I I I I I I 1 14L15 74t
9.3 l
2.11 .91
1.61 111
1
2
3
Average s i m i l a r i t y
4 -
5
34.2
Acer (gla) Alliance i 2 3 4 5 6 IDF xhl PP xhl IDF ww ESSFdk IDF xm SBS mh 08 KAM 08 KAM 04 VAN 02 NEL 07 CAR 04 CG 15 3 6 4 3 8 diag Presence class and mean percent cover
Pseudotsuga
Vegetation unit
.5 131
-
Pseudotsuga - Acer - Symphoricarpos Suballiance Symphoricarpos albus
(d)
I IV
7
5
15
IV
5
3
i
4
3
II
t I
II
t
Pseudotsuga - Acer -Cornus (ser) Association
Galium t r i f l o r u m
(d)
IIII
t I
Pseudotsuga - Betula (occ) - Acer Association Alnus t e n u i f o l i a Betula occidentalis Dactylis glomerata Equisetum hyemale Poa sp. Populus tremuloides Salix sp. Smilacina s t e l l a t a
(d) (d,c) (d) (d) (d,cd) (d,c) (d) (d,c)
]I
2
I I II I ]If
t t i 3 t
I I
t t
4 5 4 4 5 5 4 5
2 8 3 3 30 7 I 5
I
2 II
1
IV
131
t
2
t I
Pseudotsuga - Acer - Disporum Association Disporum hookeri Fragaria vesca
(d,cd) (d)
Fig. 2.
VTAB - some examples
I IV
1 I
of data reports.
II II
ECOLOGICAL DATAMANAGEMENTAND ANALYSISSYSTEM
371
PC-VTAB is now being used by the regional and headquarters ecologists, and some consulting ecologists, for the analysis of their data. Our plans for the near future include further testing of various commercial software for statistical and graphical analysis, fuzzy-logic clustering analysis, spreadsheet tabling, and xBase analysis, in order to improve the data analysis capabilities of the program ecologists. As well, we plan to develop an expert system, possibly via a neural-network simulation model, to assist in classifying new plot data.
5. Summary The evolution of the BEC system has been accompanied by changes to the database and analysis systems. These alterations were necessary to enable the completion of the classification and to match the progressive advancement in analysis hardware. The BCFS has recently developed PC-based software for its ecological classification database and corresponding analysis systems. The software is intended to be accessible to a medley of users and is continually evolving. Currently, ECOBASE is at Version 1.0 and PC-VTAB and WinVTAB are at version 1.6. The analysis software is available from the authors, upon request.
References Borland International: 1984, dBase IVTM, Software, Scotts Valley, Califomia. Britton, G.M.: 1994, 'ECOBASE- A database management system for ecology research', Software, B.C. Min. For., Victoria, British Columbia. Ceska, A. and Roemer, H.: 1971, 'A computer program for identifying species-releve groups in vegetation studies', Vegetatio 23, 255-277. Emanuel, J.: 1984, 'F405:VTAB - A vegetation classification program', Unpublished draft, The University of British Columbia, Vancouver, British Columbia. Emanuel, J.: 1990, 'PC-VTAB - A vegetation classification program', Software, B.C. Min. For., Victoria, British Columbia. Gauch, H.G. Jr.: 1977, ORDIFLEX - A Flexible Computer Program for Four Ordination Techniques: Weighted Averages, Polar Ordination, Principle Components Analysis, and Reciprocal Averaging, Department of Ecology and Systematics, Cornell University, Ithaca, New York. Hill, M.O.: 1979a, DECORANA - A FORTRAN Program for Detrended Correspondence Analysis and Reciprocal Averaging, Department of Ecology and Systematics, Cornell University, Ithaca, New York. Hill, M.O.: 1979b, TWINSPAN - A FORTRAN Program for Arranging Multivariate Data in an Ordered Two-Way Table by Classification of the Individuals and Attributes, Department of Ecology and Systematics, Cornell University, Ithaca, New York. Kayahara, G.J.: 1992, 'PC-VTAB 1.4 User's Guide', Internal report, B.C. Min. For., Victoria, British Columbia. Klinka, K. and Standish, J.T.: 1977, 'Initial data synthesis in the synecological classification; Part 1: Introduction to the synecological classification of lower systematic units', Unpublished draft, B.C. Min. For., Victoria, British Columbia. Klinka, K., Kimmins, J.P., Standish, J.T. and Phelps, S.: 1977, 'Initial data synthesis in the synecological classification; Part 2: Environmental - vegetation tables and summary vegetation tables by a computer program', Unpublished Draft, B.C. Min. For., Victoria, British Columbia.
