Species are important as the primary units of organisation in classifying taxa (Groves, 2004). Linnaean taxonomy provides a universal hierarchy in which every organism can be placed. Hence, taxonomy should be uniform in its applicability to taxa. Additionally, the process of identifying species should be objective. While the merits and requirements of classification are well recognised, actual species concepts that guide the identification of species are still debated.

The most widely used definition is the Biological Species Concept (BSC) (Table 1). Despite its pervasiveness in biological literature, it is fundamentally flawed. Firstly, it cannot be objectively applied to allopatric species (Groves, 2004). In sympatric species two populations that do not interbreed may logically be assumed to belong to different species. However, allopatric species are already isolated through geography and so it becomes necessary to make assumptions about whether they would interbreed (Groves, 2004). Additionally, the BSC does not indicate the extent of difference necessary to demarcate two species. Consequently, species could be identical except that they do not interbreed (Groves, 2004). Finally, and fundamentally, the BSC is entirely inapplicable to extinct species (Strait, 2013).

Most other species concepts are also problematic. The Recognition Species Concept (Table 1), for example, improves on the BSC in that it refers to what can be observed in the interaction between species, but cannot be applied to non-sexually reproducing taxa, to taxa primarily observed in captivity, or to extinct taxa (Groves, 2004).

The Phylogenetic Species Concept (PSC) (Table 1), however, is promising (Groves, 2004). This concept is predicated on the ability to distinguish species based upon their unique collection of heritable traits. Importantly, the PSC does not stipulate what form the differences must take (e.g. genetic distance or reproductive strategy) and so it is applicable across taxa. Additionally, it is falsifiable and observational, unlike many other concepts that require a significant amount of speculation about the evolutionary mechanisms underlying an observation (Groves, 2004).

According to Strait (2013), the PSC may be the preferred species concept employed by palaeoanthropologists, though the Evolutionary Species Concept (ESC) (Table 1) is also employed. While superficially similar, the ESC and the PSC are separated by an important practical difference: the ESC describes the mechanism underlying a pattern of evolutionary relationships, while the PSC refers only to the pattern itself (Groves, 2004). The definition of a species under the ESC is very clear but offers no practical application, which has led to the introduction of other related species concepts (3–5 in Table 1), but the applicability of these related concepts to fossil taxa is equally problematic (Groves, 2004).

While clear species concepts help define and identify natural species, species as defined in palaeoanthropology act more as communicative tools, and are subject to personal preferences (Schrenk, 2013). Hence, a description of two chronospecies (e.g. Australopithecus anamensis and Australopithecus afarensis) may be theoretically and practically useful, distinguishing two similar, phylogenetically related, morphs that followed one another chronologically, with the latter possibly evolving via anagensis (Kimbel et al. 2006). Similarly, the typological distinctions of morphospecies (e.g., H. ergaster and Homo georgicus) may be pragmatically employed for distinctions between biogeographic morphs, though despite the morphological overlap represented by these taxa possibly warranting sub-species designation biogeographic designation (“subspecies are populations, geographic segments of a species, not morphs co-occurring with other variants, and […] they differ from each other on average, not absolutely” [Groves, 2004: 1112]), sub-species are comparatively uncommon among hominins. A degree of flexibility in defining species renders palaeoanthropology (and broader palaeontology) vulnerable to political and/or unscientific allocations of species status to undeserving fossils (Schrenk, 2013; e.g. the currently sparse fossil evidence for Kenyanthropus platyops). Furthermore, the generally fragmentary/distorted nature of many fossils (e.g. OH 7 and KNM-WT 40000) may obscure differences and/or similarities between fossils that can be more easily identified in extant specimens, resulting in different taxonomic biases.

