This website is a free e-resource, from the book entitled 'Music, Lapita, and the Problem of Polynesian Origins' by Mervyn McLean. The complete text may be read either online or in paginated form as a downloadable pdf here.


Chapter 14

Sea levels


If Micronesians contributed to the peopling of Polynesia, the path they took must have included atolls. The major obstacle for believing that this may indeed have happened, lies with sea levels and associated radiocarbon dates for Micronesian settlement. Kirch (2000:174) refers to Sea levels 1 to 1.5 m higher than at present between 2000 BC and  AD 1 when atolls are assumed to have been unoccupied. This was followed by emergence of atolls when the present low sea level was reached by about  AD 1 and the atolls became available for occupation. These dates of about 4000 BP and 2000 BP respectively have been accepted by most anthropologists, ruling out Polynesian contact from Micronesians until after the latter date. Review of the scholarly literature relating to sea levels, however, throws doubt on both of these estimates as well as assumptions underlying much of the research.


Geological ages and the "ring of fire"

The coming of humans to Oceania has occupied only a tiny blip on the vast panorama of forces that shaped the environment they would ultimately occupy.

Beginning with a single super-continent that joined present-day Antarctica with South America, New Zealand and Australia, the process of continental drift separated them and created the Pacific Ocean with its tectonic plates and under-sea mountain ridges and deep-sea trenches that are still on the move, creating areas of subsidence and uplift that at times are sufficiently violent to generate tsunamis powerful enough to wipe out coastal communities and island habitats.

Other movements are so slow as not to be readily observable, as demonstrated by the little island of Niue in Western Polynesia, which has undergone a rollercoaster ride since its creation:

Around 600,000 years ago, Niue was an atoll, barely rising above the ocean surface. Then, as it was carried westward on the moving Pacific Plate toward the Tonga-Kermadec Trench, it began ascending the eastern side of the associated bulge. It has now been uplifted about 70 meters and is about halfway up the bulge. After the island has passed over the crest, it will begin to subside, eventually being drowned and perhaps ending up wedged against the western wall of the Tonga-Kermadec Trench just as Capricorn Seamount is today (Nunn 2008:22-3).

Meanwhile, around the outer fringes of the Pacific, occupied by the so-called "ring of fire" with its active volcanoes and earthquake zones, more dramatic events have occurred, including, at times, the appearance or disappearance of entire islands.[5]

The process of island formation encompasses the entire spectrum of geological time in Oceania, beginning essentially with the creation of the world and ending with the present day.


Types of island

Three broad categories of island can be distinguished, known respectively as volcanic, atoll, and makatea. Taking climate and vegetation into account,

 Oliver (1961:10-13) subdivides these into seven basic types, two volcanic,

three coral atoll, and two makatea. The term "volcanic" does not in this sense refer necessarily to volcanoes but rather to volcanic processes of island formation.

  • Weathered volcanic islands are arguably the most attractive on which to live. They are mountainous high islands with a variety of plant environments and resources. Typical examples include Hawai'i, the Society Islands, and Samoa, and the Micronesian island of Pohnpei (Ponape) is also in this category.
  • Unweathered volcanic islands include the Northern Marianas of Micronesia. They have less soil and vegetation than the weathered islands but are still capable of supporting moderate numbers of people.
  • Treeless atolls or coral islands lack good soil and drinking water, and support only sparse vegetation, making them unsuitable for permanent occupation. Canton Island in the Phoenix group of Eastern Micronesia is an example. 
  • Dry forest atolls or coral islands. In this category are most of the Marshall Islands of Micronesia, along with Tuvalu in Western Polynesia, and many of the Tuamotus and Northern Cook Islands in Eastern Polynesia. Strand flora, arable soil and fresh water, coupled with plentiful marine resources, provide support for limited numbers of people.
  • Luxuriant moist atolls or coral islands. Typical examples are Kiribati and Ulithi in Micronesia, and the Tokelau Islands in Western Polynesia. These are the idyllic coral islands of Hollywood and romantic fiction. They have sufficient soil to support both dry and wet land crops.  
  • Raised coral islands. These islands grew as a result of successive upthrusts of old coral reefs, resulting typically in the formation of a limestone plateau bounded by steep cliffs on the seaward side. This plateau, known as makatea in Polynesian languages, has thin pockets of soil with dryland vegetation, and few sources of fresh water. Coconuts and other food plants can grow fairly well in the available pockets of soil but droughts are a frequent limitation on agriculture. Examples of makatea islands are Niue in Western Polynesia and Nauru in Micronesia, both valuable for their deposits of phosphate.  
  • Also ranking as makatea is a category Oliver calls Mixed islands, combining the characteristics of volcanic islands with makatea. Examples are Guam, Saipan, and Rota in the Mariana Islands of Western Micronesia, and Mangaia in the southern Cook Islands of Eastern Polynesia.


