Enigmatic Cranial Superstructures among Chamorro Ancestors from the Mariana Islands: Comparative Geographic Variation and a Proposal About Their Meaning

Abstract

Sublime expression of three ectocranial occipital superstructures (OSSs)—occipital torus tubercles (TOTs), retromastoid processes (PRs), and posterior supramastoid tubercles (TSPs)—is virtually restricted to Oceania, with epicenters in the Mariana Islands, Tonga, Mocha Island, and perhaps other Oceanic locales such as the West Sepik Coast of New Guinea. Enigmatic in etiology, OSSs are anatomically related to entheses for the trapezius, superior oblique (suboccipital), and sternocleidomastoid muscles, respectively. Our study focuses on Latte Period (950–250 BP) Chamorro ancestors of the Mariana Islands, contextualized with other skeletal samples from Remote Oceania, Near Oceania, and the Asian and American Pacific Rims. Frequent co-variation and pair-wise patterning of multiple markedly expressed OSSs distinguishes ancestral Chamorros from all other populations, but markedly expressed individual OSSs exhibit a broad network of pan-Pacific morphological affinities. The presence of markedly developed PRs and TSPs in archaic Javanese hominins indicates deep Southeast Asian origins for these morphs, but a Northeast Asian origin for tuberculated TOTs is suggested by their earliest presence in Late Pleistocene Okinawans and Neolithic Taiwanese. The central goal of this paper is to present and evaluate evidence that OSSs are informative of both Pacific population history and the life histories of “bone-forming” Pacific Islander and Pacific Rim individuals.

Keywords:

• occipital superstructures
• bioarchaeology
• osteobiography
• Micronesia
• Pacific

ACKNOWLEDGEMENTS

For support and assistance at museum and laboratory work sites, we thank Jaymie Brauer, Susan Rodriguez, and Ian Tattersall (American Museum of Natural History); Toni Han and Lisa Armstrong (B.P. Bishop Museum); Glen Cole and John Terrell (Field Museum of Natural History); David Hunt, Carol Butler, Douglas Ubelaker, and the late Donald Ortner (Smithsonian Institution); Philippe Mennecier (Musée de l’ Homme); and Alan Haun, Diane Trembly, Sue Goodfellow, and David DeFant (Paul H. Rosendahl Ph.D., Inc.). For German and French translation assistance, we thank Jurgen Carson-Greffe, Jean-Jacques Hublin, Marion Reyes, Peter Schuup, and Susanne Wilkins. Information retrieval assistance was provided by Chris Bellessis, Arlene Cohen, Joanne Tarpley Crotts, and Moses Francisco. Map design and production was provided by Marween Yagin, Graphics Specialist (Center for Instructional Support, University of Hawaii at Manoa). For information sharing and/or critical comments on drafts of this paper, we thank three anonymous reviewers, C. Loring Brace, Frank Camacho, Sara Collins, Vincent Diego, David Frayer, Hermann Helmuth, the late W. W. Howells, Rosalind Hunter-Anderson, Marilyn Knudson, Christopher Knüsel, Marta Lahr, Koji Lum, Alan Mann, Chris Meiklejohn, Janet Monge, Darlene Moore, Walter Neves, the late Nancy Ossenberg, Phillip Rightmire, Dirk Spennemann, Mary Torcat, Alan Walker, Milford Wolpoff, and Yigal Zan. Jean-Jacques Hublin (Max Planck Institute for Evolutionary Anthropology) deserves special thanks for providing guidance, encouragement and resources for the beginning phases of this project. Shortcomings of this paper are ours alone.

FUNDING

GH’s research was supported by load allocations from the Dean of the College of Liberal Arts and Social Sciences, University of Guam. MP’s research involving Neolithic skeletons from the Tainan Science Park was supported by The National Science Council (Taiwan), and EA M-S and J-M R-A’s ongoing project “Redrawing the Polynesian Triangle: Did Polynesian settlement extend to South America?” is supported by UOO0926. The late DBH’s research was funded by a grant from The National Institute of Environmental Health Sciences, ES05064.

