National Vegetation Information System Taxonomic Review

National Vegetation Information System
Centre for Plant Biodiversity Research
Department of the Environment and Heritage, 2004

1.0 Introduction

Floristic data is the key input used to standardise National Vegetation Information System (NVIS) vegetation descriptions. The Department of the Environment and Heritage (DEH), through the Environmental Resources Information Network (ERIN), has funded this project, which aims to assess taxonomic issues in NVIS and provide recommendations regarding improvement of its floristic data. The Centre for Pla nt Biodiversity Research (CPBR) was approached by ERIN to provide this taxonomic assessment. The CPBR assessed the taxonomic profile of the NVIS database and made recommendations in the form of a report. This report is for the consideration of ERIN management and NVIS collaborators.

1.1 Project brief

1.11 Aims and objectives

The four main outcomes in brief that ERIN was looking for from the CPBR were:

  1. to assess the extent of taxonomic and nomenclatural issues in NVIS and recommend changes;
  2. to develop and test measures of taxonomic and nomenclatural data quality in
  3. the NVIS database;
  4. review the structure of taxonomic parts of the NVIS database and provide advice and options on protocols and procedures for the continuous improvement of taxonomic data in NVIS; and
  5. provide guidance material for incorporation in future NVIS manuals.

In greater detail (see also Appendix A) this would involve:

  1. Assess the extent of any immediate taxonomic and nomenclatural issues in NVIS and recommend changes for the approx. 3,600 unique species in NVIS. Priority would be on, but not limited to, major/widespread groups such as: Eucalyptus, Acacia, Chenopodiaceae, Casuarinaceae, etc (see Appendix C).

It was envisaged that the proposed changes would be presented to NVIS collaborators for adoption.

  1. Develop and test measures to periodically assess the quality of taxonomic and nomenclatural data in NVIS overall and subdivided by State/Territory and major groups, for example:
    • Extent and nature of taxonomic inconsistencies; and
    • Extent of uncertainties to be solved by: (a) simple name change, (b) re-identification of voucher and (c) field visit.

It was envisaged that these measures could be applied to the re-supplied NVIS dataset, in the future – i.e. after completion of this project.

  1. Review the structure of taxonomic parts of the NVIS database and provide advice and options on protocols and procedures for the continuous improvement of taxonomic data in NVIS:
    • Review the measures developed in #2 and suggest a reporting frequency and thresholds for action;
    • Specify corrective actions that might be needed;
    • Examine the costs and benefits of identifying and recording infra-species for vegetation surveys;
    • Review the existing NVIS structure and content with respect to taxonomic data. Examine the role of separate jurisdictional lists (which typically hold many more species than are used in the NVIS vegetation descriptions) and how these should be maintained to support updates of NVIS descriptions;
    • Examine whether the Commonwealth needs to maintain a separate list of unresolved taxonomic issues between jurisdictions, to ensure the operational capability of NVIS; and
    • Review technical options for cooperative linkages between State and Territory survey agencies with national biodiversity information initiatives – such as: the Australia’s Virtual Herbarium, SPRAT and NVIS.

The project report would focus on identifying issues and discussing the pros and cons of various possible solutions.

  1. Provide guidance material for incorporation in future NVIS manuals:
    • Advise and provide details of current standards/best practice for databasing and interchange of species data, e.g. the Herbarium Information Standards and Protocols for Interchange of Data (HISPID); and
    • Provide details of current standards and/or best practice for the taxonomic aspects of vegetation survey – especially the use of vouchers.

1.12 Deliverables

  1. Document proposed changes to each taxonomic record (in newly-created fields so that existing State/Territory data are not changed) in the NVIS database, with clear comments as to reasons;
  2. A final report addressing the Aims and Objectives (Section 1.11). The report would be for circulation to NVIS collaborators; and
  3. Provide a briefing on the project to the NVIS collaborators at a national workshop and/or ESCAVI meeting, as appropriate.

1.2 Rationale

A recent ERIN review of species names in the NVIS 2000 database indicated that approximately 5% of the 3,600 unique species names used in NVIS to be non-current. The problems may have one or more sources, ranging from simple misspellings to the use of different taxonomic concepts by vegetation surveyors.

Australia’s Virtual Herbarium (AVH), a consortium of State, Territory and Commonwealth herbaria, is considering similar issues in creating equivalent taxa across Australia. It is timely to scope the options for beneficial interaction between herbaria and NVIS, including the development of sustainable systems and processes to manage taxonomic issues in NVIS.