Atoll formation

Common to most atolls is a circular coral reef enclosing a lagoon which may or may not be occupied by a small island or islets whose height does not, as a rule exceed three or four metres above sea level. On the seaward side of the enclosing reef, however, the sea plunges to great depths. The earliest explanation of this, to be refuted by Charles Darwin, was a seemingly plausible idea that the reef must surround the crater of a sunken volcano. Darwin realised, on the contrary, that the commonly present islet in the centre of the lagoon was a remnant of a sinking mountain upon the flanks of which the coral reef was embedded. As the central structure continues to sink, the corals keep pace by building upward and outwards. All that is now known about coral islands supports this hypothesis, which remains a tribute to Darwin. In the Pacific, especially, it neatly explains the various forms that atolls take, demonstrated as follows by Fischer (1956:4) for the Eastern Caroline Islands:

  • Stage 1 is a fringing reef. Kusaie is the closest to this stage.

It is a high island with only a narrow reef around it and no large deep lagoons between the reef and the shore.

  • Stage 2 is a barrier reef, demonstrated by Ponape which is further along in the process.

As the island has subsided the sea has drowned the lower parts of the river valleys. Around most of the island the reef is separated from the mainland by a deep lagoon, although the lagoon is lacking on parts of the southeast coast.

  • Stage 3 is disappearance altogether of the central island, exemplified by Truk, which is sometimes described as a complex atoll. The land has sunk so much that the land mass has been broken up and only the tops of the highest mountains show above the water. The encircling reef is five to twenty miles distant from the main islands. The numerous atolls among the low islands illustrate the completion of the development.

Readers may have noticed that although Fischer is speaking of atoll development in the above descriptions, he has chosen high rather than typically low islands to exemplify the first two stages. This serves to illustrate two important elements of Darwin's theory which could be otherwise overlooked. First, as Darwin himself took pains to emphasise in his writing, his theory of atoll development, unlike the volcanic crater theory it superseded, accounts for barrier reefs in all situations, including continental ones such as the Great Barrier Reef of Australia. And secondly there is no suggestion that all such events took place at the same time. To the essential element of subsidence in Darwin's theory, however, must be added the complication of sea level rise and fall, considered next. 


The tides of time

Every one familiar with the sea is aware of the daily rise and fall of the tides, and perhaps also of longer-term highs and lows known as spring and neap tides. On a scale of centuries and even millennia, however, are the accumulated results of tiny increments of not more than a millimetre or two a year, though closer scrutiny may show the effect to be taking place not uniformly but in unpredictable jumps. Scientists have long known that these fluctuations result from changes in the volume of sea water, caused principally by formation and later melting of polar ice in an eons-long cycle of successive Ice Ages and Interglacials. From the perspective of human history, just two of these eras are relevant: the closing years of the last Ice Age or Pleistocene, and the most recent and still current interglacial period or Holocen, defined as beginning 10,000 years ago.


A note on method

A term to be found in most of the scientific literature on sea levels is "eustatic", referring to world-wide processes affecting sea levels, in contrast with local or regional ones. An unspoken assumption behind the term is the common sense belief that water finds its own level, and consequently, as the oceans of the world are interconnected, a change of level in one ocean will sooner or later flow through to the others. On the other hand, mariners have known for generations that sea levels are not the same everywhere, and anyone who has been through the Panama canal since its opening more than a century ago will probably be aware that the mean sea level at the Pacific end of the canal is eight inches or twenty centimetres higher than at the Atlantic end, leading to evidently erroneous speculation that the Pacific would somehow drain into the Atlantic if the canal had been at sea level.

 As a debate this is not unconnected with problems confronting the scientists whose work until the 1970s centred upon the measurement of old shore-lines in order to chart eustatic sea levels. It was not until the following decades that attempts were made to come to grips with the many non-eustatic variables that were affecting results.


Sea level chronology

The following are dates and events as accepted currently by most writers, and reported principally by Nunn (2008). All dates are approximate.

