Abstract
<p>This protocol details the extraction and characterization of starch grains from a modern reference collection in order to enable direct comparison with archaeological specimens recovered from a wide range of contexts. The methodology follows a standardized and replicable workflow that includes procedures for sample selection, cleaning, drying, mechanical extraction, and slide preparation. Particular attention is given to modern contamination control and the consistent handling of materials throughout all stages of the process. Microscopic observation is conducted under multiple magnifications, allowing detailed documentation of granule morphology, including size, shape, hilum position, and surface features. Both stained and unstained preparations are analyzed to enhance the visibility of diagnostic characteristics. Notably, the application of Lugol’s solution to enhance specific morphological features, such as lamellae, represents a methodological advance that improves the resolution and interpretative potential of starch grain analysis in archaeobotany. The resulting data are systematically recorded and coded to support subsequent morphometric analysis and statistical comparison. This approach ensures consistency, reproducibility, and reliability in the identification and interpretation of starch grains across diverse archaeological settings. Introduction The archaeological analysis of starch grains is now a routine tool for reconstructing past plant use and processing in contexts where macrobotanical remains are poorly preserved, for identifying underrepresented plant taxa or organs (e.g. underground storage organs - USO) or investigating the use of tools such as grinding stones or lithic blades (Holst et al., 2007; Mercader, 2009; Henry et al., 2011; Nadel et al., 2012; Wadley et al., 2020). Starch grains can survive on artefacts, sediments, and dental calculus, providing direct evidence of plant processing and consumption that may otherwise remain invisible in the archaeological record (Torrence & Barton, 2006; Henry et al., 2008; Li et al., 2010; Horrocks et al., 2025). Modern reference collections of starch grains are fundamental by providing the comparative framework for taxonomic identification and the evaluation of morphological variability withinand between taxa (Pagán-Jiménez, 2015; Cagnato et al., 2021; Ahituv & Henry, 2022). Traditionally, starch grains are documented unstained, allowing the observation of the natural morphological characteristics such as size, shape, hilum position, fissures, and other diagnostic features used inarchaeological starch identification (ICSN, 2011; Coster et al., 2015; Kovárník & Beneš, 2018; Brown & Louderback, 2020). The present study expands this approach by developing a reference collection that systematically documents starch grains under both unstained and Lugol-stained conditions. Unstained observations preserve the natural morphology of starch granules, whereas Lugol’s iodine increases optical contrast; together, these approaches improve the observability of small and/or faintly light‑reflective starch grains. When Lugol´s solution is applied, it produces a rapid color reaction that indicates whether starch is present in significant amounts. This response results from the interaction between iodine and starch polymers, particularly amylose. It can be used to assess the likely visibility of starch grains under the microscope and estimate the time needed for their observation (Tester et al., 2004; BeMiller & Whistler, 2009; Perez & Bertoft, 2010). The comparison of stained and unstained specimens allows the assessment of how iodine treatment affects the visibility of diagnostic features and facilitates the recognition of internal structures that may be difficult to observe otherwise (Li & Wei, 2020; Gao et al., 2021). The resulting comparative vouchers of reference aim at improving identification reliability and contribute to greater methodological consistency in starch analyses across different archaeological contexts and laboratories (Liu et al., 2014; Mercader et al., 2018; Wan et al., 2020; Louderback et al., 2022).</p>