Supraspecific taxonomy classifies species into increasingly nested hierarchical groups (Strait, 2013). While most higher taxa are defined by clades (Groves, 2004), this is not always true for genera. A clade is a monophyletic group, therefore representing species that mutually descend from a single and exclusive ancestor. The alternative to clades are grades in which species descend from a recent common ancestor and express a number of adaptations that are “functionally, behaviourally, and/or ecologically similar” (Strait, 2013: 39). Hence, clades exclusively communicate phylogeny, whereas grades provide significant information about adaptation and only broad information about phylogeny (Strait, 2013). However, others (e.g., Wood and Lonergan, 2008) would debate that grades provide any phylogenetic information. While there are merits to both systems, palaeoanthropologists have not uniformly defined hominin taxa using one system. For example, the genus Paranthropus was introduced to subsume the specimens and species that collectively exhibit extreme megadontia and the osteological scaffolding required to accommodate its highly-developed masticatory musculature (Wood and Lonergan, 2008). However, megadontia is exhibited by all australopithecines and the extreme megadontic adaptations in the ‘robust’ specimens are possibly due to convergence (Foley, 2002). If this is the case, Paranthropus would therefore not be a valid, monophyletic clade and these species could be subsumed under Australopithecus. Yet, under this scenario, some (e.g. Foley, 2002) would still argue to retain Paranthropus as a valid genus (grade), indicating that the ‘adaptive radiation’ the group of species represents is still meaningful, if not monophyletic. Frequently this is implied via the distinction between the ‘gracile’ and the ‘robust’ australopithecines (e.g. Villmoare and Kimbel, 2011). The current classification of the hominin lineage is a confusing mosaic of the clade- and grade-based systems (Strait, 2013). The logistical implications of redefining the hominin lineage to be exclusively clade-based may be a primary deterrent against attempts to implement the change.

Recently, Cartmill (2012) has suggested that all supraspecific taxonomy may be found entirely superfluous. Cartmill suggests that parallelisms are inevitable in closely related taxa due to the similarities in their morphology and the selective pressures that are exerted upon them. The difference between clades at any taxonomic level occurs at the species-level through cladogenesis, and the difference between the two clades, regardless of how adaptively distinct they may seem, is literally only the difference between those species closest to the cladogenetic event (Figure 1). Any future adaptive distinctions between clades are due entirely to subsequent anagenetic and cladogenetic evolution within the clade and due to extinctions (Figure 2). Hence, supraspecific taxa, which are often heralded for containing the suite of traits that define the adaptive shift(s) represented in the clade, are actually entirely illusory (Cartmill 2012). It is that, historically, we have only been able to observe a subsection of all diversity; that which is contained in the world’s extant taxa. The discovery of otherwise unseen diversity in all areas of life via novel, transitionary fossil evidence is arguably reducing the length of the great adaptive leaps that are represented by today’s supraspecific taxonomy. The practical implication of Cartmill’s theoretical argument is that defining hominin taxa based upon seemingly important adaptive distinctions, such as the specialised dietary adaptations of the paranthropines, will become increasingly biologically meaningless as a greater proportion of the fossil record is uncovered. Instead, it will become necessary to either accept that the monophyletic clades used to define taxa are the result of small distinctions between similar taxa, or that higher taxa can include low levels of polyphyly.

The taxon H. habilis was first described following the discovery of the distorted holotype OH 7 at Olduvai Gorge (Leakey, 1964). While OH 7 was dated to be synchronous with paranthropine species, the cranial capacity of OH 7 was estimated to be notably larger (Schrenk, 2013). Hence, H. habilis was suggested to be the first species on the trajectory towards H. sapiens (Schrenk, 2013). While many early Homo-like fossils were attributed to H. habilis, the discovery of two fossils at Koobi Fora, KNM-ER 1470 and KNM-ER 1813 questioned the uniformity of the H. habilis hypodigm. KNM-ER 1470 has a comparatively flat face, a large cranial capacity, and generally larger dentition, while KNM-ER 1813 has overall smaller dentition, cranial capacity and face compared to KNM-ER 1470 (Lieberman et al., 1996). The differences between these two specimens have led to the assertion by multiple researchers (Wood 1992; Schrenk et al. 1993) that early Homo was represented by at least two morphs. The first morph represented by KNM-ER 1813 is H. habilis s.s., while the fossil KNM-ER 1470 represents the other morph, suggested by some to represent a new species, Homo rudolfensis (Schrenk, 2013). In cases where H. rudolfensis is subsumed under H. habilis, H. habilis is considered to be extremely variable (Schrenk, 2013). In order to test the validity of the suggestion that these two morphs should represent distinct species, Wood (1991) performed a character analysis on the early Homo fossils from Koobi Fora. Wood concluded that the high degree of variation in the represented fossils could not be explained by sexual dimorphism alone as the amount of variation exceeded that observed in extant primate taxa. A subsequent study by Lieberman et al. (1996) also corroborated separate specimen groupings representing Homo habilis s.s. (including KNM-ER 1813, 1805, 1501, 1502, OH 7, 13, 16, 24, 62) and Homo rudolfensis (including KNM-ER 1470, 1590, 3732, 1801, 1802). Furthermore, the authors found that many of the morphological similarities between the specimens tentatively assigned to Homo rudolfensis and australopiths, such as their anteriorly projecting zygomatics, are probably homoplastic, which has implications for this taxon’s generic placement and/or the validity of the derived features generally associated with Australopithecus (Lieberman et al., 1996).

