It is a known axiom that the current political definitions of regions do not have any meaning in paleontological studies. Thus, the identification of an “African” taxon does not refer to the current geopolitical boarders of the African continent. The geographic extent of biogeographic regions varies if you consider the entire continent vs. a smaller region thereof. The definition of what constitutes an African taxon differs between species and genera, and whether we are studying plants or animals, and if the latter are mammals, birds or amphibians (Holt et al., 2013 and Udvardy, 1975), and the analytical and statistical methods used to produce biogeographic regions. Indeed, a reading of the current literature often obfuscates what precise definition is used by each author.

Since the original publication of the biogeographic realms by Sclater (1858), different scholars have defined biogeographic realms differently, producing different maps (Fig. 1). Furthermore, studies by researchers of fauna and flora often do not use a common terminology (Udvardy, 1975). The different definitions used by various scientists to define the African biogeographic realm are important to the topic of this paper.

The original definition of biogeographic realms was published by Sclater (1858) based on the distribution of birds, followed by Alfred Russel Wallace in 1876 (Wallace, 1876), who expanded on the original study. Wallace identified three main regions within the old world: Ethiopian, Palearctic and Oriental. Subsequent analyses such as those developed by Takhtajan (1969) and Udvardy (1975) suggested other biogeographic regions, each with a different geographic extent and a different name.

Wallace (1876), using the distribution of mammals, identified seven regions: Palearctic, Nearctic, Ethiopian, Oriental, Australian, Neotropical and Antarctic. The Ethiopian region included sub-Saharan Africa, along with the southern part of Arabia (see Fig. 1A). Takhtajan (1969), who based his biogeographic classification on flora, separated the southern African region (the Cape region) from the remaining parts of sub-Saharan Africa, but included the southwestern tip of Arabia along with the Indian subcontinent and Southeast Asia into a single region, which he named the Afrotropical region (see Fig. 1B). Udvardy (1975), grouping both fauna and flora, classified the region, which included sub-Saharan Africa only, without any part of Arabia, as Africotropical (see Fig. 1C). In all these classifications, North Africa, the southern Levant and southern Europe were included in the Palearctic region. Holt et al. (2013) used three orders of vertebrates (Birds, Amphibians and Mammals) to classify biogeographic regions. By incorporating phylogenetic informed statistical methods along with analysis of several vertebrate classes together, they identified five regions, which they named Palaearctic, Saharan–Arabian, Afro-tropical, Oriental, and Sino-Japanese. Most of sub-Saharan Africa, with the exclusion of Madagascar and Arabia, was identified as Afrotropical. They also created a new classification for North Africa and the Near East, which were now grouped into a region called Saharan–Arabian (see Fig. 1D).

Thus, when we study “African” taxa in Eurasia, it is necessary to clarify whether we are referring to the Palearctic region of Wallace (1876), the Ethiopian region of Takhtajan (1969), the Africotropical realm of Udvardy (1975), or the Afrotropical region of Holt et al. (2013) to name a few. To understand “dispersal” events, which involve movement of fauna from one region to the next, it is critical to be consistent in defining geographic borders of various regions. Thus, if we follow Wallace's classification, dispersal from North Africa into the southern Levant would not constitute dispersal from “Africa” (Fig. 1A). Indeed, Tchernov (1988) suggested that the appearance of African taxa in the Levant should not be considered as an African dispersal; a sentiment echoed by Klein (1999). Similarly, if we follow Takhtajan (1969), dispersal of fauna from the Afrotropical region could refer to dispersal from the Indian subcontinent and Southeast Asia as well as dispersal from east Africa into the southern Levant (Fig. 1B). African dispersal from the Afrotropical region as defined by Holt et al. (2013) would include only dispersals from sub-Saharan Africa into both North Africa and southern Europe (Fig. 1D).

It is not the aim of this paper to discuss the merits of each classification method, but to raise the discussion of the inconsistencies that occur and the way in which they may affect our interpretation of the dispersal of early Pleistocene out-of-Africa taxa. Moreover, there have been many more biogeographic divisions suggested in the literature (see Udvardy, 1975).

Authors need to be consistent and clear about which definitions of biogeographic regions they are using. Using the biogeographic division of Wallace (1876) or Takhtajan (1969) results in a broader definition of an African geographic region and hence inclusion of more taxa as African. Conversely, applying the classification of Udvardy (1975) or Holt et al. (2013), which both include a narrower geographic definition of African taxa, resulting in fewer Eurasian taxa identified as having an African origin.

