Autosomal studies (2021-2024)

The DNA studies featured on this page and the following ones are based on so-called autosomal DNA. In theory this will allow for a genome-wide analysis (covering all chromosomes). Which will by definition be much more comprehensive than haplogroup studies. Usually autosomal studies will be applying so-called admixture analysis. While more recent studies (post 2020) are also often investigating shared IBD segments (Identical By Descent) to make ancestral inferences. Several insightful studies making use of autosomal DNA have already been performed for Cape Verdeans. On this page I will discuss the following ones:

• A genetic and linguistic analysis of the admixture histories of the islands of Cabo Verde (Laurent et al., 2023).
• Sex-biased admixture and assortative mating shape genetic variation and influence demographic inference in admixed Cabo Verdeans (Korunes et al., 2022).
• Rapid adaptation to malaria facilitated by admixture in the human population of Cabo Verde (Hamid et al., 2021).

Follow link below for a more detailed and in depth review (incl. critique of historical framework and suggestions for new research):

• The Admixture Histories of Cabo Verde: New DNA Studies! (Fonte Felipe, 2024)

***

A genetic and linguistic analysis of the admixture histories of the islands of Cabo Verde (2023)

Link

Figure 1 (click to enlarge)

***

This study is a follow-up to “Parallel Trajectories of Genetic and Linguistic Admixture in a Genetically Admixed Creole Population” (Verdu et al., 2017). Which I discussed on the previous page. However this time their dataset has been expanded in a very useful way. For the first time featuring samples from all 9 populated Cape Verdean islands! Including Brava, Maio and Sal (n=225, see this overview). Most of the Santiago samples (40/50 from Praia) in this study are however retained from their 2017 study.¹ Due to space constraints I am only highlighting three main outcomes further below. But in fact several other fascinating topics are described in this paper. However this involves mostly preliminary findings which are unfortunately quite ambigious and open to multiple interpretation.²

Aside from expanding their Cape Verdean dataset this study also makes use of a much extended reference panel of mainland African samples to compare with for ancestral inferences. Including also historically relevant samples from Senegambia such as Jola/Diola, Fula and Wolof (see this link). The inclusion of Wolof samples was actually already suggested by the researchers in their previous study from 2017. So I am very glad they came through with this. As far as I know this is the first time autosomal DNA samples from Cape Verde are being compared with Wolof ones in a peer reviewed study. However there have already been previous research outcomes which attempt to measure genetic similarities between the Wolof and Cape Verdeans. See:

• Blood group analysis from 1957 (Jalofo = Wolof) (see this page for more discussion)
• Haplogroup analysis from 2011 which found a 41% similarity with Wolof/Lebou samples (Stefflova et al. (2011), (see this page for more discussion)
• IBD matching on 23andme from 2022 (see this page for more discussion)

Greater genetic affinity with Wolof samples among Barlavento islanders

____________________

“Furthermore, we find that all Cabo Verdeans almost exclusively share African haplotypic ancestries with two Senegambian populations (Mandinka and Wolof) and very reduced to no shared ancestries with other regions of Africa.

More specifically, we find that the Mandinka from Senegal are virtually the sole African population [for] Brava, Fogo, Santiago, Maio, and Boa Vista, and the majority of the African shared ancestry for Sal, São Nicolau, São Vicente, and Santo Antão. 

In individuals from these four latter islands, we find shared haplotypic ancestry with the Wolof population ranging from 4%–5% for individuals born on Sal (considering four or six possible sources, respectively), up to 16–22% for Santo Antão.“ (Laurent et al., 2023, p.9)

____________________

The results shown above are potentially very insightful! However you should also be aware of the various limitations which are tied to this sort of analysis. I will first discuss the findings of greater genetic affinity with Wolof samples among Barlavento islanders, especially Santo Antão. Which presumably might be connected with founding effects! I will then proceed with the minor Nigerian and Central African findings. Again very intriguing and showing differentation across the islands. But this is not a novelty. Because I myself have described this phenomenon in various blogposts since 2015.³

I have been waiting for a long time to see Wolof samples being compared with Cape Verdean samples for genetic affinity, based on autosomal genotyping. So I am very pleased that Laurent et al. (2023) performed this type of admixture analysis. But as always careful interpretation is a must! Given that it is practically impossible that Wolof ancestry does not exist for Santiago and Fogo as well. Plentiful historical records as well as linguistic evidence provide solid indications of an early Wolof presence in these foundational islands. Most likely the Wolof were forming a majority even in the first century of settlement (1460-1560). And hence they were decisive for Cape Verdean ethnogenesis in all islands ultimately!

