No agreement on what constitutes a safe and reproducible anticontamination protocol exists for ancient starch research. Protocols applied to laboratory work may represent 'symptomatic treatment' only, as contamination of archaeological materials in the field may be more extensive than realized. This paper is the first systematic study on the impact that modern starches from surface and buried soils, windborne dispersal, human motion, excavation techniques and toolkits, and field attire has on archaeological sample quality. The study area is Olduvai Gorge, Tanzania. We identify seven starch types (discrete granules, n = 788) that embody the starch contamination landscape for the region. This study also demonstrates the various diagenetic changes that buried starch granules undergo in a short time, such as cavitation, fissuring, disruption and gelatinization. There are significant differences in morphotype class representation between the topsoil starches and those collected deeper below ground at excavated sites. Diagenetically transformed granules from underground storage organs dominate in soils, while native starches from cereal endosperm (Panicoideae and Triticeae) abound above ground in airborne samples. Furthermore, we illustrate how lithic samples excavated under standard field conditions can be contaminated, and that when a sample is compromised during excavation, it may be impossible to distinguish between target and introduced starches, especially when granules are identical or morphologically similar. The paper provides field recommendations to control false positives.

The assumption that taxonomy can be ascertained by starch granule shape and size has persisted since the late nineteenth and early twentieth century biochemistry. More recent work has established that granule morphological affinity is scattered throughout phylogenetic branches, morphotype proportions vary within the genus, granules from closely related genera can differ dramatically in shape, and size variations do not reflect phylogenetic relationships. This situation is confounded by polymorphism at the species and tissue level, resulting in redundancy and multiplicity. This paper classifies morphological features of starch granules from 77 species, 31 families, and 22 orders across three African ecoregions. This is the largest starch reference collection published to date, rendering the dataset uniquely well-suited to explore (i) the diagnostic power of unique morphometric classifiers and their frequency, (ii) morphotypes that cut across taxonomic boundaries, and (iii) issues surrounding the minimum counts needed to accurately reflect granule polymorphism, variability, and identification. In a collection of 23,100 granules, taxonomic identification occurred very rarely. In the instances it did, it was at the species level, with no occurrences of a single morphotype or complement identifying all species within a family or genus. Some families cannot be uniquely identified, and morphometric types are shared despite taxonomic distance for three quarters of the taxa. However, this reference collection boasts 98 unique identifiers located in the Arecaceae,