Specimen identification to the taxon level is often in flux. Fossils are not unearthed with tags identifying the species with 100% certainty and taxonomic identifications range between tentative to confident. The taxonomic identification of some fossils is in consensus among scholars, while others tend to be more contentious.

Since the 1960s, there has been a shift in how we approach the taxonomic identification of fossil taxa in zooarcheological assemblages. These changes have affected not only the proportion of Africa taxa identified at several paleontological and paleoanthropological sites, but also the subsequent paleoecological reconstructions derived from them. At its core, taxonomic identification of fossil elements to species (or taxa) was (and still is) based on the morphology of diagnostic elements. In mammals, common diagnostic elements are dentition and epiphyses of long bones (Driver, 1992), while the ability to identify other elements depends on their breakage pattern. Thus, assemblages subjected to a myriad of taphonomic processes (Lyman, 1994) result in a high percentage of fragmented bones, and a low proportion of diagnostic elements (Belmaker, 2005). Paleoecological analyses derived from published taxonomic lists may be affected by taphonomic processes that affected the ability to accurately identify species composition and absolute number, i.e. richness. Thus, an assemblage relatively unaffected by taphonomic processes will have many diagnostic elements, allowing for taxonomic identification to the species level. By contrast, analysis of a taphonomically altered assemblage will have fewer diagnostic elements, fewer identification to the species level, and therefore a different taxonomic list, with perhaps identification only to family or genus rather than species.

Taxonomic identification is aided by the comparison of fossils to comparative specimens in museums and those published in the literature. Identification is bolstered by access to a wide variety of large collections and literature (Lyman, 2010). However, there was a lack of interregional collaboration before the fall of the former USSR and the opening of China to the west (Cox, 2010). During the period of the Cold War, much of the travel to collections and scientific communication was confined to the Western Hemisphere and the European countries in the west, Russian and Slavic speaking countries in the East, and China in the Far East, with little to no international communication among them.

Access to comparative collections was often limited or confined to countries within each socio-political block, and access to publications was limited to those, which shared similar languages and a similar political worldview. The main result for paleoanthropology was that, within each socio-political block, specimens were identified without long-ranging biogeographical comparison to similar specimens across other regions. This resulted in an increase in scientific “endemism” and the tendency to name specimens locally, often with a new name, rather than to assign fossils to a known taxon from other regions. For example, Lister et al. (2005) has noted that Mammuthus taxonomy in China, Russia and Europe may be synonymized into fewer taxa than those originally categorized, suggesting that the original taxonomic list developed in each region might be the result of local scientific practice.

The fall of the Berlin Wall in 1989 has resulted in a gradual increase in international collaboration, which has allowed researchers to study fauna across several regions as well as to personally study and compare specimens previously unavailable to them. This resulted in a more in-depth analysis of each taxonomic group and a wider geographic and temporal comparison for each specimen. This process has progressed and has become more apparent in the 21st century, with the development of the Internet and World Wide Web (WWW) (Lucky, 2000).

This process of globalization and scientific collaboration has resulted in renaming old specimens. For example, early studies named multiple species of Megantereon across the old world during the Plio-Pleistocene: M. cultridens (Cuvier, 1824); M. megantereon (Croizet and Jobert, 1828); M. nihowanensis (Teilhard de Chardin and Pivetau, 1930); M. falconeri (Pomel, 1853); M. sivalensis (Falconer, 1868). However, more recent studies have classified all these into one species, M. cultridens (Turner, 1987).

In addition, recent development of paleontological methodologies such as geometric morphometrics (Adams et al., 2013), molecular techniques and multivariate statistics may be used to revise the taxonomic affinities of a fossil species. For example, geometric morphometric analysis of the shape of upper molars in Javanese cervids has suggested new affiliations for some of the specimens (Gruwier et al., 2015), and the geometric morphometric analysis of the species Pelorovis oldowayensis suggested a taxonomic assignment to the genus Bos (Martínez-Navarro et al., 2007). The advent of molecular paleontology (Wörheide et al., 2016) allows analysis of both phylogeography and taxonomy of extant (Burger et al., 2004) and fossil taxa (Pyron, 2015). Novel statistical methods that have been developed in tandem with increased computing power allow the identification of fossil specimens that previously could not be assigned to a species (Davis and McHorse, 2013).

If we only consider species that are securely assigned to taxa (i.e. with a greater consensus among researcher and across paleontological and/or molecular studies), we obtain a more conservative assessment of the number of African taxa found in Eurasia. The opposite is also true, and the inclusion of fragmentary specimens and/or from specimens with contentious species assignment results in more lenient identification of African taxa.

