Genomic Data Reveal a Complex Making of Humans

As all individuals are genetically different, modern DNA technology has increased the ability to perform human identity testing. Individual genetic identification is important in the determination of perpetrators of violent crimes, resolving uncertain paternity or maternity and identifying remains of missing persons or victims of mass disasters.1 Next to forensic cases, genetic identification also strongly increases the opportunity to identify skeletal remains or other biological samples associated with historical figure.2 In this way, the genetic analysis can give insight in several historical and archaeological research questions. The genetic approach was successful, for example, to check the verity of the death of the legendary outlaw Jesse James3 or Louis XVII, the son of Louis XVI and Marie-Antoinette, who died during the French Revolution.4, 5 It also gave certainty about remains found at the burrial place of the Romanov family members,6, 7 about the biological relationship within the family of Austria's patron Saint Leopold III8 and about the reliability and commercial value of religious relics during the Middle Ages.9, 10

The main drawback of the genetic identification of presumptive remains from historical figures is that the DNA within these samples is often degraded and that DNA contamination can mask the original DNA of the person.11, 12 When the person and his or her close relatives died a long time ago, good quality DNA as reference material is usually not available, whereas this is rarely the case in forensic cases. Nevertheless, validation of the data remains essential to confirm the results of the genetic identification. For many of these studies, due to a lack of samples of relatives, only ‘sub-optimal’ validations can be performed, which can be very valuable but which are insufficient to confirm the identity. Sub-optimal validations can also be used to exclude identification and are feasible by several methods. One method is genetic sexing of the remains as done for the wrongly assumed remains of the Polish renaissance poet Jan Kochanowski.13 Another method is comparing genetic markers for several parts of the remains, because sometimes relics of several persons are mixed; this was found in the remains assumed to belong to the Italian poet Francesco Petrarca.14 Comparing the familial relationship between skeletons based on historical records is a third method that was successfully applied for Prince Branciforte Barresi and his family.15 The last method consists of determining the geographic region of origin of an individual based on the mtDNA and/or Y-chromosome (Y-chr) haplogroup as performed for the relics of Evangelist Luke.16


Figure 1.

Optimal validations that guarantee the identification of historical remains are possible using different approaches. One such approach consists of performing genetic tests on different authenticated objects or samples from the same individual or from a direct relative but with a completely independent origin of the samples. This was done successfully for the identification of the Polish astronomer Nicolaus Copernicus.17 The most convincing validation to identify the donor of particular remains is the analysis of the DNA of living relatives of the presumptive donor. This will also detect DNA contamination. Because these relatives are often several generations away from the individual in question, only markers on the mtDNA or Y-chr may be studied, thanks to their haploid mode of inheritance.2 This haploid mode means that any living maternal or paternal relative of the individual in question should have an identical mtDNA or Y-chromosomal type with the sample to be identified.18 This has been successfully realized in the identification of the grave of the Nazi official, Martin Bormann.19

A recent study reported the genetic identification of the remains of two kings of France by comparing their Y-STR profiles. One of these samples is a presumptive blood sample of King Louis XVI (1754–1793), who died on the guillotine during the French Revolution. After the execution, many spectators apparently soaked their handkerchief in the blood of the king. One of the many handkerchiefs so collected was apparently kept in a pyrographically decorated gourd, which is now in the possession of an Italian family. Careful examination confirmed that the gourd indeed contained a handkerchief with traces of human blood powder.20 DNA analyses revealed a 17 Y-STR profile and the confirmation of the Y-chromosomal haplogroup G(xG1,G2).20 The other sample is a mummified head attributed to King Henri IV (1553–1610), the first king of France from the House of Bourbon and a direct paternal ancestor of Louis XVI (Figure 1). During the French Revolution, the mummified body of Henri IV was excavated from its burial place in the Basilique St Denis in Paris. Unconfirmed accounts mention that it was decapitated. This ‘mummified head’ of Henri IV then would have followed a complex route over the years and came finally in the hands of collectors of royal relics and antiquaries. Based on 22 scientific and historical arguments, the head was recently identified as belonging to the French King Henri IV.21 Nevertheless, this identification remains controversial as several historical counter-arguments have been formulated.22, 23 Recently, after an initial failure to find DNA, a second effort to genotype the ancient DNA of the head was more successful. An mtDNA and a Y-chromosomal profile were obtained.24 The similarity between the six Y-STRs of a sample of the head, whereby only three alleles could be reproduced in a second PCR analysis, with the Y-STR profile of the blood sample presumably of Louis XVI was an indication for the attribution of both samples to the two Bourbon kings of France.24


Figure 2.