Notes

1 Apart from the high prevalence of markedly developed OSSs and frequent co-variation of two or more of same, ancestral Chamorros are also morphologically distinctive with regard to humeral robusticity and inferred upper body muscularity and strength. Based on a comparative study of a humeral shaft robusticity index (HSRI-2)—which expresses the shaft’s minimum plus maximum diameters at midshaft as a proportion of maximum length—ancestral Chamorros are close to the extreme upper end of known variation for AMH and fossil hominins (see Heathcote et al. 2012a:157–168, 205–207; 2012b:55–59).

2 As examples, red-slipped pottery known as Marianas Red, derived from ancestral pottery traditions in ISEA, is unique to the four largest Mariana Islands and first appears around 3500 BP (Carson et al. 2013; Moore 2002) or somewhat later (Rieth and Athens 2017). Much later, prehistoric cultivated rice (Oryza sativa L.) and latte stone architecture become distinctive cultural features of life in the Marianas. Within Remote Oceania, prehistoric cultivated rice is unique to the Marianas, while stone latte structures are unique globally. Both domesticated rice and latte are first documented around 1000 BP (Hunter-Anderson et al. 1995), though it appears that rice was not added to the crop inventory of ancestral Chamorros to a significant degree until 400 years later, with the advent of the Little Ice Age (∼600 BP) (Hunter-Anderson 2012).

3 The mutually unintelligible Chamorro and Palauan languages belong to the Western Malayo-Polynesian (WMP) subgroup of the vast Austronesian family, and stand apart from the more than 450 other Pacific region languages in not being part of the Oceanic subgroup (Blust 2000). Historical linguistic relationships and branch classifications are contentious, with most formulations fitting into one of four claims (Reid 2014), viz. that Chamorro is (a) a Philippine-type language, probably most closely related to Ilokano and Tagalog (e.g., Topping and Dungca 1973); (b) most closely related to certain languages in Indonesia (e.g., Zobel 2002); (c) most closely related to some Austronesian languages in Taiwan (see Starosta 1995); or (d) descended from a proto-language ancestral to all Austronesian languages outside of Taiwan, and not closely related to any other WMP language (see Blust 2000). Reid (2002) extends Blust’s (2000) hypothesis, contending that ancestral Chamorros spoke an Austronesian language that developed in the northern Philippines following the first migration out of Taiwan. Lack of close synchronic syntactic and phonological relationship of Chamorro to Philippine languages is interpreted to mean that Chamorro out-migrants left the Philippines before Proto-Malayo-Polynesian differentiated into dialects that later evolved into the Philippine languages of today. Zobel’s (2002) view of the linguistic position of Chamorro is in closest agreement with genetic findings (Vilar et al. 2013a), viz. that Chamorro (and Palauan) are outliers within the putative Nuclear Malayo-Polynesian branch of the Austronesian family, which includes most of the languages of Sulawesi and the Greater Sunda Islands. Further, Zobel speculated that ancestors of Chamorro speakers reached the Marianas from Sulawesi. Blust (2000) and Reid (2002) argue that there is no phonological or syntactic support for this view. Virtually all formulations agree that Chamorro (like Palauan) is an outlier language that does not share a phylogenetic subgrouping, let alone mutual intelligibility, with extant Austronesian languages, due to its evolution in relative isolation from other languages deriving from Proto-Malayo Polynesian (Bellwood 1991).

4 Chamorro distinctiveness, biological, cultural, and linguistic, surely relates to the remoteness of the Marianas archipelago, separated from various proposed ISEA homelands by more than 2000 km (Petchey et al. 2016; see also Fitzpatrick and Callaghan 2013), perhaps the longest expanse of open sea traversed anywhere in the world (Craib 1999). From where the first visitors/settlers to the Marianas originated is a matter of intense debate. The northern Philippines was recently proposed by Hung et al. (2011), based on archaeological (ceramic), linguistic, and genetic considerations. Countering this proposal, Winter et al. (2012) made a case for origins further south in ISEA, based on ceramics, linguistics, and oceanographic considerations. More recently, Fitzpatrick and Callaghan (2013) employed seafaring simulation models and concluded that the Marianas were most likely settled from Halmahera, the Bismarcks, or northern New Guinea. Most recently, Montenegro et al. (2016) situated the Solomons and New Guinea as most likely homelands of the earliest visitors/settlers to the Marianas, based on a seafaring simulation study that employed shortest-hop trajectory data.

