Paths and timings of the peopling of Polynesia inferred from genomic networks

May combine two claims.

Ancient DNA studies of Polynesia have sequenced only four samples from one western Polynesian island and three near-modern samples from one eastern Polynesian island, with low genotype density and limited overlap.

“Meanwhile, ancient DNA studies have sequenced only four samples from one island in western Polynesia and three near-modern samples from one island in eastern Polynesia, all with low genotype density, still lower between-sample genotype overlap, and different time frames17,18.”

• cites [17] Skoglund, P. et al. Genomic insights into the peopling of the Southwest Pacific. Nature 538, 510–513 (2016). sourceDisputed

  The paper’s ancient-DNA passage (section labeled Abstract; page unavailable) explicitly places its four sequenced individuals on two islands: three on Efate, Vanuatu, and one on Tongatapu, Tonga—not four on one western Polynesian island. It reports 139,461–231,944 covered SNPs per individual, but does not report three near-modern eastern Polynesian samples or the claimed between-sample genotype overlap. (LLM, conf=0.88)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured
• cites [18] Posth, C. et al. Language continuity despite population replacement in Remote Oceania. Nat. Ecol. Evol. 2, 731–740 (2018). sourcePartly supported

  The Results (‘Different genetic trajectory in Polynesia’; page unavailable) report three genome-wide samples from Ra’iātea, eastern Polynesia, dated to the 18th century. The paper also reports three Tongan samples, but does not establish the claimed total of four western Polynesian samples from one island; nor does it state that all samples had low genotype density or quantify between-sample overlap. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Standard errors for all allele-frequency-based statistics were computed using a block bootstrap with 100 replicates and a block size of 1000 markers.

“Standard errors for all allele frequency based statistics were computed using the block-bootstrap using 100 replicates and a block size of 1000 markers86.”

• cites [86] Patterson, N. et al. Reconstructing Indian population history. Nature 461, 489–494 (2009). sourceDisputed

  The Methods Summary (page not available in our copy) says standard errors were computed using a Block Jackknife, not a block bootstrap. It does not give the claimed 100 replicates or 1,000-marker block size. (LLM, conf=0.89)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured

May combine two claims.

Previous region-wide Polynesian genetics studies were limited to globin gene polymorphisms, near Polynesia, or the Society Islands, and lacked an ancestry-specific approach.

“Previous region-wide Polynesian genetics studies have considered only globin gene polymorphisms14, or have been restricted to near (western) Polynesia15 and the Society Islands16 and lacked an ancestry-specific approach.”

• cites [14] Hill, A. V. S. et al. Polynesian origins and affinities: globin gene variants in eastern Polynesia. Am. J. Hum. Genet. 40, 453–463 (1987).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [15] Wollstein, A. et al. Demographic history of Oceania inferred from genome-wide data. Curr. Biol. 20, 1983–1992 (2010). sourceNot checked — no full text of the source
• cites [16] Hudjashov, G. et al. Investigating the origins of eastern Polynesians using genome-wide data from the Leeward Society Isles. Sci. Rep. 8, 1823 (2018). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Geological analyses of Polynesian trade goods, especially adzes, indicate that remote Polynesian islands remained in trade contact for several centuries.

“Geological analyses of Polynesian trade goods, particularly adzes, indicate that the remote Polynesian islands remained in trade contact with one another for several centuries25,28,29.”

• cites [25] Kirch, P. V. et al. Human ecodynamics in the Mangareva Islands: a stratified sequence from Nenega-Iti Rock Shelter (site AGA-3, Agakauitai Island). Archaeol. Oceania 50, 23–42 (2015). sourceNot checked — no full text of the source
• cites [28] Weisler, M. I. et al. Cook Island artifact geochemistry demonstrates spatial and temporal extent of pre-European interarchipelago voyaging in East Polynesia. Proc. Natl Acad. Sci. USA 113, 8150–8155 (2016). sourceSupported
• cites [29] Collerson, K. D. & Weisler, M. I. Stone adze compositions and the extent of ancient Polynesian voyaging and trade. Science 317, 1907–1911 (2007). sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Polynesians are predominantly descended from Austronesian-speaking voyagers.

“The Polynesians are predominantly descended from Austronesian-speaking voyagers17”

• cites [17] Skoglund, P. et al. Genomic insights into the peopling of the Southwest Pacific. Nature 538, 510–513 (2016). sourcePartly supported

  The paper’s ancestry-model passage (labelled Abstract in our copy; no page given) identifies a population related to the First Remote Oceanians as the primary component in present-day Polynesians, and describes seafaring settlement of Remote Oceania. It does not establish that those ancestral voyagers spoke Austronesian languages; the Extended Data table labels some present-day sampled populations Austronesian-speaking, but not the ancient voyagers. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

May combine two claims.

The inferred settlement date for Rapa Nui is consistent with Wilmshurst and Mulrooney and agrees closely with Hunt et al.'s 1200 CE date and recent radiocarbon dates.

“Our date for the settlement of Rapa Nui is consistent with Wilmshurst and Mulrooney and also agrees closely with the date found by Hunt et al. (1200 CE) based on analyses of pollen in lake cores and soil erosion patterns42, as well as with recent radiocarbon dates of archaeological sites43.”

