First substantial evidence for Old World vultures (Aegypiinae, Accipitridae) from the early Palaeolithic and Iberomaurusian of Morocco

Remains of at least three species of large aegypiine vultures from early Palaeolithic and Iberomaurusian of Ifri n’Ammar, Morocco are the first substantial fossil record of these taxa in the Maghreb. They can be tentatively referred to the two extant species Aegypius monachus (Cinereous Vulture) and Gyps fulvus (Griffon Vulture), and to the extinct Gyps melitensis. Few fragments of remarkably large bones may belong to especially large specimens of A. monachus, but it cannot be ruled out that they are indeed remains of the so far only insufficiently known palaeospecies A. prepyrenaicus Hérnandez, 2001, originally described from the Upper Pleistocene of Spain. Two vulture species definitely occurred contemporarily; Gyps fulvus was found in the entire sequence, while A. monachus occurred only between 13,800 and 17,000 calBP. G. melitensis was found only once in an unknown stratigraphic context. All remains were found along with human artefacts in a cave deposit. Few bones show longitudinal scratches, which probably are cut marks, indicating that humans made use of the flesh, feathers and/or bones of these vultures. None of these species were previously recorded for the Upper Pleistocene of the Maghreb, and the fossil specimens provide important evidence for the former distribution of Old-World vultures in this area, which is insufficiently known and which changed dramatically during the twentieth century.


Introduction
The Pleistocene record of aegypiine vultures in the Mediterranean region is quite extensive, but virtually restricted to its northwest and northeast parts as well as to several Mediterranean islands (Sánchez Marco 2007). Remains of extant Cinereous (Aegypius monachus) and Griffon Vultures (Gyps fulvus) are frequently recorded (Jánossy 1989;Mlíkovský 1998;Sánchez Marco 2004, 2007. The extinct Gyps melitensis is distinguished from G. fulvus by its larger size and was originally described from cave deposits of Malta (Lydekker 1890). It was also reported from several other Mediterranean islands as well as mainland Spain, Monaco, France, and even Austria and Germany (Mourer-Chauviré 1975;Jánossy 1989;Louchart 2002;Sánchez Marco 2004, 2007, and references therein), though some of these records might have been based on misidentified bones of either A. monachus or G. fulvus (Mlíkovský 1998(Mlíkovský , 2002. Another palaeospecies, Aegypius prepyrenaicus Hernández, 2001 was described on the basis of a large proximal ulna (approx. 8% larger than the corresponding bone of A. monachus) from the Upper Pleistocene of Spain, but Sánchez Marco (2007) regarded it a nomen dubium.
In the Eastern Mediterranean, fossil evidence for G. fulvus and A. monachus is only reported for cave deposits of Israel (Tchernov 1962;Sánchez Marco 2004) and the Lebanon (Hooijer 1961;Kersten 1991), where both species are often found in the same horizons (Kersten 1991). Unfortunately, none of these specimens were described or illustrated, which makes more detailed comparisons with other vulture remains impossible.
In northern Africa, a single, fragmentary bone from the Pleistocene of Egypt was referred to G. africanus (Bocheński 1991). The avifauna of the cave site Haua Fteah, Libya, comprises more the 60 species of birds, but surprisingly no vulture remains are reported so far (MacDonald 1997). Bird remains are also well represented in the Pleistocene of the Maghreb, though most of them remain to be identified (Steele 2012). Remains of a large raptor, probably a vulture, are mentioned for the Middle Pleistocene hominid site of Jebel Irhoud, Morocco (Thomas 1981;Steele 2012: 119), but no further information is available.
Here we describe remains of aegypiine vultures from Ifri n'Ammar, a cave with rich Iberomaurusian and Middle Palaeolithic deposits in the Eastern Rif Mountains of Northern Morocco, which is already well known for a diverse Middle Palaeolithic mammal fauna (Mikdad et al. 2000;Hutterer 2010;Richter et al. 2010). The vulture material from the Middle Palaeolithic sections was preliminarily referred to A. monachus (Hutterer 2010), but actually represents both A. monachus and Gyps fulvus. Fossil evidence for A. monachus and G. fulvus is also relevant for reconstructing former areas of distribution for both species, which changed dramatically in the first half of the twentieth century.

