Whilst the stratigraphy of the marine deposits of the Arabian Peninsula has been reasonably stable on account of the presence of abundant marine molluscs, foraminiferans and other fossils in them, there is much less consensus concerning the terrestrial (continental) sediments (Fig. 1), with the literature showing large offsets between the estimated ages of the sedimentary units and their contained faunas. For example, the Hofuf Formation (Thralls & Hassan, 1956), was correlated by Al-Saad et al. (2002) to the Late Miocene to Pleistocene and by Sharland et al. (2004) to the Tortonian-Pleistocene, yet at Al Jadidah, close to the type area, it has yielded characteristic late Middle Miocene faunal remains aged between 14 Ma and 12 Ma (Anonymous, 1975; Hamilton et al. 1978; López-Antoñanzas, 2004, 2009; Morales et al. 1987; Thomas, 1982, 1983, 1985; Thomas et al. 1978; Sen & Thomas, 1979; Whitmore, 1987) and the fossiliferous deposits thus correlate to the Serravallian. This inconsistency raises the possibility of a mis-correlation between the so-called Hofuf Formation in Qatar (Cavelier, 1975; Al-Saad et al. 2002) and the Hofuf Formation in its type area in Saudi Arabia (Thralls & Hassan, 1956; Sharland et al. 2004), or indicates that, as currently understood, it comprises a composite unit of two or more formations.

An overview of the available evidence concerning vertebrate palaeontology of the Arabian Peninsula reveals that there are diverse localities which have yielded fossils of terrestrial animals and plants, spanning the period from Late Eocene to Recent (with gaps) but none are yet known from the period spanning the Late Cretaceous to Bartonian (Figs 1, 2).

Several named formations of continental deposits in Oman (e.g. Barzaman Fm, Ghaba Fm, Roger et al. 1994) have not yielded fossils, so there is inherently some uncertainty about their ages. One such unit which previously yielded only charophyte zoogonia, is the Marsawdad Formation which crops out widely in the Rub Al-Khali, Oman (Chevrel et al. 1992; Platel & Berthiaux, 1992a, 1992b). This unit was correlated to the Late Miocene-Pliocene by the authors because, in its type area, it overlies the Dam Formation (Burdigalian) and the Montasar Formation (which overlies the Dam Formation), and underlies supposed Pliocene to Recent deposits. Elsewhere in the Rub’ Al-Khali, the Marsawdad Formation overlies the Dammam Formation (Eocene) and Dawkah Formation (possibly Oligocene to Burdigalian). Chevrel et al. (1992) considered that the Marsawdad Formation may be equivalent in part to the Hofuf Formation of Saudi Arabia (Powers, 1968; Cavelier, 1975) but fossil mammals from several localities near the type area of the latter unit indicate that it is of Middle Miocene age, and thus substantially older than the Marsawdad sediments.

The Marsawdad Formation crops out extensively in the Rub’ Al-Khali, Oman (and probably also in Saudi Arabia) with outcrops mapped in three of the 1: 250,000 geological map sheets of Oman (Sheets, NE 40-01, NE 39-04 and NE 39-08; Chevrel et al. 1992; Platel & Berthiaux, 1992a, 1992b) (Fig. 3). The Marsawdad Formation (map symbol MPlms) is comprised of well-bedded sequences of reddish to yellowish clayey siltstone grading upwards to grey marly limestone and grey to brown micritic limestone with palaeosols. In several outcrops the surface exposures comprise tilted and folded beds (Fig. 4), probably resulting from dissolution of gypsiferous underlying strata accompanied by localised slumping, let-down structures and solution collapse (Le Blanc, 2009, figs 5.5 and 5.7).

Platel & Berthiaux (1992a) interpreted the depositional environment of the Marsawdad Formation as a ‘large continental basin surrounded by palustrine and, more rarely, lacustrine deposits, regularly encroached by fluvio-deltaic fans and very fine-grained detrital material that had been transported over long distances’. West of Montasar the formation is up to 100 metres thick, comprising 15 or more sequences 3-6 metres thick. The base of the succession is reported to be an iron-stained calcrete or a brown limestone palaeosol which locally has a travertine-like aspect.

The only fossils previously recorded from the Marsawdad Formation are charophyte oogonia (Platel & Berthiaux, 1992a) and two eggshell fragments that were collected from deflated surfaces of the formation by Rosén et al. (2021).