• cites [42] Anderson, A. J., Conte, E., Smith, I. & Szabo, K. New excavations at Fa’ahia (Huahine, Society Islands) and chronologies of central East Polynesian colonization. J. Pac. Arch. 10, 1–14 (2019). sourceCould not verify

  The Radiocarbon Dating Results from Fa’ahia section describes dates from Huahine, while General points discusses colonization estimates for other East Polynesian island groups. Searches for Rapa Nui or Easter Island settlement, Wilmshurst, Mulrooney, and Hunt’s 1200 CE estimate found no passage assessing the claimed agreement with Rapa Nui archaeological dates. (LLM, conf=0.74)

  Experimental — accuracy not yet measured
• cites [43] Hunt, T. L. & Lipo, C. P. Evidence for a shorter chronology on Rapa Nui (Easter Island). J. Island Coast. Archaeol. (2008). sourceNot checked — no full text of the source

Classified as probably not a substantial claim. May combine two claims.

Archaeological estimates for Polynesian settlement dates are debated and have recently been revised forward across eastern Polynesia by up to a millennium.

“Meanwhile, the archaeological estimates for settlement dates remain debated, and have recently been revised forward across eastern Polynesia by up to a millennium2–4,12,13.”

• cites [2] Kirch, P. V. On the Road of the Winds (Univ. of California Press, 2017). sourceNot checked — no full text of the source
• cites [3] Mulrooney, M. A., Bickler, S. H., Allen, M. S. & Ladefoged, T. N. High-precision dating of colonization and settlement in East Polynesia. Proc. Natl Acad. Sci. USA 108, E192–E194 (2011). sourceNot checked — no full text of the source
• cites [4] Schmid, M. M. E. et al. How 14C dates on wood charcoal increase precision when dating colonization: the examples of Iceland and Polynesia. Quat. Geochronol. 48, 64–71 (2018). sourceNot checked — no full text of the source
• cites [12] Wilmshurst, J. M., Hunt, T. L., Lipo, C. P. & Anderson, A. J. High-precision radiocarbon dating shows recent and rapid initial human colonization of East Polynesia. Proc. Natl Acad. Sci. USA 108, 1815–1820 (2011). sourcePartly supported

  The introduction says East Polynesian settlement dates remain debated, with long and short chronologies differing by 400–1,000 years. The Results and Discussion report a revised, later colonization chronology, but say it shortens the estimated age by up to 2,000 years relative to some earlier claims. Thus the stated ceiling of a millennium does not capture the full range reported by this source. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [13] Spriggs, M. & Anderson, A. Late colonization of east Polynesia. Antiquity 67, 200–217 (1993). sourceNot checked — no full text of the source

May combine two claims.

The matrix completion method used permits all samples, not only a reference panel, to be used for completion, which allows more accurate completion.

“Unlike earlier methods69, this method permits the use of all samples, rather than only a panel of reference samples, for the completion step; thus, far more data is used allowing for more accurate completion.”

• cites [32] Wang, C. et al. Comparing spatial maps of human population-genetic variation using procrustes analysis. Stat. Appl. Genet. Mol. Biol. 9, 13 (2010). sourceCould not verify

  The Introduction describes Procrustes analysis for comparing spatial maps, including plots whose sample sets only partially overlap. Searches for matrix completion, reference panels, missing entries, and imputation found no passage about using all samples in a completion step or improving completion accuracy. Page numbers were not provided in our copy. (LLM, conf=0.75)

  Experimental — accuracy not yet measured
• cites [68] Moreno-Estrada, A. et al. The genetics of Mexico recapitulates Native American substructure and affects biomedical traits. Science 344, 1280–1285 (2014). sourceCould not verify

  The passage on reference panels (section labeled Abstract; page unavailable) concerns haplotype sharing in Mexican samples, not a matrix-completion method. Searches for matrix completion, the completion step, and missing-data imputation found no passage saying that all samples can be used for completion or that doing so improves accuracy. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

May combine two claims.

Native American ancestry entered Polynesia particularly through emigration of admixed Hispanic individuals from Chile, which annexed Rapa Nui, and African ancestry also entered.

“Native American ancestry, particularly from emigration of admixed Hispanic individuals from Chile, which annexed Rapa Nui (Easter Island), and African ancestry also entered32.”

• cites [32] Wang, C. et al. Comparing spatial maps of human population-genetic variation using procrustes analysis. Stat. Appl. Genet. Mol. Biol. 9, 13 (2010). sourceCould not verify

  The Introduction discusses comparisons of genetic-variation maps and CNV datasets, but does not address ancestry entering Polynesia. Searches for Rapa Nui, Chilean migration, Native American ancestry, and African admixture found no passage supporting the claim in our copy. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

The block bootstrap gives better variance estimates than the jackknife for these pairwise allele frequency comparisons.

“This gives better variance estimates than the jackknife for these pairwise allele frequency comparisons34.”