Material and methods
The fossil specimens described herein were excavated during archaeological excavations of the rock shelter Ifri n'Ammar in N Morocco. Excavations were conducted between 1997 and 2015 by members of the Institut National des Sciences de l'Archéologie et du Patrimoine, Rabat (INSAP) and the Commission for the Archaeology of non-European Cultures (KAAK, Bonn) of the German Archaeological Institute DAI. The rock shelter is situated at 150 m in the eastern hills of the Plaine du Guerrouaou (34°47′03.68″N, 03°05′32.42″E), a large depression at 375 m surrounded by hills up to 1022 m elevation. The 6.3 m deep sediment filling covers a time range of 11,000-180,000 calBP. Principal publications on the archaeology of the site are by Mikdad et al. (2000), Moser (2003), and Nami and Moser (2010). All bones form part of the INSAP collections.
We follow Amadon (1977), who suggested to merge the monotypic taxa Agypius monachus, Sarcogyps calvus, Torgos tracheliotos, and Trigonoceps occipitalis into the genus Aegypius, and we follow Seibold and Helbig (1995) who applied the name Aegypiinae for a clade comprising the genera Aegypius, Gyps, and Necrosyrtes. For a more detailed discussion on the nomenclature of aegypiine vultures, see .
Measurements were taken in accordance to the guidelines of von den Driesch (1976) and Louchart (2002). The height of the os quadratum was measured from the dorsal margin of the capitulum squamosum to the ventral margin of the condylus lateralis. Measurements of bones of the fossil A. prepyrenaicus and G. melitensis as well as additional measurements of extant aegypiine vultures were taken from the literature (Louchart 2002;Sánchez Marco 2007 Description and comparisons. Quadratum-With a total height of 31.0 mm, specimen INSAP/IA01/I14/19 is comparable in size to the quadratum of A. monachus (x̅ = 30.4 mm (30.1-30.7 mm); n = 2). The dorsal margin of its cotyla quadratojugalis is strongly pronounced and laterally protruding, and the condylus medialis protrudes far ventrally (Fig. 1a). These characters are shared with Aegypius, but not with Gyps, in which the dorsal margin of the cotyla quadratojugalis is indistinct and which condylus medialis is not as elongated ( Fig. 1a-d).
Humerus-The fragmentary distal end of a huge left humerus (INSAP/IA/K15/24B; Fig. 2c-d) strongly resembles that of Aegypius, i.e., both epicondylus dorsalis and epicondylus ventralis are very robust, and sulcus humerotricipitalis as well as sulcus scapulotricipitalis are deep and marked (Fig. 2c-f;Sánchez Marco 2007;. In Gyps, the epicondyli are less strongly developed and the sulcus humerotricipitalis is much shallower (Fig. 2k). The distal width of the specimen is 50.7 mm and thus larger than corresponding measurements of extant A. monachus and of the extinct G. melitensis (Table 1).
Carpometacarpus-The highly fragmentary proximal right carpometacarpus (INSAP/IA/H13/21; Fig. 2s) bears a large pneumatic foramen in its fovea carpalis cranialis, which is characteristic for Aegypius, although quite variable in respect of size and shape (Sánchez Marco 2007;. In Gyps, the fovea carpalis cranialis is generally not pneumatised, but few specimens show a longitudinal furrow pierced by minute foramina. Due to its fragmentary preservation, no meaningful measurements can be taken from the specimen, but it is comparable in size to A. monachus ( Fig. 2s-t).
Tarsometatarsus-The fragment of a distal left tarsometatarsus (INSAP/IA02/no number) lacks trochlea metatarsi II, but its huge size and the short wing-like process of the trochlea metatarsi IV clearly indicate that it belongs to an aegypiine vulture (Fig. 3m). The width of its trochlea metatarsi III (10.7 mm) is larger than the corresponding measurements for A. monachus (x̅ = 9.7 mm (8.9-10.3 mm); n = 9) and G. fulvus (x̅ = 9.9 mm (9.3-10.5 mm); n = 10) (measurements taken from Sánchez Marco 2007). As in Aegypius, the trochlea metatarsi II extends further distal than trochlea metatarsi III (Fig. 3m), which is also the case in several species of Gyps, such as G. africanus, G. bengalensis, G. tenuirostris and Gyps rueppelli. In others such as G. coprotheres, G. fulvus, and G. himalayensis as well as G. melitensis, Terminal phalanges-Three right terminal pedal phalanges ( Fig. 3t, v-x) can be referred to Aegypiinae, because the proc. flexorius of all terminal phalanges does not project far distally as in other Accipitridae (Ballmann 1973). One of them ( Fig. 3t) probably is pedal phalanx I/2 as it bears a shallow but distinct proc. extensorius. The remaining terminal phalanges exhibit a dorsoplantarly deep proc. extensorius, a prominent longitudinal crest and broad articular facets, which is also characteristic for Aegypius. The terminal phalanges of Gyps are much weaker with proc. flexorius and proc. extensorius being less developed (Fig. 3s). Remarkably, all ungual phalanges are larger than those measured for A. monachus and G. fulvus (Table 2).