The three Bourbon males were correctly assigned to the main Y-chromosomal haplogroup R1b (R-M343) using the Whit Athey’s Haplogroup Predictor as it was confirmed by Y-SNP typing. The individuals were further assigned to sub-haplogroup R1b1b2a1a1b* (R-Z381*) based on the latest update of the Y-chromosomal phylogenetic tree of AMY 1.2.31 The 38 Y-STR haplotypes of the living donors were compared to each other (Table 1). A maximum of four mutational differences out of 38 Y-STRs was found between the living donors, namely between samples A and JH (Table 2). Next, these haplotypes were also compared with the haplotypes from the blood sample and the head (Table 1). There were 25–26 mutational differences between the 17 common genotyped Y-STRs of the living donors and the blood sample, assuming that each mutation leads to a gain or loss of one repeat unit on a Y-STR. There were eight mutational differences between the six common Y-STRs of the living donors and the head sample and even five out of the three confirmed STRs (Table 2). Based on the calculated mean mutation rate for each set of Y-STRs using the individual mutation rates measured in Ballantyne et al35 and based on the formulas of Walsh,36 the 95% confidence interval for the number of meioses between both individuals of each sample pair was calculated (Table 2). Finally, the mitochondrial DNA analysis of the head also did not support the attribution of the sample to Henri IV. According to Charlier et al,24 the donor of the head belongs to mtDNA haplogroup U5b*. Our previous study of a series of living and deceased maternally related relatives of Louis XVII showed that the mtDNA haplogroup of the Habsburgs belonged to haplogroup H.4, 5

Detailed Y-chr haplotypes and haplogroups were obtained for three living relatives of the House of Bourbon from which the French kings originated from 1589 (Henri IV became King of Navarre in 1572 by his mother’s rights, and King of France in 1589 at the death of Henri III, as the eldest descendant of Louis IX (1214–1270)) till the end of the monarchy. Based on these results, it is clear that their genealogical common ancestor (GCA) based on the official genealogy, namely King Louis XIII, was also their biological ancestor. First, the sub-haplogroups at the finest level of the Y-chromosomal phylogenetic tree were identical for all three relatives, namely R-Z381*. As there are no known recurrent mutations observed for SNP Z381 and as it is not lying in a Y-SNP conversion hotspot on the Y- chr,31 male individuals who share this Y-SNP must also share a common male lineal ancestor at the point of the SNP’s first appearance.37 Y-chromosomal sub-haplogroup R-Z381* is a subgroup of R-U106, which has been found in Western Europe with the highest frequency of around 35% in the north of the Netherlands and in Denmark but with a steep frequency fall to the south as the frequency of R-U106 is only 7% in France.38, 39, 40 The frequency of sub-haplogroup R-Z381* within Europe itself is not yet well known, except in Flanders where the frequency is around 9% within the autochthonous population.41 Second, also the 38 Y-STR haplotypes of the three individuals were highly similar. Differences were found only on four Y-STRs. These Y-STRs have a high mutation rate in comparison with all the other genotyped Y-STRs (Supplementary Table S2, Supplementary Materials). Moreover, the 17 Y-STR haplotype of the three individuals according to the AmpFlSTR Yfiler kit (Applied Biosystems), which is a part of the genotyped 38 Y-STR haplotype, is compared with the other haplotypes in the YHRD database release 43 (www.yhrd.org). The 17 Y-STR haplotype of the three living donors was found in only 3 out of 53 576 individuals in the total database, with 1 individual from Poland, 1 from Italy and 1 European American. Based on earlier studies, it is common that individuals with a matching 17 Y-STR haplotype (identical by state) are not genealogically related with each other (identical by descent) within the radiation of sub-haplogroup R-M269*, in contrast to individuals with an (almost) matching 38 Y-STR haplotype.39, 42 Based on the calculated mean mutation rate for all 38 genotyped Y-STRs and the formulas of Walsh,36 the GCA is indeed most likely their true biological ancestor as well (Table 2). This means that no non-paternity event happened along the three studied in-depth paternal lineages although some rumors that the branch of Bourbon Orleans would be illegitimate (more details in Supplementary Materials). Therefore, the genetic analysis of the three DNA donors in this study revealed the Y-chromosomal variant of the Bourbon lineage, including King Louis XIII, King Louis XIV and Louis, le Grand Dauphin (Figure 1).



