Elsevier

Journal of Human Evolution

Volume 107, June 2017, Pages 107-133
Journal of Human Evolution

The affinities of Homo floresiensis based on phylogenetic analyses of cranial, dental, and postcranial characters

https://doi.org/10.1016/j.jhevol.2017.02.006Get rights and content

Abstract

Although the diminutive Homo floresiensis has been known for a decade, its phylogenetic status remains highly contentious. A broad range of potential explanations for the evolution of this species has been explored. One view is that H. floresiensis is derived from Asian Homo erectus that arrived on Flores and subsequently evolved a smaller body size, perhaps to survive the constrained resources they faced in a new island environment. Fossil remains of H. erectus, well known from Java, have not yet been discovered on Flores. The second hypothesis is that H. floresiensis is directly descended from an early Homo lineage with roots in Africa, such as Homo habilis; the third is that it is Homo sapiens with pathology. We use parsimony and Bayesian phylogenetic methods to test these hypotheses. Our phylogenetic data build upon those characters previously presented in support of these hypotheses by broadening the range of traits to include the crania, mandibles, dentition, and postcrania of Homo and Australopithecus. The new data and analyses support the hypothesis that H. floresiensis is an early Homo lineage: H. floresiensis is sister either to H. habilis alone or to a clade consisting of at least H. habilis, H. erectus, Homo ergaster, and H. sapiens. A close phylogenetic relationship between H. floresiensis and H. erectus or H. sapiens can be rejected; furthermore, most of the traits separating H. floresiensis from H. sapiens are not readily attributable to pathology (e.g., Down syndrome). The results suggest H. floresiensis is a long-surviving relict of an early (>1.75 Ma) hominin lineage and a hitherto unknown migration out of Africa, and not a recent derivative of either H. erectus or H. sapiens.

Introduction

The taxonomic status of Homo floresiensis (Brown et al., 2004) has been controversial since the species was announced. The strata in which the remains were found are dated to between 65 ± 5 and 89 ± 7 thousand years ago (ka), while the ulna of Liang Bua 1 (LB1) is dated to between 86.9 ± 7.9 and 71.5 ± 4.3 ka and the ulna of LB6 is slightly younger (Sutikna et al., 2016). Fragmentary material from Mate Menge, Flores, dated at ∼700 ka has also recently been discovered and provisionally referred to the H. floresiensis lineage (Brumm et al., 2016; Van den Bergh et al., 2016). Bones that have been placed in the H. floresiensis hypodigm have been interpreted in three broad ways: 1) that H. floresiensis is a distinct species and derived from an early lineage of Homo (e.g., Brown et al., 2004, Falk et al., 2005, Morwood et al., 2005, Argue et al., 2006, Argue et al., 2009, Larson et al., 2007, Tocheri et al., 2007, Brown and Maeda, 2009, Jungers et al., 2009b, Brown, 2012); 2) that H. floresiensis is a distinct species, derived from Asian Homo erectus (Kaifu et al., 2011, Kaifu et al., 2015, Van den Bergh et al., 2016); and 3) that it is not a valid species, but is instead a pathological population of anatomically modern humans (i.e., Homo sapiens) that had genetic or metabolic disorders (e.g., Henneberg and Thorne, 2004, Jacob et al., 2006, Martin et al., 2006, Richards, 2006, Hershkovitz et al., 2007, Obendorf et al., 2008; Oxnard et al., 2010, Oxnard et al., 2012; Henneberg et al., 2014; for reviews to the contrary, see Groves, 2007, Aiello, 2010, Van Heteren, 2013). White (2010) mused that one might also want to consider the possibility that the unique anatomy and small body size might be the unusual result of founder effect and rapid genetic drift in a small population of modern humans colonizing Flores. However, modern humans have dwarfed repeatedly around the world and none converge on the unique proportions and anatomy of H. floresiensis (Jungers et al., 2016).

The hypothesis that H. floresiensis represents a modern human population with a genetic disorder has not withstood testing (Argue et al., 2006, Argue et al., 2009, Falk et al., 2007, Falk et al., 2009, Jungers et al., 2009b, Aiello, 2010, Brown, 2012, Baab et al., 2013, Baab et al., 2016, Van Heteren, 2013). The most recent iteration of these kinds of hypotheses proposes that LB1 had Down syndrome (Henneberg and Thorne, 2004). Baab et al. (2016), however, found minimal congruence between the phenotype of LB1 and clinical descriptions of Down syndrome, and that H. floresiensis remains a phenotypically unique, valid species with its roots in Plio-Pleistocene Homo taxa. We also test whether the phylogenetic position of H. floresiensis we obtained is attributable to Down syndrome, using similar reasoning to Dembo et al. (2015), who note that if Liang Bua 1 had Down syndrome, it should have characters diagnostic of this syndrome and characters that align it with H. sapiens.

