STARCH : QUANTIFICATION
Counting and quantification procedures
The individual starch granule is the typical unit of quantification in starch analysis, although large clusters of granules may be counted separately, particularly in cases where it is not possible to accurately count the number of granules in the cluster. Because ancient starch is usually recovered in fairly low quantities, it is common for all granules on a slide to be counted, although some researchers may stop counting a particular starch type once an arbitrary number has been reached, e.g. 100 granules (Perry 2007; Barton 2007). Alternatively, a maximum number of granules may be counted on a slide, regardless of type diversity. Qualitative scales or estimates of abundance (e.g. rare, common, abundant) may be used, such as during quick slide scans for diagnostic starch types. The addition of known quantities of exotic spore markers such as Lycopodium tablets to calculus or coprolite samples prior to processing can also be used to estimate the concentration of recovered starch and other microfossils (e.g. Boyadjin et al. 2007; Vinton et al. 2009; Wesolowski et al. 2010).
Once raw counts of the starches present in a sample have been made, a variety of quantification indices can be used to describe the ubiquity, relative abundance, density and diversity of starch types present. The choice of quantification method will vary according to the research objectives and the types of data analysis that is required to interpret the starch assemblages.
Presence or ubiquity
The number of samples in which a starch type occurs, regardless of how much starch is actually present in each sample, e.g. 17 of 25 grindstones (68%) had wheat-type starch. This method is commonly used in starch analysis to give a broad indication of how often different taxa occur on artefacts or in different contexts, which can be easily compared within and between sites. This type of analysis is less susceptible to quantitative bias from differential deposition, preservation or recovery, as the absolute number of granules in each sample is not compared.
This is based on absolute counts of granules of a particular type present divided by the total number of granules counted. This method gives an indication of how much or what fraction of an assemblage a particular starch type represents; often expressed as a percentage (e.g. 80% of the starch granules recovered from grindstone #21B were wheat-type). Assemblages from samples of different original size can be compared (e.g. wheat-type starch granules comprised 80% of the starch assemblage from context A but only 10% of the assemblage from context B). Comparisons of assemblage compositions between different contexts are less susceptible to differential preservation (cf. density ratios, below).
These are based on absolute counts of starch types per standard measure (e.g. weight, volume, surface area of artefact sampled). This approach enables counts across samples of different original sizes to be standardised for comparison. e.g. Wheat-type granules were recovered from the surface of the artefact at rate of 100 granules/gram of surface extract, but were present in the surrounding sediment at a rate of only 3 granules/gram. Comparisons of starch densities between different contexts should account for unequal preservation conditions. For density ratios, the amount of sample analysed should be measured so that it can be standardised across all samples. The most commonly used are weight (which is used for sediment samples, artefact residue extracts, charred remains and dental calculus) and sampled surface area (which has been used for artifacts, e.g. Mercader et al. 2008, and dental calculus, e.g. Henry and Piperno 2008). These should be accurately measured prior to extraction. In some cases, the dry-weight of the final extract is also recorded prior to microscopy.
The number of different starch types present, which may indicate, for example, whether an artefact was multipurpose (many starch types present) or used to process one plant type (e.g. Berman and Pearsall 2008).
Also keep in mind that sub-sampling decisions such as how much extract to mount on a slide and the number of slides to analyse per sample may affect quantitative comparisons between samples and should therefore be standardised.
Boyadjin et al. 2007; Vinton et al. 2009; Wesolowski et al. 2010; Mercader et al. 2008; Henry and Piperno 2008; Berman and Pearsall 2008