No paternal relationship was found between the living DNA donors and the donors of the blood sample of Louis XVI or the head of Henri IV. First, the Y-chr of the donor of the blood sample belongs to haplogroup G(xG1,G2) while the living Bourbon members belong to R-Z381*. Based on the time calibration of the Y-chromosomal phylogeny, the time of the most recent common ancestor (tMRCA) between individuals belonging to haplogroup G and R will be some 10 000 years ago.43, 44 Second, the strong differentiation on the 17 Y-STR haplotypes between the living donors with the blood donor also suggests at least 260 meioses between them (Table 2). Although a low number of Y-STR results were obtained for the presumptive head of Henry IV, even the six Y-STRs and the three confirmed Y-STR haplotypes showed high differences with the living Bourbon donors, suggesting at least >110 meioses between them (Table 2). Moreover, the genetic identifications of the presumptive head of Henri IV and the presumptive blood sample of Louis XVI were only based on a similar partial Y-STR profile of both samples.24 Nevertheless, no evidence for a genealogical relationship between the DNA donors of the head and the blood sample could be found based on the calculations of the tMRCA due to the low number of analyzed Y-STRs and the mutational difference observed on one Y-STR (Table 2).36 Therefore, the similarity between the partial Y-STR profiles of both samples as indicated by Charlier et al24 could have been just the result of coincidence. Finally, if the samples were indeed of both kings of France this would mean that there were at least two non-paternity events within the royal lineage of the House of Bourbon, namely that Henri IV was not the biological father of Louis XIII, next to another non-paternity between Louis, le Grand Dauphin and King Louis XVI (Figure 1). Non-paternity events—whether a man is not the biological father of his legal children—mainly occurs when the mother had sexual intercourse with a man other than the legal father or when an unreported adoption occurred. Nevertheless, the frequency of paternal discrepancy in the Western European populations is at most 3% and is probably <1% of human births.45, 46 Moreover, while there has been speculation about non-paternity cases in the family, the written records do not contain evidence for such events within this lineage of the House of Bourbon.47, 48, 49

Next to the Y-chr, the mitochondrial DNA analysis of the head of Henri IV also does not support the presumed identification of the head sample. According to Charlier et al,24 the donor of the head belongs to mtDNA haplogroup U5b* defined by three nucleotide changes at positions m.16239C>T m.16270C>T m.16311T>C (nucleotides are numbered according to the revision of the Cambridge Reference Sequence50) of the mitochondrial hypervariable region 1 (HVR1). These three diagnostic positions were confirmed in two different amplifications of the L16185-H16378 (numbered also according to the revision of the Cambridge Reference Sequence50) HVR1 fragment, proving that the results were reproducible. Henry IV was maternally related with Louis XVII—through his mother Jeanne III d’ Albret, over Anna of Habsburg to Marie-Antoinette25, 26 (Figure 2)—but this mtDNA haplogroup belongs to haplogroup H and did not show the three nucleotide changes as seen for the head sample. According to the latest mtDNA phylogenetic tree51 and its most recent update on www.phylotree.com, the MRCA of the head donor and Louis XVII lived more than several tens of thousands years ago. If the head sample was indeed attributed to Henri IV, this would mean that a non-maternity event happened somewhere between Henri IV and Marie-Antoinette, an event which is known to occur very exceptionally but without any indication in written records.52