5 The latter include a series of depopulation events, beginning with the late seventeenth-century Spanish-Chamorro wars and subsequent forced population relocations (Shell 2001; Underwood 1973) that set the stage for a severe mid-eighteenth-century population bottleneck. The total Marianas population of Chamorros, just before the first Jesuit mission in Guam (282 BP, i.e., AD 1668), is estimated to range from 24,000 to 100,000 (Shell 1999; Underwood 1973). By the time of the first Spanish census in 240 BP (i.e., AD 1710), the native population in Guam and Rota declined to under 4,000 (Shell 1999) and further to 1,700 by 192 BP (i.e., AD 1758), perhaps its lowest point in the post-contact history of the Marianas (Goetzfridt 2014). Besides direct and indirect mortality from the Spanish-Chamorro wars, the precipitous depopulation of the Marianas was due to introduced infectious disease epidemics and typhoons (Garruto 2012; Plato and Cruz 1967; Reiff et al. 2011), complicated and exacerbated by malnutrition, starvation, social disruption, and emotional trauma (Goetzfridt 2014; Leon-Guerrero 2015). Within this context of such social upheaval, birth rates would have been negatively impacted. While important in explaining the population structure of contemporary Chamorros, the impact of the eighteenth-century population bottleneck is moot to this study, for few (if any) of the ancestral Chamorros studied lived more recently than 250 BP (i.e., AD 1700) (Table 1).

6 While Valentin and colleagues (2005) referenced our protocol in describing OSSs in a single burial (dated to ̴ 1800–2400 BP) from Mangaliliu village in Vanuatu, no individual OSS scores were provided. The superstructures were described collectively as weakly developed, presumably <“2” according to our scoring scheme.

7 In a study of cranial variation across nine regional groupings, Lahr (1996) provided bar graphs indicating that strongly expressed occipital tori (scores of 5–7, according to her system) are most frequent in Fuegian–Patagonians, followed by Australians, then East Asians. She also recorded a zero frequency of pronounced OT samples of Europeans and Middle Easterners, and small prevalences in Sub-Saharan Africans, Southeast Asians, and North Africans. Lahr’s (1996) data are not included in Supplementary Data: Appendix Table 1, as frequencies and prevalences were not provided.

8 Regarding Middle Pleistocene European and East Africans, Hublin (1978c:138) described “trace” PR expressions on the occipital bone from Bilzingsleben (Germany), and Rightmire (1990:210, 217) identified “small, poorly defined” PR in the Petralona (Greece) cranium and a “low relief” outline of the superior oblique muscles in the Lake Ndutu, (Tanzania) cranium. The Spy 2 (Belgium) cranium is said to be distinguished from other Neandertal males in having a “distinct retromastoid process”, while the early modern Mladeč 5 (Moravia) cranium is described as bearing a “prominent retromastoid process” (Frayer et al. 2006:232). Magnified images of Spy 2 and Mladeč 5 (Wolpoff et al. 2001) reveal their PRs to be slight and moderate in development, respectively, using our protocol.

9 The poorly dated Sangiran individuals may range from Early to Middle Pleistocene, and probably antedate the late surviving H. erectus groups at Ngandong and Sambungmachan (who lived at least 100,000 BP and possibly <50,000 BP) by several hundred thousand years (Anton 2003; Bartstra et al. 1988; Swisher et al. 1996; see also Indriati et al. 2011).

10 This proposal finds support from Anton (2003; see also Baab 2011), whose work on morphological distinctions between Chinese and Indonesia H. erectus crania led her to conclude that intermittent isolation produced and maintained regional morphs in Southeast Asia.

11 This consistent discrepancy suggests that Waldeyer’s (1909) “weak” PR scores are equivalent to what others considered absent (see Supplementary Data: Appendix Table 2). Over-reporting is further suggested by his claim of moderate total frequencies (14.3% and 20%) in two European series, in contrast to Le Double and Dubrueuil-Chambardel (1905) and Michelsson (1911), who reported low frequencies, ranging from 1.3% to 1.4%, for three large European series. Waldeyer’s PR frequency ranges of 3.7% to 27.6% for Asian and Native American groups are also exceptional, in contrast to ranges of nil to 4.7% for three Asian/New World groups reported by Michelsson (1911) and, later, Hublin (1978c).

