Starch identification methods

Archaeological starch granules are identified by carefully examining a range of morphological, dimensional and optical features under different microscope lighting conditions and comparing them with granules from modern reference collections (see examples from the University of Sheffield Starch Reference Collection). These features include:

Most of the morphological features used to identify starch granules are best observed in regular transmitted light, although some analysts recommend using differential interference contrast (DIC) or phase contrast to improve observation of internal structure (Barton and Fullagar 2006). Cross-polarised light is used to examine the location and shape of the extinction cross and to assess the degree of birefringence. Magnifications between x400 and x800 are usually required to characterise starch morphological features accurately, although very small granules may require x1000 magnifications. In these cases, it is best to use immersion oil objective lenses to improve resolution and clarity. Starch granule size can be measured using a calibrated graticule located in the microscope eye piece or using image analysis software. Features such as three-dimensional shape, hilum position, and the form of fissures can change depending on granule orientation, so starches should always be rotated on the slide to characterise and identify them accurately (see film on right). This effect can be achieved by gently applying pressure to the coverslip, such as with a plastic pipette tip, although be careful not to press too hard as the starch granule may move out of the field of view or could be damaged.

Starch granules can also be differentiated based on their physicochemical properties, which vary between taxa depending on granule structure and composition. These properties affect the rate at which granules are hydrolysed by chemicals and biological enzymes, the temperature at which they gelatinise (become amorphous in the presence of heat and moisture), and the colour of their staining reaction with iodine (Hall et al. 1989; Reichert 1913; Torrence 2006a). These tests are rarely employed in archaeological studies, however, as they are usually destructive and their accuracy may be affected if the test conditions are not controlled properly and if granule structure and composition change through time, which is possible under certain conditions.

Starch taxonomic identification

The taxonomic level to which archaeological starches can be identified varies depending on the distinctiveness of granules from a particular taxon as well as the condition of the granules (identifications may not be possible if they are very damaged). It is critical to build a comprehensive reference collection that includes all wild and cultivated edible taxa that occur in the study region as well other economic and non-economic plants whose starches might also be present in the archaeological record. The degree of variation within and overlap between taxa can then be confidently assessed and diagnostic starch morphotypes for individual or groups of plants can be identified. Related taxa often produce similar starch types, (e.g. legumes; Triticeae [wheat tribe] cereals), which can enable at least family, tribe or genus level identifications. Because similar starches can also occur in unrelated taxa, identifications are sometimes less straightforward. Taxa can often be discriminated at higher levels of specificity using a detailed combination of size ranges, internal and external attributes, and their relative frequencies within a population (for multiple grain analyses of population signatures, see Holst et al. 2007; Piperno and Dillehay 2008; Piperno et al. 2004). Multivariate statistical tools such as discriminant function analysis can be used to test the discriminatory power of these characteristics and for subsequent classification of archaeological granules (Torrence et al. 2004). If population characteristics are used for starch identification, a sufficiently large number of archaeological granules must also be recovered to provide a representative assemblage for confident identifications.

Starches identified with the highest confidence are given binomial species or genus names. In these cases, the archaeological and comparative reference starches match in all morphometric attributes and are consistent with the regional flora and geographic area (Paz and Barton 2007). The qualifier ‘cf.’ is used to indicate tentative identifications, where an archaeological starch granule is considered likely to derive from a particular taxon but a secure identification was prevented because, for example, granules could not be rotated to view in three dimension, some diagnostic features were absent, or the granule was damaged (Perry 2007:245). Barton and Paz (2007) also describe a detailed scheme for indicating levels of confidence for starch identification that is particularly useful when reference collections are incomplete or augmented by published descriptions.

Archaeological context can also play a key role in the final identification of most parsimonious botanical source. The range of possibilities can often be narrowed down by considering factors such as ecological and geographical setting, site type and time period, specific activity area, and whether the starches are found as use-residues on artefacts (which are more likely to be used for processing economic plants). If a process of elimination is used, you should make it very explicit how you arrived at that identification.

Starch morphotype classification

Archaeological starches that do not occur in the reference collection can still be classified into morphotypes based on a common set of shared features (e.g. Barton 2005; Lentfer et al. 2002). In these cases, granules are identified by an arbitrary name, for example, Type 1, Type 2, etc.. This approach is particularly useful during exploratory studies in areas where comprehensive reference collections are still being established. The range of starch types present can still be determined (although species diversity should be interpreted cautiously owing to the potential for morphotype redundancy or multiplicity) and assemblage compositions can be compared between samples to investigate variability in the starch record (e.g. Lentfer et al. 2002). Often, a combination of approaches to starch identification and classification is used, whereby diagnostic types (i.e. those known to occur specifically and uniquely within a taxon), non-diagnostic types (i.e. generic types present in many taxa) and morphotypes not previously identified in the reference collection are identified and described within an archaeological assemblage.

Preparing and analysing starch reference collections

Reference materials can be obtained from a variety of sources, including herbaria, seed reference banks, local markets and during field trips. If plant materials are collected in the field, a dried voucher specimen should also be prepared so that it can be accurately identified by a botanist or at a herbarium (see Field 2006 for further details). Plant materials collected in the field can be placed in small, plastic vials with 70% alcohol or dried so that they preserve. Where possible, collect samples from as many parts of the plant as possible including leaves, stems, inflorescences, roots, as well as the starchy organ (seed, fruit, tuber, etc.), so that the range of starch types (both storage and transitory) produced by a plant can be determined (Lentfer 2009). Always keep a record of relevant provenience information for reference samples, including the date, location where the sample was collected (GPS co-ordinates are useful), and name of the collector.

A simple method for processing reference materials is shown below, whereby a sub-sample is homogenized in a mortar and pestle and preserved in 70% alcohol. Aliquots can removed for slide preparation as needed. Alternatively, starch samples from dissected tubers, seeds and nuts can be scraped directly onto a slide with a scalpel, although individual granules can sometimes be difficult to observe by this method as the starches are not disrupted as effectively from the tissue.