372
G.M. BRITTONET AL.
Krajina, V.J.: 1965, 'Biogeoclimatic zones and classification in British Columbia', Ecol. West. N. Amer. 1, 1-17. Luttmerding, H.A., Demarchi, D.A., Lea, E.C., Meidinger, D.V. and Void, T.: 1990, Describing Ecosystems in the Field, 2nd ed., MOE Manual No. 11, B.C. Min. Environ. and B.C. Min. For., Victoria, British Columbia. MacKinnon, A., Meidinger, D. and Klinka, K.: 1992, 'Use of the Biogeoclimatic Ecosystem Classification System in British Columbia', For. Chron. 68(1), 100-120. Mah, S., Demarchi, D. and Thomson, S.: 1996, 'An ecological framework for resource management in British Columbia', Environ. Monitor. Assess. (this volume). Meidinger, D.V. and MacKinnon, A.: 1989, 'Biogeoclimatic classification - the system and its application', In: D.E. Ferguson, P. Morgan and F.D. Johnson (eds.), Proceedings - Land Classifications Based on Vegetation: Applications for Resource Management, US Dep. Agric., For. Serv., Intermountain Research Station, Moscow, Idaho, pp. 215-222. Meidinger, D. and Pojar, J. (compilers and editors): 1991, Ecosystems of British Columbia, Special Report Series No. 6, B.C. Min. For., Victoria, British Columbia, 330 pp. Meidinger, D.V., Hamilton, E. and Fleming, T.: 1983, (revised, 1987), Ecosystem Classification Program, Data Processing System User's Guide, Version 2.0, Research Report No. RR87006HQ, B.C. Min. For. Lands, Victoria, British Columbia. Microsoft Corporation: 1990-1992, Microsoft Windows 3.1 TM, Software, Redmond, Washington, DC. Mueller-Dombois, D. and Ellenberg, H.: 1974, Aims and Methods of Vegetation Ecology, John Wiley and Sons, New York. Pojar, J., Klinka, K. and Meidinger, D.V.: 1987, 'Biogeoclimatic ecosystem classification in British Columbia', For. Ecol. Manage. 22, 119-154. SAS Institute Inc.: 1982, SAS(c) System, Software, Cary, North Carolina. Sondheim, M.: 1983, 'Validation procedures, BCSIS - Volume 4', B.C. Min. For., Victoria, British Columbia. Sondheim, M. and Suttie, K.: 1983, 'User's manual for the British Columbia soil information system, BCSIS Volume 1', B.C. Min. For., Victoria, British Columbia. Sondheim, M., Void, T. and Quesnel, H.: 1983, 'Data entry procedures for ecosystem description forms, BCSIS - Volume 2', B.C. Min. For., Victoria, British Columbia. Walmsley, M., Utzig, G., Void, T., Moon, D. and Van Barneveld, J.: 1980, 'Describing ecosystems in the field', B.C. Min. Environ. and B.C. Min. For., Victoria, British Columbia. Wrenshall, C.: 1991, ECOMENU- A Mainframe FORTRAN Program to Assist in Vegetation Analysis, Software, B.C. Min. For., Victoria, British Columbia. -
Abstract, Since 1975, the British Columbia Ministry of Forests has been systematically developing an ecosystem classification of the province, an area covering 94 million hectares. This Biogeoclimatic Ecosystem Classification (BEC) system provides a framework for resource conservation and management. To date, approximately 250 person-years have been invested in the collection, analysis and synthesis of over 8000 ecological (vegetation and environmental data) plots, and in the production of ecological field guides. The development of a database and analysis system on the micro-computer platform to support a classification system of this magnitude was a complex procedure that required judicious planning and coordination. We have developed data-processing software that permits a user to select raw data from broad provincial or regional coverage to plot- and species-level summaries, and to export the data to a variety of output formats. This paper addresses key issues for handling ecological field data on the desktop computer with emphasis on standards, operator ease-of-use, and data access.
