PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • Late prehistoric sites from a large portion of the eastern Po Valley and Venetian Plain were chosen for this study (S1 Table). The study area was limited to these recent valleys and their borders, in order to compare sites subject to similar climatic conditions. In order to focus on the economic use of animals, only assemblages from domestic settlement contexts were considered in analysis of taxon presence/abundance: sanctuaries, cult sites, and ritual deposits were excluded. Ritual assemblages often have a different distribution of taxa from those found in domestic contexts, e.g. a pronounced focus on a single species [35] (e.g. [106, 107]), which could bias results. Sites are divided into regional groups based on their location within the study area and regional similarities in material culture. Faunal samples with very long chronologies that could not be assigned to a useful temporal range were excluded. Only dated contexts were analysed; surface finds and materials recovered from plowzone were not considered. Despite efforts for good regional and temporal coverage, faunal assemblages are not evenly distributed across the study areas and time periods. The assemblages also vary greatly in size, from a few dozen to tens-of-thousands of bones. Information on some periods is entirely lacking, while others are represented only by a single site. This large range results from differences in site type, degree of preservation, ancient settlement patterns, and historical interests. Modern development obscures Iron Age sites in particular, as many major settlements of this period are buried under modern–as well as medieval and Roman–cities. Although dividing the study area into smaller sub-regions limited the size of comparative samples, such division was necessary for investigating differences within the Po Basin. Illustrations and tables have been organised to clearly demonstrate where gaps are present. Quantifying taxonomic abundance and animal size: The relative proportions of cattle (Bos taurus), sheep/goat (Ovis aries/Capra hircus), and pigs (Sus scrofa) were quantified using the number of identified specimens (NISP). Although subject to several systematic biases, NISP offers a widely employed and easy to calculate method of assessing the relative frequency of different species [108]. Inter-observer differences in identification and recording are a primary concern [109–111], as they artificially inflate the importance of taxa with a greater number of ‘identifiable’ remains. In order to better control of differences in recording practice and identification skill, rib, and vertebrate fragments were excluded from NISP counts where possible, as were any specimens not identified as a specific skeletal element. Complete and partial skeletons were also excluded from these analyses, since their inclusion would inflate NISP counts. In order to apply statistical tests with a reasonable degree of confidence, only assemblages with a minimum of 110 specimens identified to the three main domesticates were considered in NISP analyses. The statistical significance of inter-regional differences was tested using a chi-squared test in R software [112]. Changes in animal size were assessed using size-index scaled Log Standard Index values (LSI) [113]. Widely available standards were used for cattle [114], sheep and goats [115], and pigs [116]. Log ratios offer a useful method for investigating animal size, because they allow measurements from different elements to be pooled into a larger sample of indices for each dimension (length and breadth). However, this method of grouping can lead to the overrepresentation of specimens that yield several measurements, a problem that is quickly compounded if a skeleton or articulating limb is included. To avoid over representation of articulating remains, our LSI analyses included only one measurement per specimen and one specimen from an articulating group of bones. Bone widths were considered separately from lengths in order to assess two-dimensional change. Measurements of bone depth were comparatively rare and were not included. Table 1 presents the measurements used in LSI analyses in preferential order. Only one length or breadth measurement per specimen was considered, e.g. if GLl was available for a specimen, all subsequent length measurements would be excluded. This list purposefully excludes elements and measurements particularly sensitive to animal age, although it is also constrained by measurements included in the published standards. LSI values for cattle, sheep, and pigs are presented in histograms. Because accurate species distinction can be challenging for closely related taxa like sheep and goats [117], related species are also included: figures for sheep also display counts and means for goats and undistinguished sheep/goat specimens; figures for pigs also contain data identified as wild boar. Undistinguished Sus sp. were grouped with domestic pigs. The Mann-Whitney U Test, conducted in R, was used to identify statistically significant differences between populations. Only LSI values from cattle, sheep, and pig were subject to statistical analysis; goats were excluded due to a lack of reliable biometric information. Table data removed from full text. Table identifier and caption: 10.1371/journal.pone.0208109.t001 Measurements used in calculation of LSI values for each taxon. Mc = Metacarpal. Mt = Metatarsal. Metacarpal measurements were used to investigate diachronic changes in sex ratios for cattle and sheep. The balance of males, females, and castrates within a population can influence observed variation in animal biometry, especially if the sexual composition of the group changed over time. Metacarpals are the most sexually dimorphic element in domestic bovids [119–121], and consideration of metacarpal size and shape can provide information on sex representation [122, 123]. To aid interpretation, measurements were compared to data from animals of known sex: archaeological cattle from Eketorp ringfort [123], identified using DNA, and modern Shetland sheep [115, 124].
rdf:type