Insular dwarfing has been invoked to explain the morphology of H. floresiensis from a H. erectus-like predecessor. The ‘island rule’ for insular dwarfing (Foster, 1964, Alcover, 1976, Sondaar, 1977, Dayan and Simberloff, 1998, Köhler and Moyà-Solà, 2004) stipulates that body size of mammals alters when a founder population reaches an island, becomes reproductively separated from its mainland group, and faces an environment different from those of its mainland sister species. For example, a smaller body size would be expected as a response to a limited food supply, and a larger size to the absence of predation (Foster, 1964). Studies that focus on this phenomenon have disputed the universality of this ‘rule’ (Sondaar, 1977, Heaney, 1978, Wassersug et al., 1979; Lawlor, 1982; Melton, 1982, Libois et al., 1993, Dayan and Simberloff, 1998, Meiri et al., 2008), but it remains the case that rapid insular changes in body size are very common (Millien, 2006).

Van Heteren (2008) argued that the island dwarfing process in large animals typically produces paedomorphic characters, i.e. retention of juvenile characteristics. Van Heteren (2008) therefore tested for paedomorphic features in H. floresiensis and found that its orbital index falls within the range of H. erectus children but not H. erectus adults: in particular she proposed that the high orbital, dental, and brachial indices, low humeral and low tibial torsions, high gonial angle, and shortened lower limbs in H. floresiensis may be explained by paedomorphosis occurring in the dwarfing process of H. erectus, but that additional adaptations also occurred, such as the relatively small brain (to lower the daily energy requirements), while the relatively short legs and broad pelvis were adaptations to low gear locomotion and to increase stability on uneven ground (Van Heteren, 2012). There are numerous problems with this heterochronic explanation, including its generality. Paedomorphic hominids should have relatively large crania (Godfrey and Sutherland, 1996), and LB1's skull is remarkably small relative to any measure of body size. Moreover, no member of the genus Homo at any age of development has limb proportions comparable to LB1 (Jungers, 2013).

Lyras et al. (2009) performed a principal components analysis and cluster analysis of shape as represented by digitized landmarks of a stereolithographic replica of the LB1 cranium and of cranial casts of H. erectus Sangiran 17, Homo habilis KNM-ER 1813, and Australopithecus africanus STS 5; they calculated Euclidean distances among the specimens. Their results (Lyras et al., 2009: Fig. 3) show that H. floresiensis clusters with H. habilis on Function 1 (contrary to their interpretation that H. floresiensis clusters with H. erectus), while Function 2 and Function 3 do not discriminate between H. sapiens, H. erectus, H. floresiensis, or A. africanus. In their weighted pair group method of arithmetic means (Euclidean distances), however, H. erectus and H. floresiensis cluster.

Kaifu et al. (2011) undertook a detailed morphometric and morphological comparison of the LB1 cranium with H. habilis sensu lato (comprising KNM-ER 1470, KNM-ER 1590, KNM-ER 1805, KNM-ER 1813, KNM-ER 3732, KNM-ER 3735, KNM-ER 7330, OH 24), H. erectus sensu stricto (Sangiran and Trinil crania), and the Dmanisi group (putative Homo georgicus). Using 67 cranial characters that they observed on H. floresiensis, they tested four hypotheses: that H. floresiensis was descended from H. habilis, that it descended from the Dmanisi group, that it originated from something similar to early Indonesian H. erectus and subsequently dramatically dwarfed in body and brain size, and that H. floresiensis was not related to any of these species. Assessing the number of cranial characters that exclusively supported each hypothesis, they found that only three traits support a relationship between H. floresiensis and H. habilis; no characters exclusively support a hypothesis for a relationship between H. floresiensis and Dmanisi; and 17 characters supported, or are compatible with, the hypothesis that H. floresiensis was derived from early Javanese H. erectus. More recently, Kaifu et al. (2015) reiterated this hypothesis based upon a comparative study of the dentition. A re-evaluation of the endocranial value of 417 cc for LB1 (Falk et al., 2005) to 426 cc by Kubo and colleagues (2013), based upon micro-CT scans, led them to also prefer the view that H. floresiensis was an island dwarf form of H. erectus, while acknowledging the possibility that H. floresiensis may have dwarfed from a H. habilis-like species.