1. Introduction The ecological land classification system used throughout the province of British Columbia (BC) is called the Biogeoclimatic Ecosystem Classification (BEC) system (Pojar et al., 1987; Meidinger and MacKinnon, 1989; Meidinger and Pojar, 1991). Mah etal. (this volume) provide a recent account of the classification and its use. The BEC system evolved from the early ecological investigations throughout the province by V.J. Krajina (1965) and his students, and is now acknowledged as the provincial ecosystem classification standard. At present, province-wide coverage of biogeoclimatic mapping and detailed BEC field guides are available for use by both government and industry. The system has proven to be a useful framework for resource conservation and management (MacKinnon et al., 1992). The development of British Columbia's BEC was an evolutionary process. The BEC system was initiated by the British Columbia Forest Service (BCFS) in 1975 when it was implemented as a regional classification program with provincial-level support and coordination of methods. Over the years, methods for data gathering, storage and analysis developed to meet changing program needs. These methods are briefly outlined in this paper with a focus on the recent developments in the use of personal computer (PC) based systems for data storage and analysis. Environmental Monitoring and Assessment 39: 365-372, 1996. (~) 1996 Kluwer Academic Publishers. Printed in the Netherlands.
366
G.M. BRITTON ET AL.
TABLE I Evolution of BEC database and data analysis systems. Time period
Database
Data analysis
Early years 1975-1979 Middle years 1980-1992
Storage: Univ. of B.C, mainframe Format: sequential files Storage: BCSC mainframe Format: sequential files with manual index
Recent years 1993-1994
Storage: PC and Sun-UNIX system. Format: relational xBase
VTAB (Klinka and Standish, 1977; Klinka e t al., 1977; Emanuel, 1984) VEGSORT, ENTREE (Meidinger e t al., 1983) CONSYM, CONMENU (Meidinger e t al., 1983) BCSIS (Sondheim, 1983; Sondheim and Suttie, 1983; Sondheim e t al., 1983) TWINSPAN, DECORANA (Hill, 1979a, 1979b) ORDIFLEX (Gauch, 1977) PC-VTAB, WinVTAB (Emanuel, 1990; Kayahara, 1992) ECOBASE (Wrenshall, 1991; Britton, 1994)
2. Data Collection Data collection for the development of BEC in BC consists of detailed site, soil, and vegetation descriptions as well as forest mensuration measurements using the methods described in Luttmerding et al. (1990). The field data are measured by field crews normally consisting of an ecologist, a soil scientist and a mensuration technician. The sampling method is essentially subjective, with minimal bias (as described in Pojar et al., 1987). Early implementation of provincial standards for field sampling of ecosystems (Luttmerding et aL, 1990), utilizing standard field forms and definitions has resulted in good quality, consistent data from which to build the provincial database.