In light of the increasing tendency for researchers to recognise the two morphs as distinct species within Homo (Schrenk, 2013), the discovery of a fossil hominin maxilla at Olduvai, OH 65, seriously questioned the validity of Homo rudolfensis as a biologically meaningful taxon (Blumenschine et al., 2003). Blumenschine et al. (2003) outlined that this new specimen was similar to KNM-ER 1470 in its sub-nasal morphology and the position of the anterior portion of its zygomatics. Furthermore, the authors considered that these similarities in specimens, and the fact that OH 65 was discovered in Olduvai, indicated that KNM-ER 1470 and OH 7 were conspecific, rendering KNM-ER 1470 not a member of Homo rudolfensis, but rather of Homo habilis (Blumenschine et al., 2003). The authors still noted the morphological distinctiveness between these specimens and those more similar to KNM-ER 1813, indicating that there was still likely multiple species of early Homo during the Pliocene, though ‘Homo rudolfensis’ would be defunct (Blumenschine et al., 2003). Crucially, however, this reassessment relied upon the supposed similarities between KNM-ER 1470 and the distorted, sub-adult OH 7 mandible and parietals. Recently, Spoor et al. (2015) performed a reanalysis of the OH 7 fossil to correct for post-depositional distortion, and discovered that the dental arcade of OH 7 was more similar to the parallel, primitive condition seen in most australopith specimens, rather than the parabolic dental arcade of later Homo. Additionally, the authors estimated that the cranial capacity of OH 7 was greater than had been previously stated in earlier estimates. This corrected model also allowed Spoor et al. (2015) to re-examine the supposed similarities between OH 7 and specimens of Homo rudolfensis, particularly KNM-ER 1470. The authors concluded that in fact OH 7 and specimens assigned to Homo rudolfensis were morphologically distinct and that the diversity and mosaicism in gnathic morphology and cranial capacity across early Homo specimens warrants the inclusion of at least two Pliocene Homo species (Spoor et al., 2015). While the phylogenetic relationships of H. habilis and H. rudolfensis have yet to be resolved (Wood 1991, 1992; Strait and Grine, 2004), Strait (2013) suggests that these species likely differentiated early in the evolution of Homo, though the noted potential polyphyly of these early Homo species (Stringer, 2015) will likely call for substantial phylogenetic and taxonomic revisions in future years.

Increasingly, the amount of variation that is subsumed under H. ergaster is extending given recent discoveries of fossils of small brain and body size (Lordkipanidze et al. 2007; Spoor et al. 2007). H. georgicus (Gabunia et al. 2002, in Antón, 2013) was initially introduced due to the primitive facial morphology and the cranial capacity of the Dmanisi H. ergaster specimens, such as in D2600 (Antón, 2013). However, Antón (2013) suggests that the range of variation already exhibited in H. ergaster, as well as the apparent similarities between the Dmanisi (D2280 and D2282) and H. ergaster fossils (e.g. KNM-ER 3733, 3833, and WT15000), such as in the moderate size of their supraporbital tori, the presence of a thick vault, and the absence of a compound temporal-nuchal crest (Gabunia et al., 2000), suggest that they are conspecific. Given the diversity of morphologies in H. ergaster from the diminutive Dmanisi fossil D2700 (Rightmire et al., 2006), to the later, more derived morphs, such as Sangiran 17, with its distinctively thick tympanic plate and convex subnasal clivus (Rightmire, 1990), it is becoming increasingly difficult to distinguish between species of early Homo. This is further confounded by an overlap in time of early H. ergaster and other species of early Homo, for almost half a million years (Spoor et al., 2007). As H. ergaster fossils are increasingly more variable, neither body size nor cranial capacity can be used as a proxy for differentiating between H. ergaster and other early Homo fossils (Antón, 2013).

