Human stature in the Near East and Europe ca. 10,000–1000 BC: its spatiotemporal development in a Bayesian errors-in-variables model

A sample of 6098 published prehistoric skeletons consisting of long bone lengths, stature estimated from them using three different methods, as well as recalculated stature data created with other methods, was used to model tempo-spatial variance of stature in the Holocene prehistory of the Near East and Europe. Bayesian additive mixed modeling with errors-in-variables was applied, fitting a global spatiotemporal trend using a tensor product spline approach, a local random effect for the archaeological sites and corrections for mismeasurement and misclassification of covariates to obtain stature isoline maps for various time slices and diachronic stature trend curves for various regions. Models calculated for maximum long bone lengths and for stature are all largely consistent with each other, so Bayesian errors-in-variables models can be regarded as a viable means of smoothing regional and temporal variance in skeletal data as well as in estimation methods so that only robust trends become manifest. In addition to a general northwest-southeast gradient in stature, tallest stature in Eurasia and declining stature in Iberia confirms archaeogenetic insights. Transition to farming shows stable, decreasing, or even increasing stature depending on the region and the mode of Neolithization, putting into question the common assumption of a general negative effect of Neolithic lifeways on physical health. Particularly, Northern Europe experienced a rise in stature after the 4th millennium BC. Likely caused by both genetics as well as generally improving living conditions, our findings date the origin of the modern NW-SE gradient in stature to around 3000 BC.