Several chronological issues affect our interpretation of the number of African taxa in Eurasia: how do we incorporate paleontological data from sites dated back only to the informal term “Plio-Pleistocene”, only using the terminology of before/after the Plio-Pleistocene boundary? How did the revision of the Plio-Pleistocene boundary date from 1.8 Ma to 2.58 Ma affect our ability to interpret generalized dates in the literature? And what time frame do we study in relation to hominin dispersal?

The terms ‘Quaternary’, ‘Pleistocene’ and other terms in the geological timescale have been used for over 150 years (Gibbard et al., 2010), and until the expansion of radio-chronological and paleomagnetic methods, dating of the hominin specimens in Eurasia or archeological sites remained elusive, and only relative biochronological (e.g., Villafranchian) or archeological (e.g., lower Paleolithic) terminology was used. For example, the site of ‘Ubeidiya was dated in the 1960s to the Villafranchian (Stekelis et al., 1960) or the Pleistocene (Tchernov, 1968), and was associated with a glacial period between the Cromerian and the Holstein (Tchernov, 1973). In addition, the subdivisions of the Pleistocene and the Pliocene–Pleistocene border differ in the countries of the ex-Soviet Union compared to those in the West (Ljubin and Bosinski, 1995). Thus, when mining the literature, especially the older references, it may be difficult to ascertain contemporaneous sites.

Of course, there have been ongoing discussion and debates regarding the absolute date and time intervals that the terms represent (Aguirre and Rosas, 1985, Haq et al., 1977 and Van Couvering, 2004). In 2010, the date of 2.58 Ma was ratified as the base of the Pleistocene from the previous 1.8 Ma (Gibbard et al., 2010). This again critically affected how we review older literature, as it is often difficult to assign a chronometric date on the basis of the more fluid terminologies that were then common. Specifically, this affected the discussion of the co-dispersal of humans with other mammals from Africa and requires us to be cognizant of the use of relative chronology terms.

The dispersal of fauna from Africa to Eurasia did not begin in the early Pleistocene. Indeed, during the Miocene (23.5–5.0 Ma), there was a land bridge connecting the African continent with Eurasia primarily though the Syrian African Rift Valley, which was considered a main crossroad between Africa and Eurasia: a gateway for northward and southward dispersal (Thomas, 1985). A period of long-distance reciprocal biotic exchange due to the dramatic lowering of the Mediterranean Sea Level occurred during the Messinian event (5.96 to 5.33 Ma) (Tchernov and Belmaker, 2004 and van der Made et al., 2006). During the terminal Miocene (ca. 5 Ma), the developing Red Sea and flooding of the Mediterranean put an end to the geographic corridor through the Syrian African Rift Valley. The onset of the Pliocene (5.333–2.58 Ma) was marked by an abrupt transgression of the Mediterranean and reestablishment of the barrier between Africa and Eurasia. During this period, Sub-Saharan Africa became isolated from the rest of the world by the Saharan–Arabian arid belt (Thomas, 1985), and the southern Levant was increasingly isolated biogeographically. Towards the early Pleistocene (2–1 Ma), another regression of the Mediterranean terminated the quasi-isolation of the region (Tchernov and Belmaker, 2004). These early biotic exchanges did not include the genus Homo. The Miocene dispersal included primates, but that discussion is beyond the scope of this study (Tchernov et al., 1987).

Given the long history of dispersal events between Africa and Eurasia that took place during the Miocene and the Pliocene, it is crucial to focus on those dispersals that occurred in conjunction with early hominin dispersal and not earlier for the question of the ‘Migratory Wave Hypothesis’. Determining when Afro-Eurasian biotic changes become relevant to the question of human migrations is dependent on the absolute date of earliest evidence for hominin remains in Eurasia (which may include fossils, stone tools and bone surface modification).

However, the dating of the earliest hominin dispersal from Africa has changed over time due to new finds in Eurasia as well as re-dating of old sites using radiometric dating. Going back to the site of ‘Ubeidiya that was previously mentioned, found in 1959. Originally, it was dated back to the Villafranchian biochronological stage within the Lower Pleistocene with an estimated date of approximately 500,000 years ago (Haas, 1966, Haas, 1968, Stekelis et al., 1960 and Stekelis et al., 1969). By the mid-1980s, additional excavations at the site revealed fauna of African origin, leading to the dating of the site to 1.4 Ma, based on biochronology (Tchernov, 1987). Following new biochronological dating and new analysis at the site, it is currently dated back to 1.6–1.2 Ma (Belmaker, 2017). Similarly, paleontological and archeological finds in the Orce Basin in the Iberian Peninsula (Venta Micena, Baranco Leon and Fuenta Neuva 3) in the 1970s were originally dated using paleomagnetism, to the Matuyama period, between the normal chrons of Jaramillo and Olduvai (Martínez-Navarro and Toro, 2003).