Further rationale for the project is given in Section 1.3 below and in Appendix A.

1.3 Background

1.31 NVIS 2000 database

The NVIS 2000 database (Thackway et al. 2001) was collated by the Bureau of Rural Sciences for the National Land and Water Resources Audit (Audit) with collaboration of Commonwealth, State and Territory agencies across Australia.

NVIS provides a framework which:

  • specifies guidelines for the collection, compilation and monitoring of Australia's vegetation;
  • stores data on type and extent of vegetation;
  • provides an Australia-wide geographic and attribute information system for vegetation data that facilitates analysis and reporting;
  • provides access to a range of detailed and compatible mapped data sets; and
  • provides and maintains the technical infrastructure to support these activities.

This data set comprises two themes — Pre-European and Present (circa 1997) Native vegetation type and extent.

The NVIS database contains vegetation data from about 100 source datasets collected and mapped by States and Territories using a variety of methods over several decades. It is compiled from source data at a variety of scales and with varying attribution (Thackway et al. 2001; NLWRA 2001). The NVIS database is the result of the translating these data sets into the nationally consistent NVIS vegetation classification system (NLWRA 2000).

The NVIS vegetation classification system contains information on vegetation structure (growth form, height, cover) and floristics (genus and species). The NVIS information hierarchy summarises detailed vegetation association data at six levels of description. Level 1 (Class) is the most general description describing a single vegetation type (e.g. tree, tussock grass, etc.) while Level 6 (Sub-association) is the most complex, describing up to 5 vegetatio n strata/layers, 5 growth forms and 5 species per layer (ESCAVI 2003; NLWRA 2000). Table 1.31 provides details as to what structural and floristic components make up each of the NVIS hierarchical levels.

The complexity of vegetation descriptions supplied to NVIS varies between Levels 4 and 6. Most data in the NVIS 2000 database contain descriptions to Level 4, and some validation was completed across Australia for Levels 1 to 4..National Vegetation Information System Taxonomic Review

Table 1.31 The NVIS Information Hierarchy (ESCAVI 2003)
Hierarchical Level Description NVIS structural/floristic components required
L1 Class Dominant growth form for the ecologically or structurally dominant stratum
L2 Structural Formation Dominant growth form, cover and height for the ecologically or structurally dominant stratum.
L3 Broad Floristic Formation Dominant growth form, cover, height and dominant land cover genus for the upper most or the ecologically or structurally dominant stratum.
L4 Sub-Formation Dominant growth form, cover, height and dominant genus for each of the three traditional strata. (i.e. Upper, Mid and Ground)
L5 Association Dominant growth form, height, cover and species (3 species) for the three traditional strata. (i.e. Upper, Mid and Ground)
L6 Sub-Association Dominant growth form, height, cover and species (5 species) for all layers/sub-strata.

Levels 5 and 6 are the “complex” levels recommended for data compilation.

1.32 Restructured NVIS database

During 2001 to 2003, the NVIS 2000 database was restructured to make it much more

efficient, in terms of streamlining data updates and data validation, reducing the volume of data and ensuring faster query times for users. The main idea was to reduce redundancy in the database through normalisation and simplification; to ensure map units referenced a unique list of vegetation descriptions, rather than many variations on the same description. The NVIS collaborators have revised and updated the NVIS Framework to reflect the restructure and other agreed improvements (ESCAVI 2003).

Much work has also been done to validate Levels 5 and 6 in the Veg_Description table and lower level (Stratum, Taxon_Data and Growth_form) tables (see Appendix B). The emphasis was on ensuring consistent vegetation descriptions at all levels of the NVIS Information Hierarchy, as shown in Table 1.31.

The full NVIS database is stored as two components — a set of GIS spatial coverages in ArcInfo format and a full set of attribute information stored in an Oracle database. This is essential due to the complexity of both the line-work in the spatial coverages and the volume of attribute information. The relationship between the two is maintained by NVIS_ID codes identifying unique vegetation associations in both the spatial coverage and the attribute database. Vegetation mapping can be quite complex at the scales employed and a single polygon may contain up to six vegetation associations referred to as mosaics in vegetation mapping.