The island rule relates to changes in overall body mass and does not necessarily predict whether specific body regions would be most affected. Changes in body mass may be achieved in different ways, such as brain, gut, and kidney being traded-off against one another. Nor can we predict specific changes in brain volume or limb morphology after isolation on an island, because this will depend on the selective pressures involved (Niven, 2007). This notwithstanding, the question of allometric scaling of H. erectus in relation to H. floresiensis has been investigated and this explanation was abandoned early on by Morwood et al. (2005). Falk et al. (2005) showed that the brain:body size of H. floresiensis scales with australopith species, even though its endocast shape is similar to Chinese H. erectus (Indonesian H. erectus was not examined). Gordon et al. (2008) scaled six cranial vault measurements to identify the best fit between the LB1 cranium and fossil hominin groups, finding that H. floresiensis scaled with H. habilis, Dmanisi hominin D2700, and Koobi Fora cranium KNM-ER 3733 (Homo ergaster), and not with H. erectus (Sangiran 17). Baab and McNulty (2009) explored the relationship between cranial size and shape, and showed that the morphology of the LB1 cranium is consistent with the expected shape for a very small specimen of archaic Homo and quite distinct from the modern human sample. The high humerofemoral index (limb proportions) for H. floresiensis has also been supposed to be a consequence of its extreme small body size, on the assumption that there exists a negative correlation between overall body size and limb length (e.g., Richards, 2006, Holliday and Franciscus, 2009). To test this hypothesis, Jungers (2009) examined limb proportions of a range of modern humans, including pygmoids, and Australopithecus afarensis (AL 288-1), H. ergaster (KNM-WT 15000), and H. floresiensis. He discovered that intraspecific covariation between limb proportions and body size in H. sapiens was statistically insignificant, so there is no evidence for a biological ‘rule’ relating body size to limb proportions; and that A. afarensis and H. floresiensis fell well outside the range of all human samples, including the small-bodied people. He tested this result in a number of ways; the results led him to conclude that the unusual limb proportions of H. floresiensis are not simply size-related allometries that follow a scaling model for modern humans (which in any case had little support), but can be explained as primitive retentions (also see Jungers et al., 2016).

Argue et al. (2009) used cladistic analyses to test alternative hypotheses that H. floresiensis was a late-surviving member of an early hominin lineage or that it was a dwarfed descendant of H. erectus. Their sample included A. africanus, A. afarensis, H. habilis, Homo rudolfensis, H. erectus s.s., H. ergaster, and a sample of Dmanisi hominins (putative H. georgicus). Using mostly cranial (and only a few mandibular and postcranial) characters, they presented two equally parsimonious phylogenetic trees, each of which placed H. floresiensis in the early Homo section of the tree, close to H. habilis and H. rudolfensis. They recognized that the hypothesis that H. floresiensis is the result of a dwarfing process in H. erectus entails a close relationship between these two species, and they therefore tested for a sister-taxon relationship between H. floresiensis and H. erectus. This alternative placement entailed a markedly longer phylogenetic tree than the most parsimonious topology. Their results, then, led them to hypothesize that H. floresiensis derived from a more archaic lineage than the erectines and that this lineage likely existed at the same time as H. habilis and H. rudolfensis in Africa (i.e., 2.0–1.4 million years ago); they therefore proposed that H. floresiensis represents a very early migration out of Africa that arrived on Flores at an unknown but potentially very early time.

Dembo et al. (2015) applied an extensive suite of craniodental data from the literature to test a range of phylogenetic relationships among several hominin species, including H. floresiensis. Using Bayesian methods, they presented a phylogeny in which H. floresiensis is sister taxon to a branch comprising H. rudolfensis, H. ergaster, H. erectus, Homo antecessor, Homo heidelbergensis, Homo neanderthalensis, and H. sapiens. H. habilis lies outside this branch and forms a clade with Australopithecus sediba.

Most recently, Van den Bergh et al. (2016) identified mandibular fragments and isolated teeth from Mata Menge in Flores dated at ∼700 Ma that they ascribed provisionally to the H. floresiensis lineage. This material is critically important in demonstrating the antiquity, long-term stability (e.g., in body size), and thus non-pathological nature, of the H. floresiensis lineage.

