South Arabia’s prehistoric monument landscape shows social resilience to climate change

PLOS One | https://doi.org/10.1371/journal.pone.0323544 May 28, 2025 6 / 24 With excavations and radiocarbon assays in Dhofar, we refined an earlier typology of monuments [7,9,12,14] and iden- tified details of construction. Subsequent data cleaning resulted in a data set of 371 surveyed monuments and the survey tracks. (S1-Table) Although fieldwork generated an initial array of 428 observations, some monuments were too poorly preserved to assign to a chronological-cultural type from surface inspection. For those that we could assign, we used the following visual criteria. Platforms have large blocks or upright slabs forming an outer base and an overall D-shaped or oval plan. Often the largest uprights form an eastern face. Platforms have a fill of smaller stones, and typically also have a (sometimes worked) standing stone set further to the east. (Excavations inside show that the uprights were supported by interior chock stones, in turn held in place by fill.) To balance and support stones and consolidate them into a free- standing, stable platform required rapid (i.e., episodic) initial construction. An HCT appears as a conical cairn 3–5 m across with a depressed center. This depression denotes an interior (usually collapsed) chamber made from upright or stacked slabs. Usually some of the chamber facing is still visible without excavation. Tomb (HCT) construction was surely also episodic, given the exigency of burial in a hot environment populated by scavengers and carnivores. A SCAB is also

a conical structure, created by stacking boulders in concentric daises decreasing in diameter with the height of the mon- ument. Triliths have clusters of 1–3 upright stones supported by a very low, oblong, gravel platform outlined by cobbles. It would be difficult to mistake this distinctive monument for any other type (Fig 1). Each trilith platform and its adjacent hearths is repeated along an alignment. We identified boat graves by their shape in plan and by the clustering of multiple, abutting, boat-shaped elements (Fig 1), each of which likely included a burial. In the cases of agglomerated triliths and boat graves, there is no construction characteristic that requires completion in a single episode. Triliths and boat graves could and sometimes did appear as isolated, unreplicated elements; this indicates that conglomerated examples were accretive, potentially completed over many years. After analysis of excavated examples, it was possible to reassign some monuments from “unknown” and “tumulus” field designations. We reviewed original field notes and photographs for each monument, and these aided in revisions— based also on 10 years’ experience--of other typological classifications. For some monuments, there were no measure- ments of stone size and volume, leading to their exclusion from our data. We also excluded rare types (e.g., Haluf tombs n = 3, Wall tombs n = 5) from analysis, narrowing the study to the five categories that represent 90 percent of documented monuments.

Chronological and stratigraphic methods. Altogether from Hadramawt and Dhofar we obtained a dated sample of 31 monuments (47 radiocarbon determinations) to complement existing literature. The number of radiocarbon determinations falls far short of reasonable sample sizes for summed probability frequency distributions, which are problematic even in large samples [69,70]. Instead, we note that our Bayesian analysis in Hadramawt using regional geomorphology priors there provides a framework consistent with the Dhofar radiocarbon determinations [38]. Radiocarbon determinations come from excavated samples with varying potential to date the initial construction of a monument. Sometimes dated samples are from bone (human and faunal) and determine when an organism died. Other samples are carbon from charcoal, buried humus, or shell. We used stratigraphic associations to contextualize how the deposition of such samples relates to the actual event of monument construction. With multiple monuments of each type, we documented construction materials and techniques, preservation, use, and re-use, including subsequent burials and sacrifices, evidence of collapse, additions, and subsequently quarried stone [16,52,71–73]. The probability determina- tions in Fig 2 represent events prior or posterior to monument construction. Also, as a responsible reporting of results, we include a radiocarbon determination we think has unexplained error (D033-001). Another determination (D014-002) also

lies outside our expected range. (For stratigraphic details see [14,52]). More samples would surely refine and constrain our chronology, but we caution that excavation is a costly and sometimes unsuccessful path to obtain contextualized radiocar- bon samples. Nevertheless, excavation remains the only viable strategy for a classification anchored in absolute dates. Our classification and chronology largely correlate with observations across Arabia [3,13,15,17–20,34,39,40,74]. While multiple researchers have—like us—found episodes of re-use in existing monuments, we focus on the pattern of