Data-informed sampling and mapping: an approach to ensure plot-based classifications locate, classify and map rare and restricted vegetation types

Abstract

A new approach to vegetation sample selection, classification and mapping is described that accounts for rare and restricted vegetation communities. The new method (data-informed sampling and mapping: D-iSM) builds on traditional preferential sampling and was developed to guide conservation and land-use planning. It combines saturation coverage of vegetation point data with a preferential sampling design to produce locally accurate vegetation classifications and maps. Many existing techniques rely entirely or in part on random sampling, modelling against environmental variables, or on assumptions that photo-patterns detected through aerial photographic interpretation or physical landscape features can be attributed to a specific vegetation type. D-iSM uses ground data to inform both classification and mapping phases of a project. The approach is particularly suited to local- and regional-scale situations where disputes between conservation and development often lead to poor planning decisions, as well as in circumstances where highly restricted vegetation types occur within a wider mosaic of more common communities. Benefits of the D-iSM approach include more efficient and more representative floristic sampling, more realistic and repeatable classifications, increased user accuracy in vegetation mapping and increased ability to detect and map rare vegetation communities. Case studies are presented to illustrate the method in real-world classification and mapping projects.

Australian advances in vegetation classification and the need for a national, science-based approach

“…Explicitly modelling processes such as physiology, dispersal, demography and biotic interactions is believed to provide more robust predictions in species distribution models, particularly when extrapolating to novel conditions (Wisz et al 2013;le Roux et al 2014;Briscoe et al 2019). Surveying at this scale enables species discovery, the identification of rare species (Patykowski et al 2021) and mapping of rare and threatened plant communities (Tierney et al 2018;Bell and Driscoll 2021). Subsequently, the lack of national standards for data collection, analysis and classification development at this scale has serious negative implications for conservation decision making.…”

“…Six macrogroups (ecoregions) were identified for NSW (802 000 km 2 ), providing a practical map product for conservation decision-making despite data-poor areas. Bell and Driscoll (2021) outline a new method (data-informed sampling and mapping: D-iSM) for vegetation mapping that ensures that plot-based classifications identify rare and restricted vegetation types. Hunter (2021aHunter ( , 2021b highlights the importance of the temporal dimension in defining communities sensitive to inter-annual variation, e.g.…”

“…For example, a forestry classification may focus on commercial tree species and attributes such as height (Sun et al 1997). Alternatively, conservation decision-making requires an accurate assessment of understory and rare species (Tierney et al 2018;Bell and Driscoll 2021), whereas vegetation mapping focuses on key indicator species in the dominant strata, and the inclusion of rare species may hinder rather than inform mapping (Addicott et al 2018). Across all forms of classification, scale is an essential consideration for appropriate data collection and applications (Levin 1992;McKenzie et al 2008;Bell and Driscoll 2021;Hunter and Growns 2021;Patykowski et al 2021).…”