• cites [34] Novembre, J. et al. Genes mirror geography within Europe. Nature 456, 98–101 (2008). sourceCould not verify

  The Methods Summary describes leave-one-out cross-validation, and the Methods describe SNP filtering and association simulations. Neither those passages nor searches for “block bootstrap,” “jackknife,” and “variance estimates” address a comparison of the two methods for pairwise allele-frequency comparisons. (LLM, conf=0.75)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Requiring a minor allele frequency of 0.1 or lower in the merged Tonga, Samoa, and Fiji outgroup yielded 228,262 SNPs, whereas requiring minor alleles to be fixed in the Taiwan outgroup yielded 137,383 SNPs.

“(A 0.1 or lower minor allele frequency was required in the merged Tonga, Samoa, Fiji outgroup samples, yielding 228,262 SNPs, as opposed to the more stringent requirement of minor alleles being fixed in the Taiwan outgroup samples, following the procedure of Zhan et. al37, which yielded only 137,383 SNPs.)”

• cites [37] Peter, B. M. & Slatkin, M. Detecting range expansions from genetic data. Evolution 67, 3274–3289 (2013). sourceCould not verify

  The Results > Application section describes an analysis of human populations using several hundred thousand loci, but gives neither of the claimed SNP counts. Searches for the Tonga–Samoa–Fiji and Taiwan outgroups, the two counts, and the stated allele-frequency criteria found no passage addressing those filters. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

May combine two claims.

European ancestries entered Polynesia during the colonial period, beginning with Magellan in the 16th century and with significant immigration starting in the early 19th century.

“European ancestries entered Polynesia during the colonial period with the first European explorer (Magellan) arriving in the 16th century and significant immigration commencing in the early 19th century64.”

• cites [64] O’Connell, J. et al. A general approach for haplotype phasing across the full spectrum of relatedness. PLoS Genet. 10, e1004234 (2014). sourceCould not verify

  The Materials and Methods section (‘Real datasets’; page unavailable) describes cohorts in Europe and Uganda, while the other retrieved passages concern haplotype phasing or references. Searches for Polynesia, Magellan, colonial-era European ancestry, and 19th-century immigration found no passage in our copy addressing the claim. (LLM, conf=0.77)

  Experimental — accuracy not yet measured

May combine two claims.

The Tuamotu Archipelago served as a regional voyaging hub from which settlers moved north to the Marquesas and south to the Gambier Islands beginning in the mid-1100s CE.

“Stretching across central eastern Polynesia, the Tuamotu Archipelago was previously hypothesized to have served as a regional voyaging hub20,25,28, and our analysis indicates that it was from this hub that settlers made their way north to the Marquesas Islands (Nuku Hiva and Fatu Hiva in our dataset) and south to the Gambier Islands (Mangareva in our dataset) beginning in the mid-1100s CE.”

• cites [20] Emory, K. P. The Tuamotuan creation charts by Paiore. J. Polynesian Soc. 48, 1–29 (1939).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [25] Kirch, P. V. et al. Human ecodynamics in the Mangareva Islands: a stratified sequence from Nenega-Iti Rock Shelter (site AGA-3, Agakauitai Island). Archaeol. Oceania 50, 23–42 (2015). sourceNot checked — no full text of the source
• cites [28] Weisler, M. I. et al. Cook Island artifact geochemistry demonstrates spatial and temporal extent of pre-European interarchipelago voyaging in East Polynesia. Proc. Natl Acad. Sci. USA 113, 8150–8155 (2016). sourceSupported

May combine two claims.

Pre-colonial Native American genetic contact is identified only on islands inferred to have been settled via the Tuamotus, and its timing corresponds closely with voyaging dates for the region.

“It is also notable that it is only on islands that we infer were settled via the Tuamotus that pre-colonial Native American genetic contact has been identified, and its timing corresponds closely with our voyaging dates for this region32.”

• cites [32] Wang, C. et al. Comparing spatial maps of human population-genetic variation using procrustes analysis. Stat. Appl. Genet. Mol. Biol. 9, 13 (2010). sourceCould not verify

  The Introduction compares worldwide genetic and geographic maps using Procrustes analysis; it does not discuss settlement through the Tuamotus. Searches for Native American genetic contact, Polynesian islands, and voyaging dates found no passage addressing the claimed locations or timing. (LLM, conf=0.78)

  Experimental — accuracy not yet measured

May combine two claims.

Admixture between Papuan and Austronesian source populations occurred on Vanuatu and other Melanesian islands in the thousand years before Polynesian settlement and was carried into Polynesia during settlement.

“(Admixture between populations stemming from these two sources occurred on Vanuatu and other Melanesian islands in the thousand years before the settlement of Polynesia and was carried into Polynesia during its settlement17,79.)”