Here we present the results of new parsimony and Bayesian phylogenetic analyses of an expanded morphological dataset that tests the competing hypotheses for the origin of H. floresiensis. The present study uses a range of specimens for Australopithecus and Homo species, and employs characters sampled widely from the cranium, mandible, dentition, and postcranium. This has not previously been attempted and affords a new database to apply to the problem of the phylogenetic position of H. floresiensis.

Section snippets

Materials and methods

Our sample comprises 133 cranial, postcranial, mandibular, and dental characters collected from A. afarensis, A. africanus, A. sediba, H. habilis, H. rudolfensis, H. ergaster, H. erectus (Sangiran and Trinil), H. georgicus, Homo naledi, H. floresiensis, and H. sapiens (Table 1). Our comparative sample does not include Australopithecus garhi or Australopithecus anamensis. A. garhi is relatively poorly represented skeletally, missing 83% of data in our dataset. In comparison, most taxa included

Results

Both the parsimony and Bayesian analyses gave highly concordant results, whether analyses excluded H. rudolfensis (see Figure 1, Figure 2, Table 3A) or included all taxa (SOM Fig. S1, Table 3B). Support was slightly higher when polymorphisms were treated as new states (dataset A: Fig. 1), compared to when polymorphisms were scored as multiple states (dataset B: Fig. 2), suggesting the former coding method extracted a little more phylogenetic signal. In virtually all analyses, the two best

Discussion

Our objective was to test the two mainstream hypotheses for H. floresiensis:

  • i)

    that it derives from a putative population of H. erectus on Flores that dwarfed as an evolutionary response to particular conditions found on islands; or

  • ii)

    it is a late-surviving species of early Homo.

Additionally, we address some issues in relation to the pathology hypotheses for H. floresiensis.

As presented above, all the analyses (parsimony/Bayesian, dataset A/B) were generally concordant regarding this key question, so

Conclusion

We use a dataset comprising 50 cranial, 26 mandibular, 24 dental, and 33 postcranial characters to infer the relationships of H. floresiensis and test two competing hypotheses: H. floresiensis is a late survivor of an early hominin lineage or is a descendant of H. erectus. We hypothesize that H. floresiensis either shared a common ancestor with H. habilis or represents a sister group to a clade consisting of at least H. habilis, H. erectus, H. ergaster, and H. sapiens. A close relationship

Acknowledgements

We thank the Australian Research Council for providing funding under the Discovery Project Scheme (DP1096870). We sincerely thank the following people and institutions for providing access to fossils under their curation: Dr. Berhane Asfaw, National Museum of Ethiopia; Dr. Fachroel Aziz, Centre for Geological Survey, Bandung, Indonesia; Professor Lee Berger, University of Witwatersrand, South Africa; Dr. Tony Djubiantono, ARKENAS, Indonesia; Mr. Tomas Dnyango Mukhuyu and Yared, Curators,

References (136)

  • C.P. Groves

    The Homo floresiensis controversy

    Hayati Journal of Biosciences

    (2007)
  • M. Haeusler et al.

    Body proportions of Homo habilis reviewed

    J. Hum. Evol.

    (2004)
  • T.W. Holliday et al.

    Body size and its consequences: allometry and the lower limb length of Liang Bua 1 (Homo floresiensis)

    J. Hum. Evol.

    (2009)
  • W.L. Jungers et al.

    Descriptions of the lower limb skeleton of Homo floresiensis

    J. Hum. Evol.

    (2009)
  • A. Kramer

    A critical analysis of claims for the existence of Southeast Asian Australopithecines

    J. Hum. Evol.

    (1994)
  • S.G. Larson et al.

    Homo floresiensis and the evolution of the hominin shoulder

    J. Hum. Evol.

    (2007)
  • S.G. Larson et al.

    Descriptions of the upper limb skeleton of Homo floresiensis

    J. Hum. Evol.

    (2009)
  • D.E. Lieberman et al.

    Homoplasy and early Homo: an analysis of the evolutionary relationships of H. habilis sensu stricto and H. rudolfensis

    J. Hum. Evol.

    (1996)
  • I. Martinez et al.

    The temporal bones from Sima de los Huesos Middle Pleistocene site (Sierra de Atapuerca, Spain). A phylogenetic approach

    J. Hum. Evol.

    (1997)
  • M. Martinón-Torres et al.

    Dental remains from Dmanisi (Republic of Georgia): morphological analysis and comparative study

    J. Hum Evol.

    (2008)
  • A. Mounier et al.