  • 30,000-17,000 BP. Last Ice Age (Nunn 2008:20). Sea levels were substantially lower than today, allowing movement of peoples along now submerged coastal areas, or across land bridges that no longer exist. During this period Papuans may have reached as far as the Bismarck Archipelago and the northern Solomons.
  • 18,000 BP. Coldest period of last Ice Age. Sea level c.120m lower than today (Nunn 2008:30).
  • 15,000 BP. End of last Ice Age (Nunn 2008:30). As ice sheets melt, sea levels begin to rise, doing so non-uniformly in a series of steps, punctuated by periods of slower rise. Atoll building keeps pace within reach of sunlight in the "photic zone" (Nunn 2008:28), a short distance below the surface.
  • 14,200 BP. CRE-1, first of three Catastrophic Rise Events. Sea level rises 13.5m in 290 years (Blanchon and Shaw 1995).
  • 11,500 BP. CRE-2, second of three Catastrophic Rise Events. Sea level rises 7.5m in 160 years (Blanchon and Shaw 1995). 
  • 7600 BP. CRE-3, last of three Catastrophic Rise Events. Sea level rises by as much as 6.5m in 140 years (Blanchon and Shaw 1995; Nunnn 2008:76). According to Nunn (2008:78) this provoked movements by coastal-dwelling peoples of SE Asia that resulted ultimately in the Lapita expansion.[6]
  • 4000 BP. Sea levels risen to about 1.5-2.1m above current levels. At this point coral reefs are exposed at low tide forming a base for atoll formation as sea levels subsequently fall (Nunn 2008:30-1, note 62 ). Cites Grossman et al. 1998 for this date and elevation. The fall after 4000 BP results principally from a process known as equatorial ocean siphoning, which redistributes equatorial water back towards the poles (Mitrovica and Peltier 1991; cited by Dickinson 2003:489). In Fiji sea level has fallen a net 1.5-2.1 meters in the past 4,200 years (Nunn and Peltier 2001, cited by Nunn 2008:Ch.3, note 62,63. Dickinson 2003).
  • 2000 BP. Atolls believed now sufficiently developed for occupation (Nunn 2008:33, note 63). Cites Dickinson 2003.

Although widely quoted in sea level literature, there are huge margins around most of these dates, both chronologically and geographically, and results for individual islands are especially open to question as will be seen in the next section. Terms frequently to be met with in it are "highstand", referring to sea level elevations above present, and "crossover", which is a term used by Dickinson for the date at which atolls are said to have become available for occupation, calculated as the date "at which declining high-tide level fell below mid-Holocene low-tide level". (Dickinson 2003:489).


Surveys of sea level literature

Three major surveys have been published. Pirazzoli and Pluet (1991) reports upon a database of 77 regions from around the world, including islands of Oceania. Grossman et al. (1998) updates and refines results from the same database, and Dickinson (2003) reaches radically different conclusions about dates  after reinterpretation of earlier studies, including a comprehensive survey of his own, published just two years previously (Dickinson 2001). 

It can safely be assumed that the starting points for potential migrations into Polynesia, whether from Micronesia or from Melanesia, would have been high islands, and for these sea level data is irrelevant as occupation of these islands would have been possible at any time, even if present-day strand areas were absent. In this category are the Fiji Islands as the most likely portal out of Melanesia, and the Marianas, Yap, Palau, and Ponape in Western and Central Micronesia, along with the raised coral island of Nauru, halfway between Western and Eastern Micronesia, which could have acted as a way station between these two areas. It is instructive, nevertheless, to begin with them for the light they can shed on past sea levels generally.



From research carried out in 1988, Pirazzoli and Pluet (1991:145-7) document a gradual rise from c. +1 m to +2 m in Vanua Levu between 6000 and 4000 BP, followed by a sudden sea-level drop which occurred some time later than 3400 BP.

On the southern coast of Viti Levu a gradual rise is calculated to have taken place from 8000 to 4000 yr BP, when sea level reached a peak at about +1 m, followed by a fall to the present situation probably before 3000 yr BP and an almost stable state since that time.

On the face of it, this is a highly important departure from the prevailing view that sea levels were a metre or more above present 4000 years ago and did not decline to present levels until 2000 BP when atolls are said to have emerged. If the above information is correct, in the critical Fiji area reached by Lapita potters, the peak of the sea level rise is confirmed at 4000 BP, but reached and maintained present levels at least a thousand years earlier than the 2000 BP estimate, at the very time the Lapita people were moving on to Tonga and Samoa.

Two further sets of results are reported by Grossman et al. (1998), the one in agreement with the above and the other conflicting with it. One or other, therefore, must be wrong. The contrary results, all dated in the 1990s, suggest Fiji underwent a sea-level highstand 1.5 m higher than present between 3000 and 1000 BP. On this reckoning, sea levels could still have been higher than present a thousand years later rather than earlier than 2000 BP, which might therefore be regarded as a reasonable average.