• cites [17] Skoglund, P. et al. Genomic insights into the peopling of the Southwest Pacific. Nature 538, 510–513 (2016). sourcePartly supported

  The paper’s Abstract (page unavailable) dates Papuan–First Remote Oceanian mixture to roughly 1,500–2,300 years before present and finds Papuan ancestry in present-day Polynesians. It does not locate that mixture on Vanuatu or other Melanesian islands, or establish that an already admixed population carried it into Polynesia during settlement. Instead, the ancient Vanuatu and Tongan individuals had little or no Papuan ancestry, and the paper infers later movements into Remote Oceania. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [79] Wittek, P., Gao, S. C., Lim, I. S. & Zhao, L. Somoclu: an efficient parallel library for self-organizing maps. J. Stat. Softw. 78, https://doi.org/10.18637/jss.v078.i09 (2017). sourceCould not verify

  The Introduction (pp. 1–2) discusses self-organizing maps, not population history. Searches for admixture on Vanuatu and other Melanesian islands, and for Papuan–Austronesian ancestry and Polynesian settlement, found no passage in our copy addressing the claim. (LLM, conf=0.78)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

The model of IBD segment independence holds only when population size N is much greater than the number of generations T.

“In addition, the model of IBD segment independence holds only for (N >> T) with N the population size and T the number of generations97.”

• cites [97] Botigué, L. R. et al. Gene flow from North Africa contributes to differential human genetic diversity in southern Europe. Proc. Natl Acad. Sci. USA 110, 11791–11796 (2013). sourceCould not verify

  The Results section discusses detecting and comparing IBD segments, and Materials and Methods describes a sample-size correction. Neither states when IBD segments can be treated as independent; searches for segment independence and the N >> T condition found no supporting passage. Page numbers were not available in our copy. (LLM, conf=0.73)

  Experimental — accuracy not yet measured

May combine two claims.

For IBD segments shared by two present-day individuals from a common ancestor, the decay rate λ is 2T, or T/50 when lengths are measured in centimorgans.

“If we are considering recombination segments shared between two present day individuals stemming from the same common ancestor, that is IBD segments, we must adjust the rate for the number of recombination events per unit length that have occurred down both sides of the pedigree from this common ancestor, which gives a λ of 2T total92,95,98.”

• cites [92] Ceballos, F. C., Joshi, P. K., Clark, D. W., Ramsay, M. & Wilson, J. F. Runs of homozygosity: windows into population history and trait architecture. Nat. Rev. Genet. 19, 220–234 (2018). sourceNot checked — no full text of the source
• cites [95] Gusev, A. et al. Whole population, genome-wide mapping of hidden relatedness. Genome Res. 19, 318–326 (2009). sourceSupported
• cites [98] Atzmon, G. et al. Abraham’s children in the genome era: major Jewish diaspora populations comprise distinct genetic clusters with shared Middle Eastern ancestry. Am. J. Hum. Genet. 86, 850–859 (2010). sourceSupported

May combine two claims.

The recursive method for constructing the branching settlement path is an instance of the Chu-Liu-Edmonds algorithm, which guarantees a minimum spanning tree for a directed acyclic graph.

“This recursive algorithm for building the branching settlement path of the Pacific islands is a form of the Chu-Liu-Edmonds algorithm, which is guaranteed to produce the minimum spanning tree of a directed acyclic graph (see Supplementary Discussion ‘On tree-building’)87,88.”

• cites [87] Bhatia, G., Patterson, N., Sankararaman, S. & Price, A. L. Estimating and interpreting FST: the impact of rare variants. Genome Res. 23, 1514–1521 (2013). sourceCould not verify

  The Methods section on Nei’s FST and the Results sections on SNP ascertainment and FST estimators discuss population-genetic estimates, not settlement-path construction. Searches for the Chu-Liu-Edmonds algorithm, Pacific Islands tree-building, and directed graphs found no passage in our copy addressing the claim. (LLM, conf=0.78)

  Experimental — accuracy not yet measured
• cites [88] Nei, M. & Roychoudhury, A. K. Sampling variances of heterozygosity and genetic distance. Genetics 76, 379–390 (1974). sourceNot checked — no full text of the source

May combine two claims.

The inferred dates differ from archaeological dates only for the Marquesas Island group and the Southern Cook Islands, where they are earlier.

“Only in the settlement of the Marquesas Island group, dated by Mulrooney to the late 1100s CE, and the Southern Cook Islands, dated even later by Wilmshurst to the mid-1200s CE, do we find different (earlier) dates.”

May combine two claims.

Originating islands have much greater genetic diversity and dominate the first principal component when included in PCA.

“Because the originating islands have much greater diversity (as discussed below), they dominate the first principal component when included in the PCA (Supplementary Fig. 3).”

May combine two claims.

Polynesian population structure exhibits high dimensionality and is not reflective of geography.

“we find that Polynesian population structure exhibits high dimensionality (Supplementary Fig. 1) not at all reflective of geography (Fig. 1a)”

May combine two claims.

Rarotonga and Palliser appear at the center of eastern Polynesian islands, with other eastern islands radiating out from them.

“Rarotonga and Palliser appear at the center of the eastern Polynesian islands with the other eastern islands radiating out from them.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Projections onto fewer dimensions gave similar results, with some clusters beginning to disappear between 12 and 15 dimensions, as predicted by the Tracy-Widom analysis.

“(Projections onto fewer dimensions yielded similar results, with some clusters beginning to disappear in the range 12–15 dimensions, as predicted by the Tracy-Widom analysis.)”

Classified as probably not a substantial claim.

