A modern phytolith reference collection for selected native Australian plants: Implications for vegetation reconstruction

Abstract

Background and aims Phytolith analysis is widely applied in palaeoecological and archaeological research, but its interpretive strength depends on the availability of robust modern reference collections. This study expands the modern Australian phytolith reference collection by analysing 42 native plant species representing 24 families and 37 genera with emphasis on silicification patterns across major growth forms, including forbs, shrubs, trees, and C3 grasses. Methods Phytoliths were extracted from available plant parts, including leaves, stems, flowers, seeds, seed pods, cones, and roots, depending on sample availability. Morphotypes were identified following ICPN 2.0, with grass silica short cell phytoliths (GSSCPs) further classified by shape and size to examine subfamily-level patterns. Phytolith morphotype percentage data were analysed using Hellinger transformation, PerMANOVA, PCA, LDA, and hierarchical clustering to assess compositional differences among plant growth forms and grass subfamilies. Key results Phytolith production varied strongly among growth forms and plant parts. Grasses were abundant producers, whereas most forbs, shrubs, and trees were trace producers or non-producers. Leaves were the most consistent source of phytoliths, while seeds and seed pods were predominantly non-producers. Grass silica short-cell phytolith (GSSCP) morphotypes showed clear subfamily-level differentiation. Pooideae produced Rondel morphotypes. Danthonoideae produced Rondel as well as wide Bilobate types. Panicoideae and Oryzoideae exhibited a pronounced Bilobate signature, commonly associated with Polylobate and Cross forms. Non-grass taxa (woody, shrubs, and forbs) were dominated by Spheroids, Tracheary elements, Epidermal, and Polygonal sheets and other non-diagnostic forms. Phytolith assemblages differed significantly among plant families, with Poaceae uniquely producing GSSCPs, while non-grass families showed greater overlap in assemblage composition. Conclusion By expanding taxonomic and anatomical coverage, this study strengthens the capabilities of phytoliths in the reconstruction of grasslands and in general paleo vegetation in Australia, especially where other proxies such as pollen are limited.

Code and data availability

The authors explicitly state that the R scripts used for data analysis and figure generation are publicly available on their GitHub repository, which directly reproduces this paper's phytolith statistical analyses (PCA, LDA, PerMANOVA, clustering, plots). Supplementary data files contain the paper's measurements but no