The discovery of the late-surviving but primitive-looking hominin fossils on the island of Flores, Indonesia, have led to a number of possible explanations regarding its phylogenetic placement. Increasingly, similarities between the fossils of Homo floresiensis (especially the holotype LB1) and early Homo (e.g. the absence of laterally projecting “trigones” which are a distinctive feature of Homo erectus) are causing researchers to suggest a possible early dispersal out of Africa, potentially prior to the emergence of H. ergaster (Jungers, 2013). For example, the megadontia in H. floresiensis has been suggested to be most similar to that seen in H. habilis (Brown et al., 2004). While some of the similarities of H. floresiensis to early hominins could be due to reversals, their affinity to early Homo is more parsimonious (Jungers, 2013). Furthermore, in a study by Argue et al. (2006) LB1 was found to be morphologically most similar to specimens of H. habilis and H. ergaster, especially D27000. While it has not yet been established where H. floresiensis fits phylogenetically, it has been suggested by Argue et al. (2009) that it diverged early in the Homo lineage.

Recently, a new species, Homo naledi, has been defined based upon a large assemblage of hominin fossils discovered in the Rising Star cave system, in South Africa (Berger et al., 2015). Following an extensive analysis of the 1500 fossils uncovered, Berger et al. (2015) designated the specimens a novel species of Homo based upon the unique combination of cranial and post-cranial features represented. Crucially, the features that it shares with other Homo specimens are postulated to relate to functional adaptations such as striding-bipedal locomotion (presence of a linea aspera, derived ankle and foot morphology, and a long lower-limb; see also Harcourt-Smith et al., 2015), increased manual dexterity and precision (more derived carpal morphology and the relative proportions of the elements of the hand are similar to that seen in species of later Homo; see also Kivell et al., 2015), and a relaxation on the dietary selective pressures evident in Australopithecus and Paranthropus (absolutely and relatively small dentition, more gracile mandibular corpus, and laterally-oriented temporal muscle attachment) (Berger et al., 2015). However, several elements of the appendicular skeleton were described as more similar to the australopith-condition, particularly in the morphology of the shoulder and the pelvis (Berger et al., 2015). Despite the overall similarities of the cranial morphology to Homo specimens, the cranial capacity of all specimens represented has been reconstructed to be well within the range of variation observed for australopiths and considerably smaller than the majority of Homo specimens (Berger et al., 2015). Given that the assemblage has yet to be dated, the full implication of this new assemblage is largely obscured. However, early musings have indicated that a date of >2myr would likely place the Homo naledi specimens close to the base of the Homo group (Stringer, 2015). A more recent date might have equally far-reaching implications, as it would indicate that a late-surviving species of Homo with an australopith-like cranial capacity may have lived synchronously with Middle and Late Pleistocene hominins (Stringer, 2015). Furthermore, as the spatial range of Homo naledi is also not yet known, it may be that other small-bodied hominins, particularly those attributed to Homo erectus/ergaster are re-evaluated in light of these new finds (Dembo et al., 2015; Stringer, 2015).

Villmoare et al. (2015a) recently reported a new well-preserved fossil mandible, LD 350-1, at Ledi-Geraru, Ethiopia, that has been assigned to Homo. Securely dated to 2.8-2.75mya, it is currently the earliest Homo fossil and has a mosaic of primitive and derived traits (Gibbons, 2015; Hublin, 2015). While LD 350-1 may be a late surviving member of Au. afarensis, given a number of morphological similarities, such as in the overall dimensions of the dentition and mandibular corpus and the orientation of the mandibular symphysis, there are also a number of important differences that the authors interpret as distinct enough to suggest that they are not conspecific (Villmoare et al., 2015a). In particular, the absence of the distinctive wear patterns between the maxillary canine and the mandibular P₃ (Hublin, 2015) and the uniform depth of the mandibular corpus between the P₃ and the M₂ (Villmoare et al., 2015a). Importantly, many of the features that distinguish it from Australopithecus are also features that it shares with early Homo (Villmoare et al., 2015a). Given the similarities between LD 350-1 and the fossils associated with early Homo, the authors suggest that it is a species that is close to the origin of Homo (Villmoare et al., 2015a). Importantly, the fossil pushes the origin of the genus to at least 400,000 years earlier than originally thought. Additionally, given that LD 350-1 was found at 2.8-2.75mya as a likely ancestor to Homo suggests that hypotheses that include Australopithecus sediba as the ancestor to early Homo at a much later date of 1.98mya (Berger et al., 2010) are increasingly unlikely (Villmoare et al., 2015a). Furthermore, the fossil demonstrates that some of the dental and mandibular changes between Homo and Australopithecus occurred early (Villmoare et al., 2015a) and may indicate an important divergence in dietary adaptations (Hublin, 2015).