12 Jacob reported that “grade 3” TSPs were not found in male or female African Americans and Central Javanese, nor among female Alaskan “Eskimos” and Central Europeans. Indeed, only five observed sides of such TSP expressions were recorded, in 3.5% (3/86) of Alaskan “Eskimo” and 4.3% (2/46) of Central European males.

13 For example, Matiegka’s sometimes interchangeable use of AP and the German term for “asterial crest” is problematic, as the latter refer to folded bulges along the lambdoidal suture near asterion, features that we would score as an incipient, slight, or moderate TSP, at most. Drawing on his own research, Matiegka noted that Lapp, Samoyed, Chinese, Malayan, Javanese, Makassarese (southern Sulawesi), “Negro,” Kabyle, (northeast Algerian Berbers), and Egyptian mummy samples lacked APs, as well as noticeable ridges or crests in the asterionic region. In contrast, AP and/or asterial crests and ridges were reported for Native North and South Americans (e.g., Paiutes, and skulls from Florida, Santa Rosa Island [California], Chile, the Argentine pampas, and Tierra del Fuego). He reported a strongly developed asterial crest in 25% (4/16) of aboriginal Santa Rosa Islanders, a generally less developed crest in 10% of a larger cranial series from Bohemia and weakly developed processes in a small series of three Greenlandic Inuit (Matiegka 1906:368–369).

14 Interpretation of the archaeological record bearing on the peopling of the Mariana Islands is contentious. For example, Carson (2014:74) maintains that Early Unai archaeological sites provide evidence of a “formal population migration” to the Marianas, while Hunter-Anderson (2013) interprets these earliest Pre-Latte sites as bearing witness to temporary visitations by seasonal marine foragers, not settlers. Colonization was surely underway by Middle Unai times, however, as witnessed by recovery of 177 early human burials from the Naton Beach site in Guam. Combined calibrated AMS ¹⁴C age (expressed in 2σ) for Conus shell bead necklaces from four of these individuals ranges from 2740 to 2280 BP (i.e., from Middle Unai to Late Unai Periods [DeFant 2008]).

Beyond the debates about when the islands were settled is the question of in situ continuity and adaptive change vs. population replacement vs. intermediate models combining continuity with external influences from Pre-Latte through Latte Period times (see Hunter-Anderson and Butler 1995:28; Rainbird 2004:131–132; Vilar et al. 2013a). Partitioning the present Mariana Islands samples into Pre-Latte and Latte analytic units informs debate on settlement history (see Walth 2014), but cannot lead to resolution because interpretation of morphological differences between Pre-Latte vs. Latte samples is confounded by motor behavioral changes associated with the onset of latte house building and proposed physiological plastic skeletal responses to motor activities involved (see Heathcote et al. 2012b:54). Regardless of the degree to which Pre-Latte populations were ancestral to Latte Period Chamorros, the invention, development, and diversification of work involved in latte house construction complicates interpretation of Chamorro ethnogenesis, assuming the correctness of our multivariate model of OSS development, which includes mechanical induction as a co-factor.

15 Further, the presence of latte construction specialists, unevenly distributed throughout the Marianas, cannot be ruled out. While undocumented for the Marianas, clans of specialist builders are known for Kiribati (Hockings 1989) in eastern Micronesia, a similar semi-sedentary society with an incipient political hierarchy (see Heathcote et al. 2012b:60–61). Alternatively, we cannot rule out that our pooled Saipan sample might be non-representative. Earlier studies (Hasebe 1935; Schlaginhaufen 1906) offer little help, for while markedly developed TOTs and TSPs of Saipanese Chamorro male crania were illustrated, frequencies were not reported.

16 An outline of proposed muscle actions related to latte construction, including attendant megalith transport and placement, is presented in Heathcote et al. (2012b:54). The actions of the TOT-associated upper trapezius include helping with neck extension and bending the neck from side to side, drawing the clavicle and scapula backwards, elevation and rotation of the scapula, and supporting the clavicle and scapula when heavy weights are born by hands when arms are down at the side. The superior oblique (suboccipital), anatomically related to the PR, bends the neck backwards and rotates it from side to side. The TSP-associated sternocleidomastoid draws the neck forward, raises the neck while supine, raises the chest in forced breathing, tilts the neck toward the shoulder, and rotates the neck.