    Is Homo heidelbergensis a distinct species? New insight on the Mauer mandible

    J. Hum. Evol.

    (2009)
  • C. Oxnard et al.

    More on the Liang Bua and modern human cretins

    Homo - Journal of Comparative Human Biology

    (2012)
  • H. Pontzer et al.

    Locomotor anatomy and biomechanics of the Dmanisi hominins

    J. Hum. Evol.

    (2010)
  • A.J. Aberer et al.

    Pruning rogue taxa improves phylogenetic accuracy: an efficient algorithm and webservice

    Syst. Biol.

    (2013)
  • L.C. Aiello

    Five years of Homo floresiensis

    Am. J. Phys. Anthropol.

    (2010)
  • L.C. Aiello et al.

    An Introduction to Human Evolutionary Anatomy

    (1990)
  • J.-A. Alcover

    L'evolució de Myotragus Bate 1909 (Artiodactyla, Rupicaprini), un Procés biologic lligat al fenomen de la insularitat. (The evolution of Myotragus Bate 1909 (Artiodactyla, Rupicaprini), a biological process linked to the insularity phenomenon)

    Bull. Inst. Cat. Hist. Nat.

    (1976)
  • P. Andrews

    On the characters that define Homo erectus

  • S.C. Antón et al.

    Evolution of early Homo: An integrated biological perspective

    Science

    (2014)
  • D. Argue et al.

    Homo floresiensis: a cladistic analysis

    J. Hum. Evol.

    (2009)
  • E.H. Ashton et al.

    Cranial crests in the Anthropoidea

    Proc. Zool. Soc. Lond.

    (1956)
  • K.L. Baab et al.

    Homo floresiensis contextualized: a geometric morphometric comparative analysis of fossil and pathological human samples

    PLoS One

    (2013)
  • K.L. Baab et al.

    A critical evaluation of the Down syndrome diagnosis for LB1, type specimen of Homo floresiensis

    PLoS One

    (2016)
  • D.R. Begun et al.

    Events in hominoid evolution

  • L.R. Berger et al.

    Australopithecus sediba: a new species of Homo-like australopith from South Africa

    Science

    (2010)
  • L.R. Berger et al.

    Supporting Online Material for Australopithecus sediba: A New Species of Homo-like Australopith from South Africa

    (2010)
  • L.R. Berger et al.

    Homo naledi, a new species of the genus Homo from the Dinaledi Chamber, South Africa

    eLife 2015

    (2015)
  • J.M. Bermúdez de Castro et al.

    On the variability of the Dmanisi mandibles

    PLoS One

    (2014)
  • J. Braga et al.

    The posterior border of the sphenoid greater wing and its phylogenetic usefulness in human evolution

    Am. J. Phys. Anthropol.

    (1996)
  • P. Brown et al.

    A new small-bodied hominin from the Late Pleistocene of Flores, Indonesia

    Nature

    (2004)
  • A. Brumm et al.

    Age and context of the oldest known hominin fossils from Flores

    Nature

    (2016)
  • M.H. Day et al.

    Functional affinities of the Olduvai Hominid 8 talus

    Man

    (1968)
  • T. Dayan et al.

    Size patterns among competitors: ecological character displacement and character release in mammals, with special reference to island populations

    Mammal Rev.

    (1998)
  • M. Dembo et al.

    Bayesian analysis of a morphological supermatrix sheds light on controversial fossil hominin relationships

    Proc. R. Soc. B

    (2015)
  • D. Falk et al.

    The brain of Homo floresiensis

    Science

    (2005)
  • D. Falk et al.

    Brain shape in human microcephalics and Homo floresiensis

    Proc. Natl. Acad. Sci.

    (2007)
  • J. Felsenstein

    Confidence limits on phylogenies: an approach using the bootstrap

    Evolution

    (1985)
  • C.S. Fiebel et al.

    Stratigraphic context of fossil Hominids from the Omo group deposits, Northern Turkana Basin, Kenya and Ethiopia

    Am. J. Phys. Anthropol.

    (1989)
  • J.A. Finarelli et al.

    Reassessing hominoid phylogeny: evaluating congruence in the morphological and temporal data

    Paleobiology

    (2004)
  • J.B. Foster

    Evolution of mammals on islands

    Nature

    (1964)
  • Cited by (0)

    This paper is dedicated to the memory of our late friend and colleague, Professor Mike Morwood.

    View full text