Research in Namibia during the past three decades has led to the establishment of a biostratigraphic scale employing struthious eggshells (Figs 5, 9) (Pickford, 1998, 2014; Pickford & Dauphin, 1993; Pickford & Senut, 2000; Pickford et al. 1995; Ségalen et al. 2002; Senut, 2000; Senut & Pickford, 1995; Senut et al. 1994, 1998). The scale spans the entire Neogene Period, the ages of the eggshell morphotypes being based on mammal fossils found associated with the eggshells (Pickford, 2014). Studies on fossil struthious eggshells from other African countries (Kenya, Harris & Leakey, 2003; Tanzania, Harrison & Msuya, 2005; Algeria, Arambourg, 1959; Malawi, Stidham, 2004; South Africa, Stidham, 2008) and the Arabian Peninsula (United Arab Emirates, Bibi et al. 2006; Mikhailov & Zelenkov, 2020) which were also associated with fossil mammals, have confirmed the utility of the eggshells for biostratigraphic correlations (see Table 2 for references to the geology and palaeontology of the Late Miocene Baynunah Formation). We are therefore reasonably confident that the Marsawdad eggshells will yield a reliable age estimate of the deposits in which they were fossilised. Even though the preservation of the Marsawdad eggshells is not perfect, a few fragments that are lightly eroded are informative enough about surface structure to yield interesting data linking them to the genus Diamantornis rather than to the aepyornithoid type which also occurs in the Late Miocene deposits of the Arabian peninsula (Bibi et al. 2006).

Struthious eggshell fragments from the Rub’ Al-Khali, Oman, were described by Al-Kindi et al. (2021) and Maiorano et al. (2020) and attributed to Struthio camelus and an extinct species (Diamantornis laini or Struthio daberasensis).

Other eggshell fragments were reported, but not described in detail, by Rosén et al. (2021) (see Table 3). The latter authors estimated the ages of several specimens using the ¹⁴C method. Six eggshells less than 1.8 mm thick that were attributed by them to Struthio camelus syriacus yielded Recent ages but two specimens (2.6 and 3.0 mm thick) attributed by them to Struthio kakesiensis? were beyond the range of the ¹⁴C method. They commented «Such thick shelled eggs are not attributable to modern Arabian ostrich (Struthio camelus syriacus), but rather to a larger form that became extinct at least 3 Ma ago, for example, Struthio kakesiensis? ( Bibi et al. 2006 ) known from the Arabian Emirates». Two comments arise from this interpretation:(1) Bibi et al. (2006) did not report the presence of S. kakesiensis in the UAE, the two ootaxa from the Baynunah Formation listed by them being Diamantornis laini and an aepyornithoid, (2) the estimate of 3 Ma is too young, the Baynunah Formation being correlated by Bibi et al. (2006) to the Late Miocene (7.4-6.5 Ma).

The fossil eggshells from locality 25.RAK range in colour from dark chocolate brown (15 fragments) to pale chocolate (2 fragments), like specimens from near Maitan (Al-Kindi et al. 2021) and from several localities in Saudi Arabia (Lowe, 1933b). The darker specimens are all more deeply sculpted than the paler specimens (Fig. 6). Sculpting affects primarily the outer surface of the shell fragments (Fig. 8), this surface being exposed to the air when the specimens erode out of the deposits in which they were originally fossilised. The preferential orientation of the eggshell fragments is due to the curvature of the shells, their most stable position in windy environments being convex upwards.

The sculpting of the surfaces of the eggshells is due to two processes, sand-blasting (minor) and dissolution due to frequent episodes of wetting by dew (repeated many times). By these slow processes, the eggshells can lose much of their thickness and while doing so tend to develop complex systems of smoothly polished ridges and basins which pattern the surface of the shells.

The ten better preserved eggshell fragments from locality 25.RAK range in thickness from 2.4 to 3.0 mm, whereas seven deeply sculpted specimens are only 2.0 to 2.2 mm thick, but they must in any case have originally been appreciably thicker than eggs of Struthio camelus (Figs 6-8). Thickness measurements of the 17 specimens are as follows: 3 specimens: 2.0 mm, 2 specimens: 2.1 mm, 2 specimens: 2.2 mm, 1 specimen: 2.4 mm, 1 specimen: 2.5 mm, 2 specimens: 2.6 mm, 2 specimens: 2.7 mm, 1 specimen: 2.8 mm, 2 specimens: 2.9 mm, 1 specimen: 3.0 mm.