3. Database Development The large, historical collection of data on standardized field forms dictated a logical structure for the initial database model. The ecological classification field forms consist of vegetation and environment (soil and site) data for each plot sampled. The vegetation data consist of the plot number, vegetation layer codes, and the species code, which is associated with three co-variables: percent cover, distribution, and vigor. The environment data consist of the plot number, location, and many site and soil attributes (e.g., slope, aspect, landform, soil moisture and nutrient regime, etc.). Early in the program (e.g., 1975), the ecological data system was housed in a mainframe file storage and analysis system at the University of British Columbia (UBC) (Klinka and Standish, 1977; Klinka et al., 1977) (Table I). Most BEC computing was done on university mainframes due to micro-computer memory
ECOLOGICAL DATA MANAGEMENT AND ANALYSIS SYSTEM
367
and data processing limitations. Ecologists were required to travel to UBC or work through a consultant in Vancouver via mail/courier. The data files were stored on tape spools in sequential-file formats. The database system, as such, was nonrelational and catalogued by hand. In 1980, the data files and analysis programs were moved to the British Columbia Systems Corporation (BCSC) mainframe computer (Table I). This move gave regional and headquarters ecologists access to all data and programs through the provincial computer network. Differences in the mainframe operating systems required rewriting much of the analytical software and the data formats were altered slightly due to the demand to manage more environmental data (Meidinger et al., I983; Meidinger and MacKinnon, 1989). About the same time, the British Columbia Soil Information System (BCSIS) was developed as a mainframe (MARK IV) database to store site and soil data (Sondheim and Suttie, 1983; Sondheim et aL, 1983). Much of the new data being collected were entered into BCSIS and then a subset extracted for the ecology environment data files on the BCSC mainframe. In recent years, the availability of micro-computers, the increase in use and knowledge of the PC environment, and the rapid improvement in PC storage and processing power encouraged the design of a PC-based BEC database. The PC database, now called ECOBASE, is designed to work in a Microsoft Windows TM (Microsoft Corporation, 1990-1992) environment and has been programmed in C++ using code-modules which will allow for the anticipated evolution of both the database system and its corresponding operating environment. The user-interface was designed for flexibility and ease-of-use. It matches the ecologist's procedural approach, and incorporates a set of default logical-input values and extensive error-checking to provide integrity to the extracted data. The system has a variety of user-specified and default data query options to simplify the extraction of customized research data sets. The database itself is written in the commercially standard dBase I V TM (*.dbf) format (Borland International, 1984) which can be easily exported to the format of most common analytical software packages. The easy to use PC-based procedures allow scientists to easily obtain specific data for their projects and then use the data with their preferred analytical software and computational environment. For example, data is commonly extracted in ASCII format for use with VTAB, a vegetation tabular analysis program (Emanuel, 1990; Kayahara, 1992). Currently, the entire provincial database is available on a database server at BC Forest Service headquarters in Victoria and each of the six BCFS Forest Regions receive region-specific database subsets. In addition, a version of the main database is also operating on a networked Sun-UNIX minicomputer. Any additions and refinements to this provincial database will take place exclusively at headquarters. The regional database subsets are updated at least quarterly. As required, additional project-specific data subsets can be obtained on a per-request basis.
368
G.M. BRITTON ET AL. Vegetation data ]
Sorting instructions
plot number ~>
plot numbers
(~] plot number
species code cover/distribution/vigor
BEC classification
location data
full species names Fig. 1.
Environmentdata]
site data soil data
BEC database structure.
The assemblage of vegetation and environment data with site-series sorted unique plot number data (called, sorting instructions) make up the core structure of the relationally-linked database (Figure 1). The sorting instructions file, which is created to group the plot numbers into their corresponding BEC classification unit, adds the relational dimension to the database. Classifying the sample plot data, and its corresponding placement in the sorting instructions file, is determined via several iterations as a result of Braun-Blanquet-style tabular analysis (MuellerDombois and Ellenberg, 1974) in combination with multivariate techniques. The BEC database is indexed by plot number, biogeoclimatic and vegetation/site unit, and forest region, respectively. The development of the desktop computer database system was specifically focused on the client or end-user. The database system had to work adeptly with standard historical programs and analysis procedures. The program had to be comprehensive to address complex querying of the data but also simple and intuitive to use. In this way the data could be accessed without requiring an intermediary to assist in the data access process. Centralized database storage and operational administration was essential to ensure data integrity. The central-server database management system was designed to allow updating and editing of data by specifically authorized personnel only. In this manner, any changes can be traced through database administration records and the data integrity is maintained. In the near-future, the database will be connected to an extended-network in order to offer controlled accessibility for distant clients. Although not presently linked to the plot database, a geographic information systems (GIS) database of biogeoclimatic units in the form of a geographically referenced data matrix, exists for the province. In the future, we plan to have the plot database entered as points in a GIS data layer and useable with the biogeoclimatic and other map layers. This would allow extraction of data by geographic area, querying of plot data using the biogeoclimatic database, modeling of plant distributions, etc. This may involve various statistical applications such as regression modeling, and non-parametric and fractal analytical methodologies.