Conclusion
By using in-depth genealogical trees of living members of the House of Bourbon, the ‘true’ Y-chromosomal variant of Bourbon males of the French dynasty lineage was reconstructed. This Y-chromosomal variant was different from the ones of the presumptive head of Henry IV and of the presumptive blood sample of Louis XVI. As the genetic identifications of these samples were only based on the similarity between the partial Y-STR profiles of both samples, these identifications can no longer be accepted. Moreover, matrilineal genealogical data revealed that the observed mtDNA variant for the head sample was not the one that could be expected. These discrepancies might indicate that the samples analyzed were actually contaminated with non-authentic DNA. As contamination is indeed always possible when dealing with samples with high DNA degradation,11, 12, 53 contamination cannot be excluded here, especially for the head as the first attempt to collect DNA from the sample was not successful.21 Alternatively, they would support the hypothesis that the French kings were not the donors of these biological samples. Y-chromosomal analysis on the already identified heart of Louis XVII, the son of Louis XVI,4 might further remove all doubts about the identification of the blood sample considered to be of Louis XVI and might also solve any controversy about the paternity of Louis XVI of this child. As was already stressed in earlier studies,5 DNA typing of living relatives who are paternally or maternally related should be a requisite to solve historical cases with ancient DNA. This is an essential extra quality control for DNA contamination as well as a guarantee for a more objective identification, as illustrated here by this study of the Bourbon family.

European Journal of Human Genetics (2014) 22, 681–687; doi:10.1038/ejhg.2013.211

Sweet potato originated in tropical America (16). When, from where, and how it subsequently reached Oceania have been the subjects of extensive debate. Its presence in precontact archaeological sites scattered throughout Polynesia has long been considered as direct evidence for prehistoric contact between Polynesia and America (17⇓⇓⇓–21). Also, the lexical similarity between terms for sweet potato in Polynesian languages (“kuumala” and its derivatives) and the terms for this plant (“kumara,” “cumar,” or “cumal”) found among Quechua speakers of northwestern South America supports the hypothesis that humans introduced sweet potato from South America to Polynesia (22), against the alternative hypothesis of natural long-distance dispersal of seeds (23). Finally, the tripartite hypothesis, first proposed by Barrau (24), developed by Yen (25), recently updated by Green (26) and reviewed by Clarke (27), posits both prehistoric and historic dispersal events, putting forward a tripartite origin to explain the distribution of the species (Fig. 1). In the first origin, the Polynesian sweet potato (Kumara lineage) was introduced by Polynesian voyagers who collected it somewhere from the western coast of South America, between 1000 and 1100 A.D. (26, 27). They may have rapidly diffused it throughout Polynesia, in already populated islands such as Hawaii, Easter Island, and some other islands of eastern Polynesia and then into New Zealand, around 1150–1250 A.D., with the original colonists. Other independent prehistoric introductions (from northern Colombia or even Central America) have also been hypothesized, but these lack support (23, 28). Also, the possibility of an early westward dispersal of sweet potato carried by Polynesians to Tonga, Samoa, and eastern Melanesia is suggested by early historical accounts (26, 29). The second and third origins of sweet potato in Oceania arose from European contact during the sixteenth century. Spanish “Manila-Acapulco” galleons introduced the Mesoamerican Camote to the Philippines around 1500 A.D. (Camote lineage), whereas Portuguese traders introduced to present-day eastern Indonesia the Batata from the Caribbean and Central America (Batata lineage). From these points, secondary dispersal events, mediated by local traders, European travelers, or both, may have distributed sweet potato in the western Pacific, probably early in the New Guinea highlands (around 1700), and much later (mid-19th century) into western and eastern Melanesia (Fig. 1). Two additional European introductions may also have contributed to early diffusion in the Pacific, one by Mendaña’s voyage to the Marquesas and Solomon Islands in 1595 A.D. and another by Queiros in Espíritu Santo (part of present-day Vanuatu) in 1606 A.D. (30).