Filtering sites not located in Polynesian segments for any individual removed 60,377 SNPs, or 8.75% of 689,899 SNPs, leaving 629,522 SNPs for population allele frequency statistics.

“This filtering resulted in the loss of 60,377 SNPs (8.75% of the total 689,899 SNPs), leaving 629,522 SNPs across all populations for computation of population allele frequency statistics.”

May combine two claims.

Using Tonga, Samoa, and Fiji as an outgroup reduced standard errors for the range expansion statistic on eastern Polynesian islands, while the general ordering of islands was the same as with the Taiwan outgroup.

“The latter calculation reduced the standard errors for the range expansion statistic on islands settled subsequent to western Polynesia, i.e. the eastern Polynesian islands; nevertheless, the general ordering of islands in the range expansion was the same for both calculations (see comparison Supplementary Fig. 14).”

May combine two claims.

Modern Pacific islanders are often admixed, carrying European ancestry and occasionally Native American and African ancestries.

“As discussed above, modern Pacific islanders are often admixed, possessing European and occasionally Native American and African ancestries (Supplementary Fig. 32).”

May combine two claims.

The inferred settlement dates for Fatu Hiva (1140 CE) and the Tuamotus (1110 CE) agree with genealogical oral histories.

“our settlement dates (1140 CE on Fatu Hiva in the Marquesas, or 28.4 generations before 1989) agree with the genealogical oral histories of many Pacific islanders themselves (1005 CE, or 29 generations before 1875, on Fatu Hiva) 27. In the Tuamotus our dates (1110 CE, or 29.3 generations before 1989) agree even more closely with island’s oral histories (1125 CE, or 28 generations before 1965)44.”

• cites [27] Handy, E. S. C. The Native Culture in the Marquesas (The Bishop Museum, 1923).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [44] Mulrooney, M. A. An island-wide assessment of the chronology of settlement and land use on Rapa Nui (Easter Island) based on radiocarbon data. J. Archaeol. Sci. 40, 4377–4399 (2013). sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

A few Pacific island individuals, especially in the Marquesas, had more than 5% East Asian ancestry in the ADMIXTURE results.

“A few Pacific island individuals, particularly in the Marquesas, were found to also have >5% East Asian ancestry in the ADMIXTURE results.”

May combine two claims.

F4 tests found deviations from the settlement tree only for three branches: Mangareva-Raivavae with migration from Tahiti, Mangareva-Palliser with migration from Tahiti, and North Marquesas-Palliser with migration from Tahiti and also from the Cooks.

“We find deviations from our settlement tree only for 3 of its branches: Mangareva – Raivavae (migration from Tahiti), Mangareva – Palliser (migration from Tahiti), and North Marquesas – Palliser (migration from Tahiti and also from the Cooks).”

May combine two claims.

Cross-validation error showed an elbow at K = 7 ancestry clusters, with larger numbers of clusters giving little improvement.

“An elbow61 was found in the cross-validation error plot at K = 7 clusters, with larger numbers of clusters delivering little improvement (Supplementary Fig. 32).”

May combine two claims.

The terminal islands in the settlement sequence have high levels of homozygosity and significant IBD.

“We have shown that these particular islands also have high levels of homozygosity (Supplementary Figs. 25–27), which should increase the power to detect trait associations, and significant IBD, enabling IBD mapping, another useful approach51.”

• cites [51] Buck Te Rangi Hīroa, P. H. Vikings of the Sunrise (J. B. Lippincott, 1938).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

The genotype frequency matrix had 14 significant dimensions, determined using a Tracy-Widom distribution and verified with a scree plot.

“The number of significant (p > 0.05) dimensions for the genotype frequency matrix, described above, was determined (n = 14) using a Tracy-Widom distribution57 and verified via a scree plot70.”

May combine two claims.

The estimated λ values confirm the drift statistics: Mangareva is the most recently connected island to Rapa Nui, and Samoa is the most archaic connection.

“The pattern confirms the results of our drift statistics; Mangareva is the island most recently connected to Rapa Nui, while Samoa, the root of the expansion into remote Polynesia, is the most archaic connection.”

May combine two claims.

IBD sharing tends to increase along the inferred settlement chain, but the signal contains substantial noise.

“Trends of increasing IBD sharing along the course of the inferred settlement chain (see the map in Extended Data Fig. 5) are evident, but there is significant noise.”

May combine two claims.

Removing outlier inter-island IBD segments does not significantly change the estimated island settlement dates but increases the standard error.

“Not removing these outliers does not change our island settlement dates significantly, but, by distorting the tail of the exponential decay distributions, does increase our standard error”

May combine two claims.

Rapa Nui has the largest ψ value, indicating it has the most potential parent islands and is terminal in the range expansion.

“Begin with the island with the most potential parents (at the end of the range expansion) (Fig. 2B) or in other words the largest ψ, Rapa Nui.”

May combine two claims.

Removing λ entries with standard errors above 0.07 yields a more precise generation matrix but leaves some missing entries.

“Removing all entries corresponding to λ values that have standard errors above .07 (representing errors larger than 15% of the average λ value), creates a matrix of more precise generation values, but with some missing entries.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Using the outgroup-F3 metric instead of the pairwise difference metric yields the same settlement sequence.