Remarks. In respect of the postcranial skeleton, A. monachus is very similar to the slightly smaller and less robust A. tracheliotos. The latter is clearly distinguished from the former by its proportionally more elongated tarsometatarsi Manegold and Zelenkov 2014), but this character is not discernible in the present material. It seems nevertheless reasonable to refer the above listed specimens from Ifri n' Ammar to A. monachus, because of their large size and robustness.
Humerus-Although only fragmentarily preserved, the shaft of a left humerus (INSAP/IA02/H15/23B; Fig. 2g-h) can be referred to G. fulvus, because both sulcus humerotricipitalis and sulcus scapulotricipitalis are shallow and indistinct. In contrast to Aegypius, both sulci do not reach towards the level of the proximal margin of the fossa m. brachialis (Fig. 2h, l). The width of the fragmentary shaft just proximal of fossa m. brachialis is 23.5 mm, which is comparable to the size of G. fulvus. The epicondylus ventralis and the attachment site for the proximal articular ligament of the distal right humerus (INSAP/IA10/N14/25A) (Fig. 2i) and the epicondylus dorsalis of another specimen (INSAP/IA97/ M14/18) (Fig. 2j) are as indistinct as in Gyps (Fig. 2k), and the specimen is thus clearly distinguishable from the corresponding bone of Aegypius (Fig. 2e-f; . Radius-As in all Aegypiinae, the facies articularis ulnae of the fragmentary distal right radius (INSAP/IA97/ M13/22; Fig. 2o) shows a pair of deep depressions with the cranial one being highly pneumatic. It shows several Measurements are taken in accordance to Louchart (2002;Fig. 9) with (1) being the dorsoventral height of articular facet, and (2) being the proximal height from distal tip of proc. flexorius to dorsal margin of articular facet (all measurements in mm) features characteristic for Gyps, such as a more distinct caudal depression (but see Sánchez Marco 2007), a distinct furrow cranial of the facies articularis radiocarpalis, and a more distal ligamental attachment site on its caudal surface ( Fig. 2o-p). With a distal width of 19.7 mm, specimen INSAP/IA97/M13/22 falls within the size range of G. fulvus (x̅ = 19.2 mm (18.5-19.9 mm); n = 5).
Os carpi ulnare-All three ossa carporum ulnaria from Ifri n'Ammar show characteristic features of Aegypiinae, such as a pneumatised facies articularis ulnocarpalis (Fig. 2q). Compared to A. monachus, the incisura metacarpalis is rostrocaudally broader as it is the case with G. fulvus (Sánchez Marco 2007). All specimens fall well within the size range of G. fulvus and are smaller than the average os carpi ulnare of A. monachus (Fig. 2q-r; Table 3).
Femur-Two small fragments of a proximal right femur probably belong to the same individual (INSAP/IA/ K15/24B). One of them (Fig. 3a) shows several pneumatic foramina craniomedially of the crista iliotrochantericus, which is characteristic for Accipitridae (Holdaway 1994), as well as a crescent-shaped impressio iliotrochantericus, which is found in Aegypius and Gyps (Fig. 3b). Based on its proximal width, this femur appears to be slightly smaller than that of G. fulvus, but distinctly smaller than A. monachus (Table 3).
Tibiotarsus-In contrast to Aegypius and virtual all the remaining Accipitridae (Ballmann 1973), the distal end of tibiotarsus of Gyps is craniocaudally deep and in distal view it appears to be square rather than oblong ( Fig. 3g; Lydekker  1890). This feature is discernible in all three distal tibiotarsi from Ifri n'Ammar ( Fig. 3e-j). The margins of the condyli are more or less abraded in all specimens, so that meaningful measurements are hardly possible, but they all seem to fall in the size range of G. fulvus (Table 3).