Fig. 1. Prehistoric and historical dispersal of sweet potato in Oceania, as postulated by the tripartite hypothesis. This map summarizes the tripartite hypothesis, as updated by Green (26) and reviewed by Clarke (27). Dispersal events and the terms commonly used to designate sweet potato in the different regions [compiled mostly from Yen (25)] are represented. Archaeological and early historical records providing strong evidence for the presence of sweet potato in Oceania in the prehistoric period are also indicated.

Summary of the Situation in Tropical America. As discussed in a previous study (34) and summarized here in Fig. 2 A–C, combining the use of chloroplast and nuclear microsatellite markers, we established the existence of distinct gene pools in the Northern and Southern regions of the neotropics. Neotropical sweet potatoes are characterized by two distinct geographically restricted chloroplast lineages (in total 21 haplotypes), which correspond quite well to two nuclear genetic clusters, K1 and K2, identified by K-means clustering grouping (Fig. 2 A–C): most accessions from the Southern region (hereafter the Southern gene pool) exhibit chloroplast haplotypes of group 1 (79.7%) and belong to the nuclear cluster K1 (83%). Sweet potatoes from the Northern region (hereafter the Northern gene pool) carry mostly chloroplast DNA haplotypes of group 2 (85.6%) and belong to the nuclear cluster K2 (91.46%).


Fig. 2. Geographical distribution of nuclear and chloroplast genetic variation in sweet potato through space and over time. (A) Proportion of individuals belonging to each chloroplast lineage [lineage 1 in orange (Cp 1); lineage 2 in blue (Cp 2)] for each sampling site (country or archipelago). Only sites with four or more accessions were represented. Area of the circle is proportional to the square root of the sample size for the site. Values by region correspond to the frequency of chloroplast lineage 1. (B) Proportion of individuals belonging to each nuclear cluster (cluster K1 in orange; K2 in blue) for each sampling site, as determined by the DAPC analysis. Only sites with four or more accessions were represented. Area of the circle is proportional to the square root of the sample size for the site. Values by region correspond to the mean K1 ancestry, as determined by the Bayesian clustering. (C) Three bar plots showing for each individual: (i) the probabilities of membership in nuclear clusters K1 and K2 as determined by DAPC (Top); (ii) the probabilities of membership in nuclear clusters K1 and K2 as determined by the Bayesian clustering method implemented in Structure (Middle); and (iii) the individual’s chloroplast lineage (Bottom). Each individual is represented as a vertical bar, with colors corresponding to membership probabilities in clusters K1 (orange), K2 (blue), and chloroplast lineages 1 (orange) and 2 (blue). Individuals are organized by geographical origin, S for the Southern neotropical region, N for the Northern neotropical region, P for Polynesia, M for Melanesia, NG for New Guinea, and SEA for South-East Asia. (D) Genetic constitution of herbarium specimens collected from the 18th century to the early 20th century. Lower and upper halves of each circle represent the chloroplast lineage and nuclear cluster (as determined by DAPC), respectively.

Despite this clear-cut phylogeographic pattern, neotropical gene pools have secondarily come into contact, as shown by the admixture revealed with both chloroplast and nuclear markers (Fig. 2 A–C): in the Southern region, we detected some accessions with chloroplast group 2 haplotypes (16/65), some assigned to the nuclear cluster K2 [8/65 with the discriminant analysis of principal components (DAPC) method (37), 3/65 with the Bayesian clustering method implemented in Structure 2.3.3 (38, 39)], and some that were admixed individuals (3/65 with DAPC, 34/65 with the Bayesian method). Also, in the Northern region, we identified several accessions with chloroplast group 1 haplotypes (31/82), some attributed to cluster K1 (5/82 with DAPC), and some that were admixed individuals (2/82 with DAPC, 49/82 with the Bayesian method). This situation suggests that clones were exchanged between both regions and recombined with local material.