“although the same settlement sequence is also returned when using the latter metric instead (Supplementary Fig. 12)”

May combine two claims.

The expansion finally reached the Marquesas, Raivavae, and Easter Island, with Easter Island settled around 1200 CE via Mangareva.

“finally to the widely separated, but genetically connected, megalithic statue-building cultures of the Marquesas (Te Henua ‘Enana) Islands in the north, Raivavae in the south, and Easter Island (Rapa Nui), the easternmost of the Polynesian islands, settled in approximately 1200 CE via Mangareva.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The Austronesian-speaking settlers of Fiji, Tonga, and Samoa populated the widely dispersed islands to their east through long voyages.

“The Austronesian-speaking settlers of the western Pacific (Fiji, Tonga, and Samoa) went on to people the widely dispersed islands in the vast ocean to their east through extraordinary voyages of exploration and settlement2,20.”

• cites [2] Kirch, P. V. On the Road of the Winds (Univ. of California Press, 2017). sourceNot checked — no full text of the source
• cites [20] Emory, K. P. The Tuamotuan creation charts by Paiore. J. Polynesian Soc. 48, 1–29 (1939).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

Austronesian linguistic origins trace to Taiwan.

“who trace their linguistic origins to Taiwan9”

• cites [9] Gray, R. D., Drummond, A. J. & Greenhill, S. J. Language phylogenies reveal expansion pulses and pauses in Pacific settlement. Science 323, 479–483 (2009). sourceNot checked — no full text of the source

May combine two claims.

Recent inter-island movement can introduce longer IBD segments and shift divergence time estimates toward the present, so a truncated exponential was fitted.

“Recent movement between islands, such as post-settlement trade contact, can introduce small numbers of longer, inter-island IBD segments, shifting the estimated divergence time towards the present, so we fit a truncated exponential.”

May combine two claims.

Each successive principal component captures genetic drift of a particular island or island group, indicating that genetic variance between islands is dominated by founder effects rather than diffusion clines or migration gradients.

“Instead, each successive principal component captures the genetic drift of a particular island or island group (Fig. 1b, c and Supplementary Fig. 2), illustrating that genetic variance between these islands is dominated by their founder effects, not by diffusion clines or migration gradients.”

May combine two claims.

The dataset contains both terminal and internal nodes of the population tree, and their hierarchy is known from the ψ-statistic.

“When constructing a population tree, this means that our dataset contains not only the terminal (leaf) nodes, but also the internal nodes, and we know their hierarchy from the ψ-statistic.”

Classified as probably not a substantial claim. May combine two claims.

Quality control filters removed individuals with more than 1% missing genotypes, sites missing in more than 1% of individuals, and 18 SNPs with extreme Hardy-Weinberg equilibrium deviations.

“Quality control filters were applied across all sampled individuals using the Plink 1.9 package53, removing individuals with >1% of genotyped sites missing (mind .01), removing genotyped sites missing in >1% of individuals (geno .01), and removing sites (18 SNPs) with extreme deviations from Hardy-Weinberg equilibrium (p-value less than 10e-110).”

Classified as probably not a substantial claim.

Individuals with a kinship coefficient greater than 0.2, indicating first-degree relatives, were iteratively removed.

“REAP55 was used to determine kinship coefficients using the ADMIXTURE clustering results discussed below; individuals with a kinship coefficient of >.2 (first degree relatives) were iteratively removed.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The final leg to Rapa Nui had been suggested by linguistic similarities between Mangarevan and Rapanui and by similarities in traditional stone ceremonial platforms.

“This final leg had been suggested by some based on similarities between the Mangarevan and Rapanui languages48, and by similarities in their traditional stone ceremonial platforms49.”

• cites [48] Dickinson, W. Pacific atoll living: how long already and until when. Geol. Soc. Am. Today 19, 4–10 (2009). sourceNot checked — no full text of the source
• cites [49] Fischer, S. R. Mangarevan doublets: preliminary evidence for proto-southeastern Polynesian. Ocean. Linguist. 40, 112–124 (2001). sourceNot checked — no full text of the source

May combine two claims.

Settlers fanned out from Rarotonga northeast to the Society Islands around 1050 CE, and thence to the Tuamotu Archipelago by 1110 CE.

“Settlers also fanned out from Rarotonga northeast to the Society Islands (represented by Tahiti in our dataset but also containing the culturally significant island Raʻiātea) around 1050 CE, thence northeast to the Tuamotu Archipelago (represented by Mataiva in the Palliser group in our dataset) by 1110 CE.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Local approvals for engagement with the Rapa Nui community were obtained from the mayor of Easter Island, and the study was registered with CONADI under indigenous law No. 19.253.

“Local approvals for engagement with the Rapa Nui community were obtained from the mayor (Pedro Pablo Edmunds Paoa) of the municipality of Easter Island, and the study was registered with the National Corporation for Indigenous Development (CONADI), in accordance with the indigenous law No. 19.253.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The UNESCO International Declaration on Human Genetic Data and the Declaration of Helsinki were followed throughout the study.