ECOLOGICAL DATA MANAGEMENT AND ANALYSIS SYSTEM
369
4. Data Analysis As mentioned previously, until 1992 (see Table I), most of the classification analyses were accomplished with a mainframe analysis package at the University of British Columbia and then later through the BCSC provincial network. Through the provincial computing system, a complete analysis system was available to the ecologists (Meidinger et aL, 1983, 1987). The analysis programs included ORDIFLEX (Gauch, 1977), TWINSPAN and DECORANA (Hill, 1979a, 1979b), CERO (Ceska and Roemer, 1971), and FASTCLUS (SAS Institute Inc., 1982). Data were analyzed with a variety of programs and then presented using standard tabular display programs. The programs for tabular analysis and final presentation of vegetation and environment/mensuration data were, respectively, VEGSORT and ENTREE. The final tables represent a classification of plots using vegetation composition and site factors. In addition, the analysis system had software to test a classification hierarchy sensu Braun-Blanquet (Mueller-Dombois and Ellenberg, 1974). The program, entitled CONMENU, used vegetation constancy tables produced by VEGSORT and allowed the user to produce hierarchical vegetation tables and test diagnostic values for species in associations, alliances, and orders. Due to the 1992 changes in the data edit and access programs at BCSC and the requirement to rewrite the front end of the whole system, most of the programs were transferred to the Sun-UNIX mini-computer at headquarters and we embarked upon the development of a PC-based analysis and tabling system. The resulting software, entitled PC-VTAB (Emanuel, 1990; Kayahara, 1992), performs the analyses that were available on the mainframe VTAB program at UBC and the VEGSORT, ENTREE, and CONMENU programs at BCSC. At present, both DOS (PC-VTAB) and Windows TM (WinVTAB) versions of this software are available. VTAB performs more than just tabular analysis. VTAB's main screen consists of various menus and subprograms. The file menu allows the user to enter and edit the vegetation and environment data, extract specific plot data via sorting instructions files and edit species codes/names files. The report menu produces various data summaries such as vegetation tables, environment tables, diagnostic tables and summary tables. The data analysis menu produces a similarity matrix, spectral histograms, and has a data export function to ORDIFLEX, Cgroup, or CONSYM format to facilitate the use of other analytical software. For assistance, on-line help is provided through the help menu. Some examples of the VTAB reports: long vegetation table, diagnostic table, and a similarity matrix are shown in Figure 2. WinVTAB, the Microsoft Windows 3.1TM version of VTAB, is currently undergoing extensive revision and should be wholly Windows TM capable soon. The Windows TM environment permits the analysis of a larger number of plots, site units and species - limited only by the amount of available dynamic memory.
370
G.M. BRITTON ET AL.
Pseudotsuga Acer(gla) Alliance Long Vegetation Table IDF xhl 08 KAM S• Douglas maple - Dogwood PLOT AVERAGE RC I BR I BR I 85D NUMBER VALUES 60221AI2OI All91 LOB0 ST SPECIES P MC PERCENT COVER A1 Pseudotsuga menziesii 26.7 2.1 8 441 5 441 Pinus ponderosa 20.0 ,8 1 431.5 141 Populus balsamifera 13.3 1.1 6 44110 731 Thuja plicata 6.7 .3 I I A2 Pseudotsuga menziesii 93.3 16.6155 113 741 74140 Betula papyrifera 40.0 5.2 5 2 441 241 s Picea engelmanniiXglauca 33.3 12.0 I I
Thuja p l i c a t a 26.7 Populus balsemifera 26.7 Pinus ponderosa 20.0 Acer glabrum 20,0 Abies lasiocarpa 13,3 Populus tremuloides 13.3 S a l i x sp. 6,7 Alnus t e n u i f o l i a 6.7 Prunus v i r g i n i a n a 6.7 A3 Acer glabrum 26.7 Betula p a p y r i f e r a 26.7 Picea engelmanniiXglauca 20.0 Pseudotsuga menziesii 20.0 Thuja p l i c a t a 6.7 BI Acer glabrum l 60.0 Cornus s t o l o n i f e r a l 26.7 Picea engelmanniiXglauca l 20.0 Betula papyrifera ] 13.3 Rosa nutkana I 13.3
Diagnostic Table for
9.8 4.0 1.7 .65 2.3 1.0 1.0 .0 .0
Number of plots
[
1 23110 73120
.5 ]31 ] 2 441 2 24 I I I [.5 231 I
VTAB 1.6 S i m i l a r i t y Matrix based on Goldstream c o e f f i c i e n t Pseudotsuga Acer(gla) Alliance
.6 14 I .5 241
5,31 5.51
I I I I I
1,51
1.II .51
I I I I I
2142.4 3 I 40.3 34.5 4 I 35.0 24.2 30.0 5 I 31.1 25.8 39.0 33.9 6 I 24.1 22.9 36.9 30.8 61.5 +. . . . . . . . . . . . . . . . . . . . . . . . .