 Even greater non-convergence of results, however, emerges from work by Dickinson 2003:494 who places the inferred end of highstand in Fiji at 1200 BC and "crossover date" when atolls would have emerged at AD 500 or 3200 BP and 1500 BP respectively, both much later than earlier estimates. With such disparity of results, some explanation is in order before proceeding further:


Problems of interpretation

In view of all that is known about vulcanism, plate tectonics, and a host of other variables that can affect apparent sea levels, it is surprising that little or no account was taken of these in earlier work on sea levels. Instead, it seems that elevations of former shore lines were interpreted solely on a model of world-wide eustatic change taking place uniformly and at similar rates everywhere. 

Pirazzoli and Pluet begin their world survey with the observation that if all of the 1960s and earlier results were accepted

sea level would have had to be, at one and the same time, higher, equal, and lower than at present and remain stable, though at the same time fluctuating (Pirazzoli and Pluet 1991:4).

Nor was this all. Subsequent efforts to make sense of disparate results have run into similar difficulties, largely, it would seem, because of neglect of local and regional variables. Thus, in their summing up, Pirazzoli and Pluet observe:

The average of all curves since 10,000 yr BP is a relative sea-level drop of about 8 m.  This value, which reflects the predominance of glacio-isostatically uplifted areas in the sample of curves available, is obviously no more representative of the global eustatic situation than would be, for the last century, the average of all tide-gauge records . . . The high variability from place to place confirms indubitably that it is impossible to estimate directly, from field data alone, the amount of the eustatic change since 10,000 yr BP (or since any other time) (Pirazzoli and Pluet 1991:231). 

Also to be noted are some useful observations about the role of models in the interpretation of sea level data. The authors warn:

. . . when predictions are compared to field data, discrepancies usually exist and it is difficult to ascertain whether they are due to non isostatic causes (tectonics, ocean dynamics changes, etc.) or to insufficiently accurate assumptions used by the models (Pirazzoli and Pluet 1991:34).

Using an augmented version of the same database, Grossman et al. (1998) reach similar conclusions. The need for a regional approach even to so basic a concept as mean sea level is apparent from one of their findings that because of gravitational anomalies relative to the earth's centre, sea level can vary by up to 180 m from place to place, with the result that "mean sea level (msl) in the world’s oceans more closely resembles the pitted surface of a golf ball rather than a smooth sphere." Inclusive of this effect:

The dominant forces that influence Holocene sea-level movements are: (1) climatic and oceanographic vari­ation; (2) glacio- and hydroiostatic redistribution of Earth’s mass in response to ice-sheet advance and re­treat; (3) paleogravitational variation in the geoid; and (4) changes in the morphology of ocean basins and margins due to tectonism.

Only the second of these is relevant to eustatic change.

Additional to the above, examination of the database led the authors to the fairly pessimistic conclusion that:

Errors including those from measurements of elevation, age, tides, tectonics, wind and wave set-up, storm deposition, and environ­mental interpretation, make the uncertainty of the paleosea-level estimates comparable in magnitude to the marginal changes we want to understand.

Finally, taking account of variables such as those alluded to by the above authors, Dickinson 2003 is an attempt to reconcile both field data and theoretical predictions of sea levels with radiocarbon dates not only of emergent reefs but also of island occupations as recorded by archaeologists. Adjustments were made both to the occupation dates and to dating from reef samples to take account of differing proportions of radiocarbon in sea water compared with terrestrial carbon. Other adjustments to data were also made, including elimination from the sample of materials judged uncertain or unacceptable for other reasons (Dickinson 2003:490).


The Micronesian connection


As elsewhere in remote Oceania, the probable direction of settlement in Micronesia is from west to east. The oldest known settlement dates are in the Western Micronesian areas of the Mariana Islands, Yap, and Palau at about

the same time as Lapita settlement in the Bismarcks, but these dates are pushed back more than a millennium earlier than Lapita by paleoenvironmental evidence suggesting colonisation of the Marianas by 4800 BP, and Palau by 4500 BP (Wickler 2004:29). It will be noted that this is considerably before sea levels are thought to have begun their decline to modern levels.  


Mariana Islands

There are two reasons for highstand not to have operated as a barrier to settlement in the Marianas. First, Guam and Rota are high islands which would have been ready for occupation at any time, and, second, they were subject to tectonic uplift, cancelling out any sea rise there might have been.

At Rota, Pirazzoli and Pluet (1991:143) report a 3-4.5m highstand at 4000-3000 BP and a lesser one at 2500 BP. " In Guam, a standstill at +1.8 m lasted from 5100 to 2900 yr BP." Grossman et al. (1998) report 1.8m between 6000 and 4200 BP but attributed entirely to tectonic uplift. Dickinson (2003:493) places the end of highstand at 3200 BP and crossover at 2100 BP, with human occupation occurring perhaps 300 yrs after sea level began to fall, long before the so-called crossover date which is irrelevant in the Marianas because of its combination of high island status and presence of uplift during the critical period that preceded occupation.