LD 350-1 has a more primitive morphology of the anterior portion of the mandibular corpus than to specimens attributed to either Homo habilis or Homo rudolfensis though it does display some similarity to Homo habilis, such as in the curvature of the row of the anterior dentition (Villmoare et al., 2015a), which suggests that they may belong to the same lineage, but separated by ~1myr (Gibbons, 2015). Nevertheless, LD 350-1 has been conservatively assigned to Homo species indeterminate pending the discovery of additional fossils (Villmoare et al., 2015a; see also Hawks et al., 2015 and Villmoare et al., 2015b).

The phyletic divergence of Homo has frequently been attributed to adaptive shifts due to dramatic environmental change and the innovation of lithic technology (Antón, 2013). Many authors suggest that it is the increasing aridification (deMenocal, 2004) or some other environmental correlate (e.g. Vrba’s (1993) turn over hypothesis) that explains the synchronous emergence of Homo and Paranthropus. However, Grove (2011) suggests that it may have been selection for phenotypic plasticity that lead to behavioural flexibility in Homo. The period that Grove (2011) associates with a peak in selection for phenotypic plasticity is 2.7-2mya. According to Schrenk (2013), much of the systematics of early Homo are drawn from biogeographic perspectives with respect to the environmental and faunal context in which it speciated. For example, the synchronic appearance of Homo and Paranthropus has been suggested to be due to environmental changes between 2.8-2.5mya in which open habitats extended with more tough food resources (Schrenk, 2013). The expansion of arid environments during this period has been suggested to have influenced the selective pressures differentially in these two lineages (Hublin, 2015). In one, leading to the paranthropines, they became more specialised for processing tough and fibrous vegetation, which selected for the pronounced megadontia and the highly developed masticatory musculature in the associated specimens (Hublin, 2015). In the other lineage, which includes the putative early members of Homo, the hominins developed increasingly omnivorous diets with greater reliance on meat, possibly facilitated by increased encephalisation and the development of - or improvement upon - lithic technologies (Hublin, 2015). Hence, it is suggested that these environmental stressors caused the emergence of Homo and Paranthropus during the Plio-Pleistocene transition (Schrenk, 2013).

Homo habilis has been repeatedly associated with the innovation of lithic technology (Napier, 1962). However, the discovery of OH 62 as reported by Johanson et al. (1987), suggested that some aspects of the morphology of H. habilis were more similar to Australopithecus africanus than other known Homo specimens (Schrenk, 2013). Additionally, in a recent study by Skinner et al. (2015), the structure of the trabecular bone of the carpals of Au. africanus, in addition to other taxa of Pleistocene hominins, indicates that they may have used lithics. While the majority of fossil evidence has thus far indicated an emergence of lithic technologies to coincide with H. habilis (Semaw et al., 2003), recent indirect evidence of tool use was dated to 3.4mya, and possibly associated with Au. afarensis (McPherron et al., 2010). Additionally, the recent discovery at Lomekwi 3, West Turkana, of stone tools dated to 3.3mya and associated with non-Homo fossils further indicates that lithic technologies pre-date the emergence of Homo by at least 0.5myr (Harmand et al., 2015). Hence, if lithics did originate with Australopithecus then this innovation clearly did not require the full suite of Homo-like hand morphology and/or cranial capacity (Skinner et al., 2015). Furthermore, behavioural studies of chimpanzee (Pan troglodytes) tool use, particularly in percussive nut-cracking (Sakura and Matsuzawa, 1991), question whether the innovation of stone tool technologies may have originated in the hominin clade at all. For example, it has been suggested that the knapping techniques that may have been employed to produce the ‘Lomekwian’ tools are similar to those used by chimpanzees during nut-cracking (Harmand et al., 2015). Hence, it may be that the Lomekwian tools represent a transitional technological stage between a method of accidental flake production similar to that observed in contemporary chimpanzee behaviour, and the freehand knapping techniques exhibited in the later Oldowan technology (Harmand et al., 2015; Bril et al., 2015). Consequently, the production and use of various forms of lithic technologies may have been incorporated gradually within various pre-Homo lineages, with perhaps a pre-adaptation originating in their hominoid ancestry.