17 While the historical record is mute on the physical labors involved in operating traditional Chamorro sailing canoes (proas), Gladwin (1970) provides such an account for Puluwatese crew members on a traditional outrigger sailing canoe. Puluwatese canoes are good proxies, due to design similarities to Chamorro proas (Hornell 1936). Thus, the work involved in maneuvering them must also be similar. Among Puluwatese navigators, routine labors include paddling, hoisting the sail, bailing water, steering with a big steering paddle, and adjusting the sheet (sail). The most rigorous work occurs when wind changes necessitate tacking, when the sail is moved from one end of the canoe to the other. Illustrations of this process Gladwin (1970:104–105) show 11 steps from releasing the tack line at one end of the canoe to securing it to the end stay at the other end. Crew members engage in rig carrying, heaving, and pulling movements. A movement that especially loads the head and neck dynamically is overhead rig heaving with the head and neck forwardly flexed. Thus, sailing canoe voyaging would involve the SCM in ways with the potential to transform the TSP site (in predisposed individuals).

18 Unpublished images of aboriginal Northwest Coast skulls from British Columbia, provided by Christopher Knüsel (Université de Bordeaux), illustrate quite robust crania, but with only moderate (non-tuberculated) developments at the TOT (and PR) site.

19 Unpublished images of two Fuegian Haush (courtesy of Walter Neves, University of São Paulo, Brazil) and published illustrations of Fuegian-Patagonian skulls (Hrdlička et al. 1912:plate 38b) reveal that while one Patagonian and two Haush skulls bear exceptionally robust OTs, none bear TOTs.

20 Should systematic surveys verify the latter, such a distribution would contribute to the debate on the peopling of the Americas. Intriguingly, Neves et al. (1999a) have demonstrated morphometric affinities between the Haush and three other Tierra del Fuegian groups with two Polynesian groups (Mokapu, Hawaii, and Moriori from the Chatham Islands) accessed from Howells’s (1989) global craniometric database. While outside the purview of this paper, Lahr (1995), Munford et al. (1995), Hernandez et al. (1997), Neves et al. (1999a, 1999b), Matisoo-Smith and Ramírez (2010), and Ramírez-Aliaga (2010, 2011), among others, provide alternative interpretations of distinctive Fuegian-Patagonian craniofacial morphology and the wider issue of pan-Pacific contact between Remote Oceanic and South American populations.

21 These findings may bear on the “dual structure model” of Japanese population history (K. Hanihara 1991), which posits a more direct ancestral connection between the Jomonese and modern Ainu of Hokkaido, Sakhalin, and the Kuriles, as well as Ryukyuans (Hudson 1999), than to modern Japanese. According to K. Hanihara (see also Brace et al. 1989), modern Japanese of the main islands are a differentially admixed product of two ancestral groups: Neolithic Yayoi migrants from North East Asia and the longer-established Jomon autochthons of proximate Southeast Asian Paleolithic heritage. In comparative craniodental and craniometric and studies, respectively, T. Hanihara (1992) and Pietrusewsky (1992) offer support for the dual origin model. In contrast, Turner (1992) found evidence for Japanese admixture with Jomonese-Ainu, but also a close relationship between modern Japanese and Northeast Asians (“Sinodonts”), based on his study of dental nonmetric traits. Concerning affinities of prehistoric Japanese to Remote Oceanic populations, Brace (e.g., Brace and Hunt 1990) and K. Hanihara (e.g., K. Hanihara et al. 1993) found craniodental and craniometric support, respectively, for Jomon-Pacific Islander population historical connections. Such close Jomon-Pacific Islander affinities are not supported by Pietrusewsky (e.g., 1996, 2006a), in independent craniometric studies.

22 Distribution may include Tierra del Fuego, if any of Matiegka’s (1906) attributed “strong” AP expressions in Fuegians are morphologically equivalent what we score as markedly developed TSPs.