125 741 3 441 I 6 44115 74 I I I I I I I I 1 14L15 74t
9.3 l
2.11 .91
1.61 111
1
2
3
Average s i m i l a r i t y
4 -
5
34.2
Acer (gla) Alliance i 2 3 4 5 6 IDF xhl PP xhl IDF ww ESSFdk IDF xm SBS mh 08 KAM 08 KAM 04 VAN 02 NEL 07 CAR 04 CG 15 3 6 4 3 8 diag Presence class and mean percent cover
Pseudotsuga
Vegetation unit
.5 131
-
Pseudotsuga - Acer - Symphoricarpos Suballiance Symphoricarpos albus
(d)
I IV
7
5
15
IV
5
3
i
4
3
II
t I
II
t
Pseudotsuga - Acer -Cornus (ser) Association
Galium t r i f l o r u m
(d)
IIII
t I
Pseudotsuga - Betula (occ) - Acer Association Alnus t e n u i f o l i a Betula occidentalis Dactylis glomerata Equisetum hyemale Poa sp. Populus tremuloides Salix sp. Smilacina s t e l l a t a
(d) (d,c) (d) (d) (d,cd) (d,c) (d) (d,c)
]I
2
I I II I ]If
t t i 3 t
I I
t t
4 5 4 4 5 5 4 5
2 8 3 3 30 7 I 5
I
2 II
1
IV
131
t
2
t I
Pseudotsuga - Acer - Disporum Association Disporum hookeri Fragaria vesca
(d,cd) (d)
Fig. 2.
VTAB - some examples
I IV
1 I
of data reports.
II II
ECOLOGICAL DATAMANAGEMENTAND ANALYSISSYSTEM
371
PC-VTAB is now being used by the regional and headquarters ecologists, and some consulting ecologists, for the analysis of their data. Our plans for the near future include further testing of various commercial software for statistical and graphical analysis, fuzzy-logic clustering analysis, spreadsheet tabling, and xBase analysis, in order to improve the data analysis capabilities of the program ecologists. As well, we plan to develop an expert system, possibly via a neural-network simulation model, to assist in classifying new plot data.
5. Summary The evolution of the BEC system has been accompanied by changes to the database and analysis systems. These alterations were necessary to enable the completion of the classification and to match the progressive advancement in analysis hardware. The BCFS has recently developed PC-based software for its ecological classification database and corresponding analysis systems. The software is intended to be accessible to a medley of users and is continually evolving. Currently, ECOBASE is at Version 1.0 and PC-VTAB and WinVTAB are at version 1.6. The analysis software is available from the authors, upon request.
References Borland International: 1984, dBase IVTM, Software, Scotts Valley, Califomia. Britton, G.M.: 1994, 'ECOBASE- A database management system for ecology research', Software, B.C. Min. For., Victoria, British Columbia. Ceska, A. and Roemer, H.: 1971, 'A computer program for identifying species-releve groups in vegetation studies', Vegetatio 23, 255-277. Emanuel, J.: 1984, 'F405:VTAB - A vegetation classification program', Unpublished draft, The University of British Columbia, Vancouver, British Columbia. Emanuel, J.: 1990, 'PC-VTAB - A vegetation classification program', Software, B.C. Min. For., Victoria, British Columbia. Gauch, H.G. Jr.: 1977, ORDIFLEX - A Flexible Computer Program for Four Ordination Techniques: Weighted Averages, Polar Ordination, Principle Components Analysis, and Reciprocal Averaging, Department of Ecology and Systematics, Cornell University, Ithaca, New York. Hill, M.O.: 1979a, DECORANA - A FORTRAN Program for Detrended Correspondence Analysis and Reciprocal Averaging, Department of Ecology and Systematics, Cornell University, Ithaca, New York. Hill, M.O.: 1979b, TWINSPAN - A FORTRAN Program for Arranging Multivariate Data in an Ordered Two-Way Table by Classification of the Individuals and Attributes, Department of Ecology and Systematics, Cornell University, Ithaca, New York. Kayahara, G.J.: 1992, 'PC-VTAB 1.4 User's Guide', Internal report, B.C. Min. For., Victoria, British Columbia. Klinka, K. and Standish, J.T.: 1977, 'Initial data synthesis in the synecological classification; Part 1: Introduction to the synecological classification of lower systematic units', Unpublished draft, B.C. Min. For., Victoria, British Columbia. Klinka, K., Kimmins, J.P., Standish, J.T. and Phelps, S.: 1977, 'Initial data synthesis in the synecological classification; Part 2: Environmental - vegetation tables and summary vegetation tables by a computer program', Unpublished Draft, B.C. Min. For., Victoria, British Columbia.