Pedal phalanges-Pedal phalanx II/1 (INSAP/IA97/ M14/23; Fig. 3p) is not abbreviated as in most Accipitridae  fulvus (a, b, d, e, g, h, i, j, l, m, o,  Remarks. All species of Gyps are very similar osteologically, but G. fulvus is larger than white-backed (G. africanus) and slightly larger than Rüppell's Vulture (G. rueppelli). The last mentioned species are today restricted to Africa south of the Sahara, though G. rueppelli has been regularly reported in small numbers in North Africa and on the Iberian Peninsula in recent years (Ramírez et al. 2011). Thus, it seems most likely that the fossil specimens described above are remains of G. fulvus, but because of the fragmentary nature of the material this identification remains tentative. Three specimens (  Description and comparisons. Femur-The distal femur is only imperfectly preserved, lacking most of its condylus lateralis and sulcus intercondylaris. In contrast to Aegypius but as in Gyps, the femur lacks a distinct tubercle proximal to fossa poplitea , only a narrow and shallow, more distally situated crest is discernible (Fig. 3c). The distal width of the femur measures 38.0 mm, which is wider than any distal femur measured for G. fulvus (Table 3). According to Lydekker (1890) the width of one distal femur of G. melitensis from Malta is 44.0 mm, but Sánchez Marco (2007) gives only 35.7 mm for the same specimen (Table 3).  Remarks. Because it lacks a distinct tubercle proximal to the fossa poplitea and because of its large size, this specimen is tentatively referred to G. melitensis, which was not reported previously for the Maghreb. This femur shows several cut marks at its caudomedial margin. Description and comparisons. Costa vertebralis-The dorsal fragment of a sternal rib is comparable in size with that of A. monachus and G. fulvus, and shows a similar pattern of pneumatic foramina on its cranial and caudal surface. The fragmentary nature of the specimen does not allow identification beyond the subfamily level, and because the sternal ribs of the Bearded Vulture (Gypaetus barbatus) are also pneumatic, referral this fragment to Aegypiinae is only tentative.
Humerus-Fragments of a huge proximal left humerus (INSAP/IA97/L16/15; Fig. 2a) can be referred to Aegypiinae, because it shows numerous minute foramina along the caudodistal margin of caput humeri, which is characteristic for this taxon (Fig. 2b). However, Aegypius and Gyps apparently are indistinguishable on the basis of the proximal end of their humeri by means of discrete characters. The fossil specimen is clearly larger than G. fulvus and its proximal width even exceeds that measured for the largest specimen of our restricted sample of comparative skeletons of A. monachus (Table 1), though it might belong to G. melitensis or a large fossil species of Aegypius (see below). Fragments of two further humeri (INSAP/IA01/I13/6 and IA01/K13/13) are smaller (Table 1) (Table 3). One fossil specimen (INSAP/no number) is distinctly wider than the caput femoris of the latter species (Table 3). All these specimens are much too fragmentary for any identification beyond the subfamily level.
Tibiotarsus-The huge size and overall similarity allow the assignment of a distal right tibiotarsus (INSAP/IA/ K16/24) to Aegypiinae. Only its cranial surface is preserved and its condyli are heavily abraded, so that no meaningful measurements are possible. The intercondylar tubercle on the pons supratendinosus is quite indistinct, and the position of the tuberositas retinaculi medialis seems to be more similar to the situation in G. fulvus, but the lack of more diagnostic characters makes a more detailed identification impossible. The same applies for a distal left tibiotarsus (INSAP/ IA10/N13/25B; Fig. 3k), which lacks both condyli and the area of the incisura intercondylaris, so that differentiating characters of Aegypius and Gyps are no longer discernible.
Fibula-The proximal fragment of a right fibula (INSAP/ no number) is comparable in size and morphology to the corresponding bone of the larger aegypiine vultures, but its poor preservation does not allow more meaningful measurements and comparisons.