“The guidelines of the UNESCO International Declaration on Human Genetic Data and the Declaration of Helsinki were followed throughout the study.”

May combine two claims.

The reconstruction of the settlement path is independent of the date estimates and is more robust to later sporadic contact than IBD distributions.

“Note that our reconstruction of the settlement path is independent of these date estimates, which are overlaid upon it, and is more robust to later sporadic contact than IBD distributions are”

Probably not a substantial claim: may not pass S3 (load-bearing).

No statistical methods were used to predetermine sample size.

“No statistical methods were used to predetermine sample size.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The study was approved by the Council of Polynesian Elders for the Rapa Nui community and by local educational institutions.

“This study was also approved by the Council of Polynesian Elders for the community of Rapa Nui, along with local educational institutions, including the Lyceum Hoŋa’a o te Mana and the Lorenzo Baeza school for adults.”

May combine two claims.

The Tuamotu hub and other critical atolls had only recently emerged above falling sea levels (900 CE) and finished solidifying topsoil and forests.

“At this time the widely scattered Tuamotu hub and other critical atolls in the expansion path (e.g. Nororotu in the Austral group), would have only recently emerged above falling sea levels (900 CE) and finished solidifying their topsoil and forests44,47”

• cites [44] Mulrooney, M. A. An island-wide assessment of the chronology of settlement and land use on Rapa Nui (Easter Island) based on radiocarbon data. J. Archaeol. Sci. 40, 4377–4399 (2013). sourceNot checked — no full text of the source
• cites [47] Walworth, M. The Language of Rapa Iti (Univ. Hawaii, 2015).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

From Rarotonga, settlers continued south around 1190 CE to Rapa Iti and separately east to the smaller Cook Islands (Mauke and Atiu).

“From this base, we find that settlers continued south around 1190 CE to Rapa Iti (a branch recently hypothesized from linguistic evidence46) and, separately, east to the smaller Cook Islands (Mauke and Atiu in our dataset).”

• cites [46] Di Piazza, A., Di Piazza, P. & Pearthree, E. Sailing virtual canoes across Oceania: revisiting island accessibility. J. Archaeol. Sci. 34, 1219–1225 (2007). sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Rarotonga's clouds made it visible from long distances and probably facilitated its discovery.

“These clouds on its prominent mountain make the island visible for long distances at sea and probably facilitated its discovery45.”

• cites [45] Pirazzoli, P. A. & Montaggioni, L. F. Late Holocene sea-level changes in the northwest Tuamotu islands, French Polynesia. Quat. Res. 25, 350–368 (1986). sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

After merging reference and sample genotype data, unambiguous strand inconsistencies were flipped and ambiguous SNPs were removed, leaving 689,899 SNP sites.

“After merging reference sequence data with sample genotyped data, strand inconsistencies were flipped when unambiguous, while ambiguous SNPs were removed, leaving 689,899 SNP sites.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Research ethics approval and permits were obtained from Stanford University IRB, Oxford University Tropical Research Ethics Committee, and the Scientific Ethics Committee of the Catholic University of Chile.

“research/ethics approval and permits were obtained from the following institutions: Stanford University Institutional Review Board (IRB approval No. 20839), Oxford University Tropical Research Ethics Committee (reference No. 537–14), and the Scientific Ethics Committee of the Catholic University of Chile (reference No. 1971092).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The average genotype call rate was 98.5% for all newly genotyped samples.

“The average call rate was 98.5% for all newly genotyped samples.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Written informed consent was obtained from every participant in the study.

“Written informed consent was obtained from all participants”

Probably not a substantial claim: may not pass S3 (load-bearing).

Community engagement, including pre-participation presentations and post-participation return of results, was conducted throughout the project.

“Community engagement, including pre-participation presentations and post-participation return of results, were conducted throughout the project.”

May combine two claims.

Because allele frequencies drifted during settlement, some alleles minor in Taiwan became major by western Polynesia, so repolarizing with Tonga, Samoa, and Fiji increased resolution and reduced standard errors for downstream islands.

“Because allele frequencies drifted during the Pacific island settlement process, some minor alleles in Taiwan would have become major alleles by the time the settlers reached western Polynesia (see Supplementary Discussion ‘On psi‘), so the intermediate repolarization using Tonga, Samoa, and Fiji as an outgroup increased the resolution of the range expansion statistic (reduced standard errors) for”

May combine two claims.

A genome-wide haplotype vector assembled by concatenating chromosomes switches phase arbitrarily at chromosome boundaries and is therefore already a mixture of an individual's two true parental haplotypes.

“Thus, a genome-wide haplotype vector constructed by assembling all chromosomes sequentially into a single row vector will switch phase arbitrarily across chromosome boundaries and so is already a mixture of an individual’s two true parental haplotypes.”

May combine two claims.

Ancestries introduced through colonial settlement did not necessarily follow the same island settlement process, founder sizes, or dates as original Polynesian settlement, so they must be distinguished from Polynesian ancestry.

“Because ancestries fully (or partially) introduced via colonial settlement did not necessarily follow the same island settlement process (or founder sizes and dates) as the original Polynesian settlement, such ancestries need to be distinguished, necessitating an ancestry-specific approach65, cf. Supplementary Figs. 10, 20.”