372
G.M. BRITTONET AL.
Krajina, V.J.: 1965, 'Biogeoclimatic zones and classification in British Columbia', Ecol. West. N. Amer. 1, 1-17. Luttmerding, H.A., Demarchi, D.A., Lea, E.C., Meidinger, D.V. and Void, T.: 1990, Describing Ecosystems in the Field, 2nd ed., MOE Manual No. 11, B.C. Min. Environ. and B.C. Min. For., Victoria, British Columbia. MacKinnon, A., Meidinger, D. and Klinka, K.: 1992, 'Use of the Biogeoclimatic Ecosystem Classification System in British Columbia', For. Chron. 68(1), 100-120. Mah, S., Demarchi, D. and Thomson, S.: 1996, 'An ecological framework for resource management in British Columbia', Environ. Monitor. Assess. (this volume). Meidinger, D.V. and MacKinnon, A.: 1989, 'Biogeoclimatic classification - the system and its application', In: D.E. Ferguson, P. Morgan and F.D. Johnson (eds.), Proceedings - Land Classifications Based on Vegetation: Applications for Resource Management, US Dep. Agric., For. Serv., Intermountain Research Station, Moscow, Idaho, pp. 215-222. Meidinger, D. and Pojar, J. (compilers and editors): 1991, Ecosystems of British Columbia, Special Report Series No. 6, B.C. Min. For., Victoria, British Columbia, 330 pp. Meidinger, D.V., Hamilton, E. and Fleming, T.: 1983, (revised, 1987), Ecosystem Classification Program, Data Processing System User's Guide, Version 2.0, Research Report No. RR87006HQ, B.C. Min. For. Lands, Victoria, British Columbia. Microsoft Corporation: 1990-1992, Microsoft Windows 3.1 TM, Software, Redmond, Washington, DC. Mueller-Dombois, D. and Ellenberg, H.: 1974, Aims and Methods of Vegetation Ecology, John Wiley and Sons, New York. Pojar, J., Klinka, K. and Meidinger, D.V.: 1987, 'Biogeoclimatic ecosystem classification in British Columbia', For. Ecol. Manage. 22, 119-154. SAS Institute Inc.: 1982, SAS(c) System, Software, Cary, North Carolina. Sondheim, M.: 1983, 'Validation procedures, BCSIS - Volume 4', B.C. Min. For., Victoria, British Columbia. Sondheim, M. and Suttie, K.: 1983, 'User's manual for the British Columbia soil information system, BCSIS Volume 1', B.C. Min. For., Victoria, British Columbia. Sondheim, M., Void, T. and Quesnel, H.: 1983, 'Data entry procedures for ecosystem description forms, BCSIS - Volume 2', B.C. Min. For., Victoria, British Columbia. Walmsley, M., Utzig, G., Void, T., Moon, D. and Van Barneveld, J.: 1980, 'Describing ecosystems in the field', B.C. Min. Environ. and B.C. Min. For., Victoria, British Columbia. Wrenshall, C.: 1991, ECOMENU- A Mainframe FORTRAN Program to Assist in Vegetation Analysis, Software, B.C. Min. For., Victoria, British Columbia. -