Assessing the Relative Contribution of Camote, Batata, and Kumara Lines of Sweet Potato Introduction into Oceania in the Modern Sample. Sweet potato varieties in Oceania represent a subset of neotropical diversity for both kinds of markers. They include representatives of both neotropical chloroplast lineages 1 and 2 (in total eight different haplotypes, including those most common in tropical America) and also some private haplotypes (eight), all rare and derived from the most common ones by only one or two mutation steps (Fig. S1 and Dataset S1). Nuclear-diversity indices [except for observed heterozygosity (Ho)] were slightly lower than those calculated for either of the two neotropical gene pools (Table S1).


Fig. 3. Global patterns of genetic differentiation. Neighbor-joining tree based on the Lynch distance for the global dataset. Individuals are labeled according to their geographical origin, in orange, blue, cyan, green, yellow, magenta, and black for accessions from the Southern region, the Northern region, Polynesia, Island Melanesia, New Guinea, South-East Asia, and “other regions” (Micronesia, Madagascar, Madeira), respectively. Herbarium specimens from Polynesia are indicated by red open circles.

The unrooted neighbor-joining tree revealed a relatively high degree of phylogeographic structure among Oceanian varieties (Fig. 3). As expected under the tripartite hypothesis, western Pacific varieties (varieties from South-East Asia, Melanesia, and New Guinea) are differentiated from those of Polynesia. However, incongruities between chloroplast and nuclear data prevent, in some cases, the definitive attribution of some Oceanian varieties to an original gene pool in tropical America (Fig. 2C).

From Ancient to Modern Patterns: Reshuffling of the Initial Genetic Background of Sweet Potato in Oceania.

In the Eastern Pacific. In contrast to all herbarium specimens up to the early 20th century, most of the contemporary varieties from eastern Polynesia (80.5%) carry a chloroplast lineage 2 haplotype, supporting the long-standing contention of ethnobotanists that later introductions have replaced initial sweet potato varieties in this area (35, 40). The relative stability of genetic constitution reflected by herbarium accessions from the end of the 18th century to the beginning of the 20th century from this area may indicate that early European reintroductions (e.g., by traders, whalers, and travelers) did not immediately erase the initial genetic heritage. At first, these movements may have redistributed clones already present in the region rather than introducing new genotypes, extending the dispersal of varieties (which the Polynesians themselves had originated) throughout Polynesia and probably even further westward (24, 35). The shift probably took place during the 20th century, surely before the 1960s, when Yen began his collections.

Differentiation patterns observed nowadays reveal that varieties from eastern Polynesia form a heterogeneous group, with a few accessions closely related to the Southern gene pool, some to the Northern gene pool, and some to varieties predominant in the western Pacific, whereas some others are intermediate and even form a well-differentiated cluster (Fig. 3 and Fig. S3). This group includes some clonal lineages (i.e., varieties with the same haplotype and only very few differences between genotypes) and also cultivars, which, although genetically closely related, may be derived from independent sexual recombination events. This pattern of genetic differentiation suggests that later reintroductions did not simply replace initial ones but rather reshuffled the initial genetic background. Although mainly clonally propagated, sweet potato reproduces sexually, and its volunteer seedlings are sometimes incorporated by farmers and multiplied as new clones (24, 41). In the Marquesas and Tahiti, sweet potato long remained of only secondary importance. In the early 20th century, only a few native varieties, poorly adapted to the humid tropical environments of eastern Polynesia, were recorded (42). Strong genetic bottlenecks, such as those that accompanied prehistoric diffusion of the crop by Polynesians who likely introduced a limited number of autoincompatibility groups, may have greatly limited sexual reproduction and subsequent selection of seedlings by farmers as new clones. Later European reintroductions may have allowed the initiation of local diversification. Hawaiian varieties exhibit a quite similar pattern to that shown by eastern Polynesian varieties, and most of them are part of the eastern Polynesian cluster (Fig. S3). Furthermore, some clones are shared between both regions (eastern Polynesia and Hawaii), attesting to the circulation of vegetative propagules between the two areas. Nevertheless, in Hawaii, early references by European explorers and traders described the sweet potato as a plentiful food (26), and in the early 20th century, Handy (43) characterized a great number of local varieties. Did independent prehistoric introductions, or early historical introductions by Spanish explorers (Fig. 1), broaden the initial genetic base, thus providing quite rapidly opportunities for widespread recombination and local selection of variants in this archipelago? Interestingly, early 20th Century herbarium specimens already exhibit a “mixed” constitution. Unfortunately, we identified no old Hawaiian herbarium specimens to confirm this hypothesis. Also, none of the New Zealand Kumara varieties groups with the eastern Polynesia/Hawaii cluster (Fig. S3). Two cultivars were clearly associated with the Southern gene pool on the basis of both kinds of markers and may represent true original Kumaras. Two other accessions, recognized by Maori informants to be native Kumaras (i.e., pre-European introductions) are associated with the Northern gene pool and may, instead, represent later reintroductions.