The discovery in 1991 (Dzaparidze et al., 1992) of hominin remains associated with lithic artefacts in lower Pleistocene deposits from Dmanisi (East Georgia, Caucasus) was originally reported to be as old as 1·6–1·8 Ma (Dzaparidze et al., 1992 and Gabunia and Vekua, 1995). Subsequent dating of the site to 1.85–1.78 Ma (Ferring et al., 2011) marked this site as the oldest confirmed evidence for hominins out of Africa with extensive hominin remains, lithic and faunal material with evidence of human butchery (Lordkipanidze et al., 2007). Thus, up to 2010, many of the discussions of the co-occurrence of early Homo and African taxa focused on the dispersal of hominin during the Olduvai–Matuyama reversal, and the onset of early Pleistocene grosso modo. Taxa that appear in Eurasia after that date were considered as “African taxa” with a putative dispersal together with early Homo.

However, there have been continuously publications of new discoveries suggesting that early Homo may have dispersed earlier than that into Eurasia. In Yiron, Israel, cores and flakes have been retrieved from the gravels in a major crevice apparently below a lava flow of 2.4 Ma (Ronen, 1991 and Ronen et al., 1980) and the ‘Erq el Ahmar’ Formation (Horowitz, 1979) located about 14 km south of the Sea of Galilee, in the Jordan Valley, and accumulated prior to the ‘Ubeidiya Formation. Artifacts found in the deposits (Tchernov, 1995) were suggested to predate the early Pleistocene. Similarly, some artifacts found in Pabbi Hills, Pakistan, were suggested to be dated back to 2.0 Ma (Dennell et al., 1988). These early pre-2.0 Ma were contested on stratigraphic, archeological and chronological grounds (Bar-Yosef and Belmaker, 2011).

Recently, Olduvai Chron paleoanthropological sites such as Longgupo (China), ca. 2.2 Ma (Han et al., 2017), and Masol (India), ca. 2.5 Ma (Dambricourt Malassé et al., 2016, Moigne et al., 2016 and Tudryn et al., 2016), have been found. If these dates are indeed confirmed, the anthropogenic origin of sites such as Yiron, ‘Erq el Ahmar and Pabbi Hills should be reconsidered. Most importantly, if the anthropogenic origin of these sites were confirmed, the date for the earliest presence of hominins out of Africa would be pushed back to the earliest Pleistocene (2.58–1.8 Ma) or even the late Pliocene (> 2.58 Ma). This will expand the discussion of the ‘Migratory Wave Hypothesis’ to include dispersal events dated to 2.1 Ma (for example the wolf–Equus event (Azzaroli, 1983 and Azzaroli et al., 1988) dated to the Middle–Late Villafranchian transition). However, at the current state of research today, studies that focus on Afro-Eurasian biotic exchanges in the middle Pliocene (such as the site of Bethlehem, Palestine (Rabinovich and Lister, 2016) or Kvabebi, Georgia (Agustí et al., 2009 and Martínez-Navarro, 2004) may not be pertinent to the discussion at hand, and while very informative on biogeographic routes and ecological process, they may not be as critical for the evaluation of the co-dispersal of large mammals with humans.

The determination of the cut-off date is somewhat of a moving target; it is important to clearly delineate which dispersals are being discussed within the context of faunal biogeographical processes (with their own merit), and which dispersals are analyzed in conjunction with early Homo dispersals.

In his seminal work on the fauna of ‘Ubeidiya, Tchernov (1986) assigned several taxa to a paleotropical biogeographic kingdom. It is unclear which specific author he followed when assigning his divisions. The paleotropical taxa were assigned either to the Ethiopian region (a sub division of the Paleotropics): Crocuta crocuta, Herpestes sp. Kolpochoerus olduvaiensis, Giraffa gen. indet., Hippopotamus gorgops, and Pelorovis oldowayensis, while Felis cf. chaus, Bos sp. and Hystrix indica were assigned to the Oriental region (another sub division of the Paleotropics). In addition, several species (Gazella sp., Oryx sp., Equus tabeti) were identified as coming from the North African or Saharan–Arabian region. However, this also exemplifies how the definition of biogeographic realms can have a direct impact on how we quantify African species. If we accept a maximizing definition of Paleotropics as reflective of “African” taxa, then the number of African species in ‘Ubeidiya is 12. On the other hand, if we use another definition, such as that by Holt et al. (2013), who applied a phylogenetic framework to traditional biogeographic divisions, Afrotropical taxa are sub-Saharan taxa and do not include species from North Africa or Arabia.