Two of the 17 Marsawdad specimens are relatively unaffected by sand-blasting and dissolution by dew, so they provide reliable information about the surface texture of the shells. In both cases, there appear to be no pores on the outer surface, thereby resembling the large smooth surfaces that occur between the circular pore complexes in eggs of Diamantornis laini (Bibi et al. 2006; Pickford, 1998, 2014; Pickford & Senut, 2000; Pickford et al. 1995; Ségalen et al. 2002; Senut, 2000; Senut & Pickford, 1995; Senut et al. 1994, 1998). The larger of the two well-preserved Marsawdad fragments (Fig. 7) shows the edge of a depression on one side, which could represent the margin of a circular pore complex as in shells of D. laini (Figs 5, 9).

Over the past nine decades, fossil struthious eggshell fragments have been reported from a variety of localities in the Arabian Peninsula (Lowe, 1933a, 1933b; McClure, 1984; Bibi et al. 2006; Hofmann et al. 2018; Al-Kindi et al. 2021; Mairoano et al. 2020; Rosén et al. 2021; Buffetaut, 2022). The most common eggshell fragments reported from the peninsula belong to the extant ostrich, Struthio camelus (Struthio syriacus in Lowe, 1933a, 1933b). Specimens are known from the vicinity of Maitan, Oman (Al-Kindi et al. 2021), from various sites in Saudi Arabia (Lowe, 1933a, 1933b; McClure, 1976, 1978, 1984) (Table 2) as well as generally over much of the peninsula (Boug & Islam, 2018). These eggshells are thin (less than 2 mm) and generally show the pore arrangement typical of the extant ostrich. Near Maitan, such eggshells were exploited by Neolithic societies to fabricate beads (Al-Kindi et al. 2021).

A single fragment of an oospecies with a shell thickness of 2.6 mm (probably of Diamantornis laini but could belong to Struthio daberasensis) was found near Maitan, Oman (Al-Kindi et al. 2021; Maiorano et al. 2020). The eggshells of D. laini from Namibia are considerably thicker than those of S. camelus, ranging in thickness from 2.4 - 3.0 mm. In addition, the pores in eggs of Diamantornis laini are concentrated into circular slightly depressed pore complexes, with large expanses of smooth surfaces devoid of pores between the pore complexes, thereby differing from the eggs of aepyornithoids which have pores liberally scattered over the surface of the eggs (Bibi et al. 2006).

Restudy of the eggshell fragment (NHMUK A 2043) from Shuqqat Al Khalfat, Saudi Arabia (near 21°57’13’’N - 49°45’41’’E) described by Lowe (1933a, 1933b) as suggested by Buffetaut (2022) was undertaken in March, 2023 (Fig. 10). It is 3.1 - 3.2 mm thick and the apparent lack of pores in the fragment suggests appurtenace to D. laini. On this basis the specimen indicates that the deposits from which it was collected are probably of Late Miocene age.

Arambourg (1959) described fossil struthious eggshell fragments from Bou Hanifia, Algeria, reporting that they range in thickness from 2.5 to 3.0 mm. Examination of the fossils housed in the Muséum National d’Histoire Naturelle, Paris (inventory n° 1951-9-295. Fig. 11) reveals that the outer surfaces appear to be devoid of pore structures, and they are thus similar to the surfaces between the circular pore pits of specimens of Diamantornis laini. Remeasurement of the specimens yielded the following data: 1 specimen, 2.0 mm; 1 specimen, 2.1 mm; 3 specimens, 2.2 mm; 1 specimen 2.3 mm; 2 specimens, 2.4 mm; 2 specimens, 2.5 mm; 1 specimen, 2.6 mm; 1 specimen, 2.7 mm.

Recent reassessment of the Bou Hanifia mammals (Pickford & Chaïd-Saoudi, in prep.) suggests that they correlate best with the Late Turolian to Ventian, rather than to the Vallesian, and the fossil eggshells found in the same deposits accord with this reinterptation of the age of the faunas. Thus Bou Hanifia and Marsawdad could be roughly contemporaneous.