Two new barium-bearing minerals: gurimite, Ba3(VO4)2 (IMA2013-032) and hexacelsian, BaAl2Si2O8 (IMA2015-045) were discovered in veins of paralava cutting gehlenite-flamite hornfels located in the Gurim Anticline in the Negev Desert, Israel. Gurimite and hexacelsian occur in oval polymineralic inclusions in paralava and are associated with gehlenite, pseudowollastonite or wollastonite, rankinite, flamite, larnite, schorlomite, andradite, fluorapatite, fluorellestadite, kalsilite, cuspidine, aradite, zadovite and khesinite. Gurimite and hexacelsian form elongate crystals <10 μm thick. The minerals are colourless and transparent with a white streak and vitreous lustre, and have (0001) cleavage, respectively good in gurimite and very good in hexacelsian. Fracture is irregular. Density calculated using empirical formulas gave 5.044 g cm–3 for gurimite and 3.305 g cm–3 for hexacelsian. Mean refractive indexes, 1.945 and 1.561, respectively, were also calculated using the empirical formulas and the Gladstone-Dale relationship. The minerals are uniaxial and nonpleochroic. The following empirical crystal chemical formulae were assigned to holotype gurimite: (Ba2.794K0.092Ca0.084Na0.033Sr0.017)∑3.020(V1.8275+S0.0916+P0.0515+Al0.040Si0.005Fe0.0053+)∑2.017O8,and holotype hexacelsian: (Ba0.911K0.059Ca0.042Na0.010)∑1.022Al1.891Fe0.0723+Si2.034O8. The Raman spectrum of hexacelsian is similar to the one of the synthetic disordered β-BaAl2Si2O8. The Raman spectrum of gurimite is identical to that of synthetic Ba3(VO4)2. The electron back-scattered diffraction (EBSD) pattern of gurimite was fitted to the structure of its synthetic analogue with the cell parameters of R3m, a = 5.784(1),c = 21.132(1) Å, V = 612.2(2) Å3, Z = 3, giving a mean angular deviation = 0.43° (good fit). The Raman spectra of hexacelsian and its EBSD pattern suggest that natural hexacelsian corresponds to disordered synthetic β-hexacelsian P63/mcm, a = 5.2920(4) Å, c = 15.557(2) Å, α = β = 90°, γ = 120°. We suggest that after relatively fast crystallization of the main constituents of the paralava, gurimite, hexacelsian and also other Ba-bearing phases crystallized from residual melt enriched in incompatible elements that filled interstices between crystals of the main constituents.

In this response to the reply by Shahack-Gross and Finkelstein (2017), we present additional data of our research at Horvat Haluqim. This includes phytolith percentages and multicellular phytolith stomata in a thin section of a layer in Terraced Field 12, dated by radiocarbon (14C) to the Late Bronze–Early Iron Age. We also show thin-section evidence of aggrading sediment laminations in this terraced field. A new 14C date is given of the Early Islamic Period in Terraced Field 7, as differences in terrace wall architecture are highlighted. We revisit the interpretation by Shahack-Gross and Finkelstein in relation to herd management. Our 14C dates attest that terrace agriculture based on runoff/floodwater irrigation occurred in the Negev Highlands during several periods, including the Iron Age.

Shahack-Gross and Finkelstein (2015) further developed their theory, based on microarchaeology, that there was no agriculture in the Negev Highlands during the Iron Age. We critically evaluate their article in this rejoinder and propose that their conclusion is an example of overinterpretation from a small amount of indirect data. Based on phytoliths in two courtyards and a few rooms, i.e. structures not related to farming, they construed the absence of agriculture during the Iron Age in an area of 2000 km2. We present new radiocarbon, macroarchaeological, and microarchaeological data of Horvat Haluqim, showing that agriculture in the Negev Highlands based on runoff/floodwater capture and related terrace wall construction did not begin with the Roman–Byzantine period. Terrace agriculture in the Negev is older and includes also the Iron Age.

Soils in similar geomorphic settings in hyperarid deserts (< 50 mm yr−1) should have similar characteristics because a negative moisture balance controls their development. However, Reg soils in the hyperarid southern Negev and Namib deserts are distinctly different. Soils developed on stable alluvial surfaces with only direct input of rainfall and dust depend heavily on rainfall characteristics. Annual rainfall amount can be similar (15–30 mm), but storm duration can drastically alter Reg soil properties in deserts. The cooler fall/winter and dry hot summers of the southern Negev Desert with a predominance brief (≤ 1 day) rainstorms result in gypsic-saline soils without any calcic soil horizon. Although the Namib Desert receives only 50–60% of the southern Negev annual rainfall, its rainstorm duration is commonly 2–4 days. This improves leaching of the top soil under even lower annual rainfall amount and results in weeks-long grass cover. The long-term cumulative effect of these rare rain-grass relationships produces a calcic-gypsic-saline soil. The development of these different kinds of desert soils highlights the importance of daily to seasonal rainfall characteristics in influencing soil-moisture regime in deserts, and has important implications for the use of key desert soil properties as proxies in paleoclimatology.

Stone-built installations known as ‘leopard traps’ are found throughout the deserts of the Middle East. They have generally been considered to be recent in date, and to have been built by Bedouin or other local communities to trap carnivores that threatened their flocks. But how much older might they be? Survey in the hyper-arid ‘Uvda Valley of the southern Negev Desert in Israel discovered 23 ‘leopard traps’, 19 of them clustered in a relatively small area. This study describes the architecture and function of these structures and presents the first optically stimulated luminescence ages for two of them. These results demonstrate that the traps are ancient and were already in use before the late fourth millennium BC, not long after the adoption of herding by the desert dwellers.

This study presents an assessment of the potential application of Mn content in rock varnish laminae as a paleoclimate indicator. To investigate the environmental controls on varnish formation, we determined Mn composition in rock varnish formed on flint artifacts produced during the earliest Holocene from eight coeval prehistoric sites in the Negev desert, Israel. These sites lie along a north–south annual rainfall transect ranging between 120 and 30 mm yr− 1. The varnish is ~ 100 times enriched in Mn relative to the content in the desert dust source material. Chemical profiles across the varnish display 4–6 distinct Mn peaks in all sampled sites, pointing to systematic fluctuations within the varnish along a wide range of environmental settings. The mean Mn contents in the various sites range between 10.7 and 15.6 at.%, yet within this range, the Mn content in the Negev varnish does not show a correlation with mean annual rainfall. As moisture is needed for Mn mobility, wetting cycles by dew or light rain, which are not adequately represented by the mean annual rainfall amounts but control the number of wetting–drying cycles may explain the variance within the results from the arid and hyperarid Negev varnish.