Tarsometatarsus-The lateral fragment of a proximal tarsometatarsus (INSAP/IA/K16/24) can be referred to Aegypiinae, because of its large size and the lack of a tuberculum m. fibularis brevis. A large, but heavily abraded distal left tarsometatarsus (INSAP/IA02/K13/22; Fig. 3n) probably is also referable to an aegypiine vulture, but the poor preservation prevents any further identification.
Remarks. Most of the specimens listed above are fragments of rather indistinctive bones. Others are insufficiently preserved so that the identification beyond the subfamily level is not possible.

Discussion
It is evident that at least three different vulture species are represented in the bird material collected at Ifri n'Ammar. Seven remains are referable to A. monachus, 30 specimens can be tentatively referred to G. fulvus. A distal femur with features characteristic for Gyps clearly exceeds the corresponding bone of G. fulvus in size, and is most likely referable to G. melitensis.
Four bones, a distal humerus (INSAP/IA/K15/24B) and three pedal phalanges (INSAP/no number; INSAP/IA/ K12/16; INSAP/IA02/I16/21) show diagnostic characters of Aegypius, but are larger than the corresponding bones of A. monachus. However, our comparisons are based on a small sample of comparative skeletons of A. monachus that might not reflect the actual size range within this species. All fossil specimens mentioned above might belong to large individuals of A. monachus. Alternatively, they might indicate the presence of a separate, now extinct Aegypius species, such as A. prepyrenaicus. This palaeospecies was described from the Upper Pleistocene of Spain on the basis of a particular large proximal ulna (Hérnandez 2001), but Sánchez Marco (2007) regarded it as nomen dubium. A fragmentary proximal humerus (INSAP/IA/L16/15) and a left caput femoris (INSAP/no number) fall within the same size category as the aforementioned bones, but cannot be positively referred to either Aegypius or Gyps due to the lack of diagnostic characters. They might either belong to large specimens of A. monachus, or to G. melitensis or even to the enigmatic A. prepyrenaicus. Given the fragmentary nature of the material and the insufficient data on the size range of A. monachus, however, we abstain from further speculations on the presence of a fourth species of vulture at Ifri n'Ammar.

Palaeoecological implications
Bones of Aegypius monachus and Gyps fulvus co-occurred in the Ifri n'Ammar for more than 2.000 years, from level 16 to 24. This is equivalent to a calibrated age of about 14,735-14,378 calBP (level 17) to 16,411-16,159 calBP (level 24). Bones of G. fulvus were found in almost all levels, from the top (level 5) to level 27, the beginning of the Palaeolithic (Hutterer 2010). In a diagram in Mikdad et al. (2000: fig. 38), records of Gyps in the upper levels 12-15 were misidentified as Aegypius. The period of overlap of Aegypius monachus and Gyps fulvus found here coincides with a high concentration of eggshells of Struthio camelus in the cave (Hutterer 2010), and therefore probably with a sufficient supply of food for two species of vultures.

Evidence for the manipulation of bird bones
Extremitas omalis of left coracoid (INSAP/IA10/P14/27A), distal humerus INSAP/IA10/N14/25A and distal tibiotarsi INSAP/IA97/M14/23 und INSAP/IA10/N13/25B show cut marks. This is not surprising as all the larger bones found in the Ifri n'Ammar were left by humans and had been manipulated while preparing animals as food, or in the course of other unknown practices.

Implications on distributional changes
The modern distribution of aegypiine vultures in the Maghreb changed dramatically during the first half of the twentieth century. G. fulvus was a common breeding bird in Morocco, Algeria and Tunisia, but witnessed a dramatic decline and local extinction since 1900 (Thomsen and Jacobsen 1979;Cramp and Simmons 1980;Isenmann and Moali 2000;Isenmann 2005). Evidence of A. monachus for the Upper Pleistocene is especially noteworthy, because this species does no longer breed in the Maghreb region (Thiollay 1994). Its former status in North Africa is uncertain, because it was frequently confused with the Lappet-faced Vulture (A. tracheliotos) (Cramp and Simmons 1980). It formerly bred in Morocco, but apparently was restricted to few colonies in the Tangier Peninsula (Heim de Balsac and Mayaud 1962;Thévenot et al. 2003). Reports on vulture colonies in the north of Algeria from the late 19th and early twentieth century might pertain to this species (Sharpe 1878), although they are generally regarded as former breeding evidence for A. tracheliotos (Heim de Balsac and Mayaud 1962;Isenmann and Moali 2000).