In the Western Pacific. In contrast, the lack of geographic structure among western Pacific varieties (Fig. 3 and Fig. S3) suggests a common genetic background consistent with the tripartite hypothesis. Moreover, this pattern suggests a wide circulation of varieties across this region, also attested to by the presence of several shared clones (Dataset S2). Still, the mean value of K1 ancestry found in Melanesia was slightly higher (0.343 ± 0.113) than that observed in New Guinea (0.276 ± 0.116) and in South-East Asia (0.279 ± 0.097), which both showed values very similar to that observed in the Northern gene pool (Table S1). The greater contribution of the Southern gene pool in Melanesia likely reflects early introductions of true Kumara clones, as also suggested by the common use of cognates of this term (“kumara/kumala”) to designate sweet potato in eastern Melanesia (28, 35). Also, New Guinea varieties appear to be slightly differentiated from those of other areas in the western Pacific. These genetic data, combined with the presence of many names seemingly unrelated to each other (24), likely reflect multiple processes of local diversification in this island. Also, the frequency distribution of pairwise Manhattan distances for western Pacific genotypes (Fig. S4) suggests that the high diversity in this region appears to have resulted, as suspected by Yen (25), from intensive recombination among introductions and from the use by farmers of plants issued from true seed, rather than from clonal evolution and selection by farmers of somatic mutants.
Did Genes and Names Disperse Together? The mixed clonal/sexual reproductive system of sweet potato permits frequent recombination of newly introduced genotypes with local material. In these recombined genotypes, the name likely follows the phenotype. Distinctive phenotypes may reflect alleles at loci that are under strong selection and may show no correlation with the widely reshuffled neutral genetic background. Thus, old names may continue to be applied to varieties that have been strongly affected by modern plant movement and local recombination. This is what likely happened to most of the “native” Polynesian varieties originating from South America, which were progressively admixed with material introduced much later. An illustration of this evolutionary scenario may be offered by the group of ancient Polynesian sweet potatoes named as Convolvulus chrysorrhizus dulcis by D. Solander (and followed by J. Forster) because of the bright yellow color of the internal root flesh. In 1960, Yen identified two very similar groups of yellow varieties (yellow internal flesh and fusiform root) with a pan-Polynesian distribution (the “re’amoa” and “hererai” groups) and considered by local people to be native Kumaras (25, 26). Whereas the “native” yellow variety described by Banks and Solander carries a chloroplast lineage 1 haplotype, all of the contemporary yellow varieties bear a chloroplast lineage 2 haplotype and are very closely related, probably forming a clonal lineage or a group of genetically closely related varieties. The neutral genetic basis likely has shifted, whereas the initial phenotype, and probably the initial names, appear to have been well conserved. Thus, genotypes and names do not always migrate together, and information on each gives access to complementary parts of the plant’s dispersal history.

Roullier, Caroline, et al. "Historical collections reveal patterns of diffusion of sweet potato in Oceania obscured by modern plant movements and recombination." Proceedings of the National Academy of Sciences 110.6 (2013): 2205-2210.