1.33 Taxonomic issues in the restructured NVIS database

Prior to this project, very little work has been done on validating the entries in the Taxon_Lists and Taxon_List_Origin tables (Appendix B). What validation had been done included comparison of species names in NVIS with the taxonomic lists in the DEH’s Species Profile and Threat (SPRAT) database. The SPRAT taxonomic module contains listings of Australia’s plant names that have been validly published. These published name lists are kept current by manual updates from the CPBR’s Australian Plant Name Index/What’s Its Name (APNI/WIN) database. The SPRAT database has fields to note whether taxa are “current”; this facility is used by DEH for implementation of the threatened species component of the Environment Protection and Biodiversity Conservation Act 1999 (Commonwealth), but has no other status. The validation thus checked:

  • spelling errors; and
  • indicative taxonomic “currency”.

Early validation work by ERIN indicated that the SPRAT database needed updating. Changes were made for approximately 50 species. ERIN then re-ran the validation and made the necessary spelling corrections in the vegetation descriptions, lower tables and in the Taxon_Lists tables. During this process statistics of spelling errors by State were not recorded. These (and other) suggested corrections have been agreed by most States and Territories and form a key basis for this review.

The comparison of NVIS species names with SPRAT “current” flag indicated that 206 (or approximately 5%) of the 3,916 unique taxon names used in NVIS to be non-current. In any particular case, this is probably the result of one or more of the following reasons:

  • nomenclatural inconsistencies, such as annotations to indicate doubtful identifications;
  • old vegetation survey data, using outdated species concepts;
  • differing species concepts between vegetation surveyors;
  • observer differences in their ability to identify plant specimens; and
  • errors in SPRAT.

The NVIS collaborators have not, however, given permission for the Commonwealth to change taxon names, because there is no one agreed national taxonomy for vascular plants in Australia. So no changes were made to the database as a result of the currency check. The error rate can be taken as indicative of taxonomic issues that need to be resolved by NVIS collaborators.

1.34 Taxon Lists in the restructured NVIS database

The Taxon_List_Origin table (see Appendix B) contains a description of the source of each taxon list, of which there are 20 in the NVIS 2000 database. The Taxon_Lists table (Appendix B) contains the detailed lists, with one row per taxon.

Neither the structure nor contents of the Taxon_Lists table have been changed since the NVIS 2000 database was collated. The original purpose of this table was as a lookup table to convert each taxon’s Source_Code in other parts of NVIS to a full scientific name. The table contains lists of taxa supplied by each State and Territory. In some States, this comprises a list of taxa for each NVIS input dataset and, in other States, the table contains multiple versions of a complete State list.

The end result is massive duplication and redundancy at the national level as shown in Table 1.34. Of the 59,507 rows (taxa) in the Taxon_Lists table, there are only 23,198 distinct taxa, thus representing considerable duplication. A total of only 5,159 taxa (rows) in the Taxon_Lists table are used in the vegetation descriptions, meaning that 54,348 (59,507-5,159) taxa are redundant. Of course some of these may be used to lookup taxa for new vegetation descriptions, but that is a lot of maintenance for a “just-in-case” scenario. It should also be noted that a total of only approximately 18,000 vascular taxa are present in Australian flora (Orchard, A. 2003, pers. comm., 31 October).

Table 1.34 Counts of taxa (species and infraspecies) in the NVIS Taxon_Lists table.
State/Source No of Vegetation Descriptions No of Unique Taxa Used in Vegetation Descriptions Total Number of Taxa
Commonwealth     73
ACT 343 78 564
NSW 969 843 4,776
NT 1,005 467 5,674
QLD 921 1,880 32,071
SA 538 694 5,444
TAS 169 341 2,766
VIC 568 856 5,829
WA   0* 2,306
Total   5,159 59,513

* WA data were provided at NVIS Level 4, which does not use species data.

When the total of 5,159 taxa were further investigated, only 3,916 were found to be distinct. This represents further duplication within the taxonomic component of the NVIS database and arises from inherent inefficiency in the database design and update protocols. Even further efficiencies might be gained from examining content issues for each taxon, for example:

  • standardising (or removing) notations for doubtful identifications and qualifiers — e.g. sp. aff., sens. lat., complex, etc.;
  • standardising taxonomy to an agreed national set of current names — meaning that some of the 3,916 species names would be “synonymised”; and
  • standardising (or removing) other attributes in the Taxon_Lists table.

CPBR is well equipped to advise ERIN and the NVIS partners on the above taxonomic content issues and some of the structural issues. It should be noted that at the time of the CPBR review, the number of records used to generate vegetation descriptions rose from 5,159 to 5,447. All CPBR analysis thus relates to this latter figure.