• cites [65] D’Arcy, P. The Chinese Pacifics: a brief historical review. J. Pacific Hist. 49, 396–420 (2014). sourceNot checked — no full text of the source

May combine two claims.

Raivavae, North and South Marquesas, and Rapa Nui share a common genetic source in the Tuamotu archipelago and had monumental stone statue traditions.

“Together with even more distant North and South Marquesas and Rapa Nui, each also with inferred settlement stemming from the Tuamotus, Raivavae had an ancient tradition of carving monumental anthropomorphic statues in stone. No other Austral island had these50; indeed, such immense sculptures are found only on those far-flung islands that we now, however, show to had a common genetic source in the”

• cites [50] Flenley, J. & Bahn, P. The Enigmas of Easter Island (Oxford Univ. Press, 2003).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

From western Polynesia, voyagers reached Rarotonga around 830 CE, passing from Samoa along a route shared with Fiji and Tonga.

“From western Polynesia, Polynesian voyagers reached Rarotonga in the Cook Islands around 830 CE, having passed from Samoa along a route shared with the settlement of Fiji and Tonga.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Because drift is independent among separated small island populations, some deviation from Hardy-Weinberg equilibrium is expected in Polynesia, so a typical Hardy-Weinberg threshold was not applied.

“The independence of drift between these separated, small island populations leads us to expect some deviation from Hardy-Weinberg in Polynesia, so we do not apply a typical threshold here.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The experiments were not randomized and investigators were not blinded to allocation during experiments and outcome assessment.

“The experiments were not randomized and the investigators were not blinded to allocation during experiments and outcome assessment.”

May combine two claims.

The dataset has T values of 25-30 generations and N values in the thousands, so it avoids the problematic regimes for the IBD model.

“Fortunately, our dataset has T values of 25–30 generations and N values in the thousands21, so we do not fall into these problematic regimes.”

• cites [21] Hunt, T. & Lipo, C. The Statues that Walked (Free Press, 2011).Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

The Tahiti migrations align with Tahiti's recent role as the capital of French Polynesia.

“consistent with Tahiti’s recent role as the capital of French Polynesia.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

The per-generation recombination rate is constant over time, so the distance from an ancestral population to any sampled contemporary descendant is the same when measured by segment length distributions.

“This works because the per generation recombination rate is constant over time, so the distance from an ancestral population to any of its sampled contemporary descendants is the same when measured by segment length distributions.”

May combine two claims.

The decay rate of the IBD length distribution changes steadily each generation and does not depend on unknown fluctuations in historical population sizes.

“In addition, the shape of the IBD length distribution (decay rate) changes steadily each generation. It does not depend, as genetic drift does, on the fluctuations, which are generally unknown, of the historical population sizes.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The Poisson process model of recombination holds for small IBD segments (T > 5), but for more recent relatedness the finite size of chromosomes must be considered.

“The model of a Poisson process of recombination events along a continuous genome holds for small IBD segment lengths, viz. T > 5, but for more recent relatedness, leading to very long IBD segments, one must consider the finite size of the chromosomes themselves when computing the IBD length distribution101.”

• cites [101] Palamara, P. F., Lencz, T., Darvasi, A. & Pe'er, I. Length distributions of identity by descent reveal fine-scale demographic history. Am. J. Hum. Genet. 91, 809–822 (2012). sourceNot checked — no full text of the source

May combine two claims.

Changing the ancestry normalization factor alters the amplitude but not the decay rate λ of the exponential IBD length distribution, showing that λ is robust to ancestry normalization errors.

“This alters the amplitude, but not the decay rate shape parameter λ, of the exponential in Supplementary Fig. 22A, or, equivalently, the intercept, but not the slope of the lines in Supplementary Fig. 22B; thus demonstrating graphically that λ is robust to noise (errors) in ancestry normalization.”

May combine two claims.

Founding bottlenecks create some intra-island IBD, causing some recombination events to be non-productive.

“However, because of the founding bottlenecks for each island, there is some intra-island IBD shared between pairs of individuals on the island (Supplementary Figs. 25, 26), so some recombination events will be non-productive.”

May combine two claims.

The ultrametric inequality holds because all samples are contemporary and are leaf nodes of the ancestry tree dating to the present.

“This holds in our case because all samples were taken from contemporary populations, and so all are leaf nodes of the ancestry tree dating to the same period (the present).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Markers were bootstrapped together as long contiguous blocks to preserve the effects of linkage on the variance of the estimates.

“The markers are bootstrapped together as long contiguous blocks to preserve the effects of linkage on the variance of the estimates35.”

• cites [35] Nei, M. & Kumar, S. Molecular Evolution and Phylogenetics (Oxford Univ. Press, 2000). sourceNot checked — no full text of the source

May combine two claims.

The migration graph is acyclic, a result derived from the formal definition of ψ.

“That our graph is acyclic can be shown (proof derived in Supplementary Discussion ‘On the acyclicity of psi’) from the formal definition of ψ, which defines our edge directionality.”