Phytolith evidence for the pastoral origins of multi-cropping in Mesopotamia (ancient Iraq)

Multi-cropping was vital for provisioning large population centers across ancient Eurasia. In Southwest Asia, multi-cropping, in which grain, fodder, or forage could be reliably cultivated during dry summer months, only became possible with the translocation of summer grains, like millet, from Africa and East Asia. Despite some textual sources suggesting millet cultivation as early as the third millennium BCE, the absence of robust archaeobotanical evidence for millet in semi-arid Mesopotamia (ancient Iraq) has led most archaeologists to conclude that millet was only grown in the region after the mid-first millennium BCE introduction of massive, state-sponsored irrigation systems. Here, we present the earliest micro-botanical evidence of the summer grain broomcorn millet (Panicum miliaceum) in Mesopotamia, identified using phytoliths in dung-rich sediments from Khani Masi, a mid-second millennium BCE site located in northern Iraq. Taphonomic factors associated with the region’s agro-pastoral systems have likely made millet challenging to recognize using conventional macrobotanical analyses, and millet may therefore have been more widespread and cultivated much earlier in Mesopotamia than is currently recognized. The evidence for pastoral-related multi-cropping in Bronze Age Mesopotamia provides an antecedent to first millennium BCE agricultural intensification and ties Mesopotamia into our rapidly evolving understanding of early Eurasian food globalization.


Supplementary Figures
. Graph Tables  Table S1. Comparison of sediment sample location, mineralogy, and micro-remain concentrations .......... 3  Table S2. Sources for archaeological sites in Fig. 1 Fig. S1. Table S1 contains the FTIR (Fourier Transform Infrared) spectroscopy assessment of sediment main mineral components, calcite types, and whether clays were altered by heat (burned) (see Additional methods below). Bioturbation was carefully avoided during sampling and well-defined horizontal layering indicates contextual integrity and minimal micro-remain translocation.
In general, phytolith preservation from Trench Y82 was good. Phytolith concentrations ranged from 0.4 to 45.9 million/gram of acid insoluble fraction (AIF) (median: 15.9 million/gram of AIF), percent weathered phytoliths was low (<10.1%), percent anatomically connected phytoliths (multicellular structures) was high (median 18.5%), delicate morphologies were present (median: 22.8%), and there were no indications of preservation trends by elevation ( Fig. S1) 1,2 . Sediments in trench Y82 alternate between unaltered geogenic fills low in micro-remain concentrations (Facies B) and sediments that are comparatively rich in organic content, micro-remains, and are burned (Facies A). Faecal spherulites were found in high concentrations in sediments with heat altered clays (Facies A: 28.1 ± 10.0 million/g sediment; SD), indicating layers contained discarded dung fuel or burned animal pen accumulations.

Additional Methods
FTIR: Sediment mineralogical analysis was performed using a Thermo Scientific Nicolet iS5 FT-IR Spectrometer in the 4000 and 400 cm -1 spectral range at 4cm -1 resolution. We mixed approximately 1 mg of sample with 80 mg of potassium bromide (KBr) in an agate mill. Main mineral components were determined using the wavelengths of the strongest absorption peaks following 3 and referencing the standards from the Kimmel Center for Archaeological Science, Weizmann Institute of Science. Calcite type was determined using the grinding curve method established by 4 . Clay thermal alteration (heating) was established following 5 as well as the new local thermal alteration references reported in 1 .

Phytolith concentrations and morphologies:
All samples were assessed for percent organic content using the loss-on-ignition method (550°C for 2 hours) 6 . Subsequently, the same samples were treated with 3N HCl following Albert and Weiner 7 to determine the acid insoluble fraction (AIF). Assessment of phytolith concentrations is in millions per gram of AIF. Table S1. Comparison of sediment sample location, mineralogy, and micro-remain concentrations. Main mineral components are ordered by relative peak height in each FTIR spectrum. Data in first seven (of nine) columns was previously reported by 1 .  Figure S1. Graph of relative abundances (%) of phytolith morphologies and micro-remain concentrations (million/g of AIF or sediment, respectively) in each sample arranged by elevation. SRP_2-20 is a surface control sample. Grass inflorescences are reported as overall percent and contain both wild and domesticated species. Selected data from 1 .     Table S5. 2 Undulation pattern η-type (levels I-III) 59 3

Sample Location Phase Elevation
Ending structure Cross "finger-type"     Figure S3. Map illustrating Euclidean distances from Khani Masi to the natural growing range of Panicum miliaceum (outlined dark gray area; >120mm precipitation May-October) 37,77 . The 47 km radius circle (green) indicates the maximum one-way travel distance for plant matter consumed by sheep/goats based on average herd speeds and digestion times 78 . The green line indicates the Euclidean distance to nearest area in which P. miliaceum can be cultivated with precipitation. Dashed lines show later, reconstructed silk road corridors (after 79 ). This figure was generated in Esri's ArcGIS 10.6.1 (http://www.esri.com/software/arcgis) using a digital elevation hillshade derived from 7.5-arc-second Global Multi-resolution Terrain Elevation Data 2010 (GMTED2010) courtesy of the USGS and Esri World Imagery (Sources: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community).