If we apply a more stringent definition of biogeographic realms following Holt et al. (2013), Equus tabeti, Gazella sp., and Oryx sp. assigned as African by Tchernov (1986), are assigned to the Saharan–Arabian region. Similarly, taxa from South and Southeast Asia (Felis cf. chaus, Bos sp. and Hystrix indica) by Tchernov (1986), maybe assigned to the oriental region by Holt et al. (2013), are excluded from the list of African taxa.

Of course, there is no absolute method to guarantee that taxonomic identifications are indeed correct and different views and opinions are prevalent in the taxonomic literature. We can therefore envision a continuum between taxonomic identifications in consensus across researchers and which are based on methods with a high probability of correct identification and those identifications, which are highly contested. Two issues need to be addressed: •

the taxonomic identification of a fossil specimen;

There are fossil specimens in Eurasia that are securely assigned to species and have a clear evolutionary link to African ancestors. These include Theropithecus oswaldi, Paleorovis oldowayensis, Kolpochoerus olduvaiensis, cf. Alcelaphus, Giraffa camelopardalis, and Herpestes sp. representing only six taxa. The identification of other fossil specimens as African species is contested by some. I have detailed below the main arguments for and against the identification of specific fossils as those of species of either African origin or non-African origin. The aim of this section is not to favor one opinion over the other, but to exemplify how taxonomic issues may plague our ability to quantify the number of taxa of African origin in Eurasia.

Identification of intercontinental dispersal events has its merit when discussing the co-dispersal of hominins and African taxa. As presented in the “Migratory Wave Hypothesis”, it is crucial to confirm that we are comparing the first appearances of African taxa in Eurasia, which appear in Eurasia together or after the dispersal of hominins, but not earlier. Today, the accepted and confirmed event for the first hominin dispersal from Africa is that at Dmanisi, Georgia, dated back to 1.78–1.82 Ma (Ferring et al., 2011). While some suggestions for early dispersals have been put forth (Han et al., 2017), they are still unverified. To evaluate the “Migratory Wave Hypothesis”, some may consider only dispersals from 1.8 Ma and younger, while others may support the analysis of sites as old as 2.58 Ma.

Several species that had been identified as part of an early Pleistocene dispersal event have been revised based on chronological considerations, to predate the early Pleistocene. The five bovids (Sivatragus bohlini, Oryx sivalensis, Vishnocobus patulicornis, Sivacobus palaeindicus, and Damalops palaeindicus) found in South and Central Asia were originally assigned an early Pleistocene age. A reanalysis of the taxonomic identification of the species led O’Regan et al. (2011) to suggest that these bovids either appeared earlier than the late Pliocene/early Pleistocene in Asia or represent locally evolved species from previous dispersals. Indeed, all these species are endemic to South Asia, albeit assigned to a genus or family of African (sensu latu) origin. This supports Vrba's suggestion that the Hippotragini dispersal occurred earlier than 2.7 Ma when sea level was lower, creating a land bridge between Africa and Asia across Arabia (Vrba, 1995).

The Giraffa sp. identified at Bethlehem, originally dated back to 2.5 Ma, is now viewed as a mid-Pliocene dispersal, as the site has been recently dated back to 3.5–3.0 Ma (Rabinovich and Lister, 2016). This suggests that if this were indeed an out-of-Africa event, it predates the time frame of interest. Moreover, there are taxonomic reasons to doubt if Giraffa sp. is indeed an African taxon. There are several Pliocene Giraffa ssp. dated back to 4.0 Ma (Sen et al., 1998). Therefore, the find in Bethlehem may represent an expansion of these Asian populations. Similarly, the find of Giraffa sp. in ‘Ubeidiya may also be associated with this earlier dispersal and needs not reflect a later Pleistocene dispersal. Indeed, tooth measurements have suggested that the species in ‘Ubeidiya may be assigned to G. jumae (Robinson and Belmaker, 2010). In contrast, the Giraffa species in Latamne, Syria, is indeed the extant species G. camelopardalis present in Africa from 1.2 Ma (Guérin et al., 1993). This would suggest an early Pleistocene dispersal of this species.