Yap and Palau

Comparative information is not available for Yap and Palau, but they occupy the same island chain as the Marianas and their history may not have been much different.



As a homeland habitat for future excursions into Eastern Micronesia, the central Caroline Islands are a mix of high and low islands. Primary among them is the high island of Ponape (now Pohnpei) which would almost certainly have been the first to be occupied. Another high island of significance is Kusaie (now Kosrae), mentioned earlier, the furthest east island of the group which would have been a convenient springboard into Eastern Micronesia when settlement of this area began. Also to be noted is presence of atolls in the group, which would have ensured thorough familiarity with atoll environments as soon as these became available for exploitation. 



This is the garden island which has emerged in every chapter of the present book as the most likely starting point for migrations into Eastern Micronesia.

It is mountainous and fertile, with plentiful water and abundant natural resources. Pirazzoli and Pluet (1991:143) report a sea level rise of about 0.8m at 5500 BP but provide no further detail. Grossman et al. (1998) were able to find only inconclusive sea level evidence for this area, and Dickinson (2003) makes no mention of it at all. As a high island, however, crossover points have no relevance for Ponape and, like other such islands, it could have been occupied at any time.



East of the Carolines lie the Eastern Micronesian atoll chains of the Marshall Islands and Kiribati (formerly Gilbert Islands), together with Tuvalu (formerly Ellice Islands) in Western Polynesia, though long associated under colonial administration with Kiribati as the Gilbert and Ellice Islands Crown Colony, leading to much intermingling of cultural traits between the two. But this is not the only connection.

In this area, referred to generally in sea level literature as the central equatorial Pacific, Grossman et al. (1998) calculate a middle to late Holocene sea level highstand peaking at 1 to 2 m above present between 5000 and 1500 BP.

By contrast, for the Marshalls, Kiribati, and Tuvalu collectively, Dickinson (2003:494 Table 1) assesses the end of highstand at 2200 BP, and crossover not until AD 1100. Later in the paper, however (Dickinson 2003:498), he sites occupation dates for the Marshall Islands of AD 100-400, a full 1000 years earlier than his newly calculated crossover date that, by definition, is meant to mark the earliest date by which occupation would be possible. Dickinson does not acknowledge this as a conflict but concedes: 

Evidence for human occupation well before the crossover date implies that habitable islets had begun to grow on atoll reefs while mid-Holocene paleoreef flats were still intertidal.

In explanation, Dickinson attributes this anomaly to higher tides in the central Pacific than elsewhere which would favour greater deposition of reef materials.

Important to note is that both this example of pre-crossover settlement and Dickinson's explanation of it is particular to the very area that would have been a path for Micronesians if they contributed to the settlement of Polynesia, blowing a hole in the entire concept of crossover as a necessary precondition for settlement, which anthropologists have believed for years, primarily on the authority of Dickinson.

Taking all of the information in the present chapter into account, it is now clear that global estimates of past sea levels cannot be trusted, and local conditions frequently over-ride any such trends. Dickinson's revelation that higher tides in the region of Kiribati and Tuvalu favour increased deposition of reef materials is one such case, and has been confirmed by some remarkable recent research which can readily be extrapolated into the past.

The commonly held assumption of atoll erosion during periods of sea rise has been overturned by the findings of a paper published in 2010 by Webb and Kench. These authors examined aerial photographs and satellite images of central Pacific islands taken over the last 20 to 60 years when sea levels have been rising at a rate of about 2 mm a year. Contrary to the idea that these islands are slowly disappearing under the sea, the results revealed that 43 percent of the islands actually increased in land area, and only 14 percent underwent reduction. 

Other studies say little of what happened to atolls during the long period of sea level rise after about 4000 BP. In light of the Webb and Kench results, however, it is not unreasonable to suppose that this too may have been a time of island building. Erosion by wind and tide of newly formed coral would be replaced by further coral growth, and the debris would be washed up somewhere else, adding to the volume of land and creating areas that would be exposed as extensive tidal flats as soon as sea rise ceased. Floating coconuts would sprout and root, and other vegetation would follow, and it would seem probable that such an area would become suitable as a way-station, if not for permanent occupation, well within the following 2000 years.


>>> Chapter 15. Discussion and conclusions

[5] Considered at length by Nunn 2008.   


[6] CRE 3, together with the previous two CRE events is deduced from changes in Atlantic corals of the West Indies but is evidently deemed to have affected the Pacific equally.