23 If Swisher et al. (1996) are correct in their indirect dating of the Ngandong and Sambungmachan fossils to between roughly 27,000 and 53,000 BP, genetic continuity of a late-surviving population of Indonesian hominins with modern Pacific Rim and Oceanic populations would be problematic, given the advanced H. erectus morphology of these fossils (see Hawks et al. 2000).

24 Given the longstanding bias “towards all things ‘Polynesian’… [in] scholarship on prehistory in the Pacific Islands” (Terrell 2011), a multitude of migration models have been developed concerning the spread of humanity into the non-Micronesian islands of Remote Oceania (see Bellwood 2011; Gosling et al. 2015; Greenhill and Gray 2005; Kirch 2010; Pietrusewsky 2012; Pietrusewsky and Douglas 2016; Sheppard 2011). Historically, treatment of the colonization of Remote Oceania has most often focused on Lapita expansion and the Polynesian diaspora (see Green 2003), and the competing settlement scenarios vary from one another along parameters of demographic viability, directionality, speed of migration, and the natural and cultural environments at the destination (Sheppard 2011:806), factors that surely apply to the peopling of Island Melanesia and Micronesia.

25 Remote AMH ancestors, including some who appear to have carried the genetic potential for developing marked expression of OSSs (through assimilation of Indonesian H. erectus and/or Late Pleistocene northeast Asian genes), were hunter-gatherers, probably of variable lineage origins stemming from several late Pleistocene out-of-Africa AMH colonization events (see Reyes-Centeno et al. 2014). By 50,000–60,000 BP, AMHs began colonizing Sunda (the extension of the SEA continental shelf that included the Malay Peninsula, Sumatra, Borneo, Java, and surrounding islands) and, perhaps somewhat later, Sahul (the great southern landmass which later became New Guinea and Australia) (Morley 2017), but earlier arrivals, dating to the Middle/Late Pleistocene boundary, are suggested by other studies (Corny et al. 2017). Whereas the first occupation of Near Oceania began during the Late Pleistocene, the first expansion into Remote Oceania began much later, during post-glacial Holocene times, from multiple origination points (Montenegro et al. 2016), initially out of southern China, Taiwan, the northern Philippines, and Indonesian archipelago (Ko et al. 2014; Lipson et al. 2014; Oxenham and Buckley 2016). Southeast Asian voyaging populations—those who were not a part of the Lapita cultural complex—ventured southward, then east into western Micronesia (the Marianas, Palau, and, perhaps, Yap) (Clark et al. 2006; Fitzpatrick and Callaghan 2013; Hung et al. 2011). Contact with the Mariana Islands by sea-goers from ISEA occurred ∼3500 BP (Vilar et al. 2017) or somewhat later and, around the same time or somewhat earlier, there was movement of sister ISEA populations of Austronesian-speakers associated with the Lapita expansion into the Bismarck Archipelago (Rieth and Athens 2017), the Solomons, and eastern island Melanesia, eventually reaching Tonga and Samoa in Western Polynesia by ∼3000 BP (Petchey et al. 2010; Vilar et al. 2017). After a pause of approximately 1,000 years, navigators migrated to the furthest reaches of Remote Oceania (Rapa Nui, Hawai’i, and New Zealand [Bellwood et al. 2011; Kirch 2010]), as late as 800 BP (Hunt and Lipo 2006; Pietrusewsky 2012).

26 Such morphological continuity of these archaic humans to Southeast Asian, Oceanic, and Pacific Rim AMHs is supportive of the Assimilation Model of modern human origins, whereby “small but not insignificant anatomical contributions” from archaic to modern humans occurred (Smith et al. 2016:126), in this case, among more immediate ancestors of Oceanic and Pacific Rim AMH populations.

27 To date, aDNA analysis has been conducted (but not fully reported) on only one group from our study, viz. Latte Period remains from the Grand Mariana Resort site, Anaguan (Garapan), Saipan (Table 1). Perzinski and Dega (2016) report that their aDNA study was handicapped by poor biomolecular preservation. Geneticist John Dudgeon was able to partially sequence 13 individuals for ancient mtDNA, and all exhibited Haplogroup E variants with probable affiliation with ISEA E1 and E2 haplogroups. There was no evidence for the Chamorro motif or any other Oceanic B4 lineage (Dega et al. 2017).