On first sight the outcomes seem to be suggesting a shift in the composition of Upper Guinean roots when comparing people from the Barlavento islands with Sotavento islands. It is indeed remarkable that people from Santiago, Fogo but also neighbouring Brava and Maio are exclusively showing Mandinka as best fitting Senegambian population. While Wolof affinity seems to increase on a gradient among Barlavento islanders. Peaking for the western-most island of Santo Antão (see this map). Perhaps mere coincidence but this finding is actually also somewhat in line with this pioneering research from 1957, which I mentioned above already.

I actually have a strong hunch this outcome is firstmost reflecting a modeling issue/artefact. Still I am curious to know how this might relate to these old statements of mine from 2014:

____________________

“[Wolof:] Pending on future DNA research founding effects might be substantial. Therefore most widely spread among all segments of Cape Verdean population?“(Fonte Felipe, 2014)

“So we can assume that for a majority of Cape Verdeans the African part of their ethnogenesis dates mostly from the period before 1731 at the very least and most likely especially the 1500’s playing a crucial part. […]

For a subset of Cape Verde’s population there was still continued African geneflow also in the 1700’s and 1800’s leading perhaps to some differentiation in African ethnic origins between islands as well as socially defined subgroups on the same island.”

“could mean that the mulattos or mestiços from the census might have “timecapsuled” the African part of their mixed ancestry dating from the 1400’s-1600’s more so than freed blacks/forros. Many of these mixed people would later on be prominent among those sent out to to settle the islands of the north in the late 1600’s/1700’s” (Fonte Felipe, 2014)

____________________

More follow-up research is needed to see how robust this finding from Laurent et al. (2023) might turn out to be. Again by default this type of analysis leads to simplified results. Because after all their modelling is set up in such a way that only 4 or 6 source populations are taken into consideration. I suppose more granularity could be obtained by further expansion of the Senegambian reference panel.⁴

At first I was surprised that apparently there was no genetic affinity found with the Fula and Jola samples which are included as well in the African reference panel of this study. Perhaps a telling sign of so-called over-smoothing. Whereby similar types of DNA get grouped in one lump category because it has a greater pull (mostly Mandinka in this case). Because again there is practically no way that Cape Verdeans would not have any meaningful ancestral connections with the Fula or Jola people (from Casamance)! Upon further inspection it appears the Fula samples were removed from this particular analysis (see this comment). Which is the reason they are not considered as a source candidate in figure 1. But this still does not explain the absence of any Jola similarity. I find it disappointing that the authors did not explain this more clearly in their text.

Even more importantly I think another approach is required. In particular the more commonly used analysis of shared IBD segments. Strongly indicative of recent ancestry, when reaching a certain threshold. This kind of IBD analysis is actually also the basis for the genetic groups being reported on 23andme, a personal DNA testing company. For more discussion see the 23andme section on this website.

Minor “Nigerian” and Central African lineage among Cape Verdeans

Table 1 (click to enlarge)

Source: my survey findings for 100 Cape Verdean 23andme results (2021). Notice the subgroup averages of 3.1% “Nigerian” and 2.4% “Central & Southeast African” for Barlavento. Somewhat higher than for other subgroups. However such minor scores are sometimes also found for Brava, Fogo and Santiago. 

***

____________________

“Finally, we find limited (1–6%) shared haplotypic ancestry with East Western […] or South-West Central […] African populations in all Cabo Verdean islands, except Fogo and Brava, and the specific populations identified and their relative proportions of shared haplotypic ancestries vary across analyses.” (Laurent et al., 2023, p.9)

This potentially indicates differences in shared ancestries with different continental African populations across islands in Cabo Verde, which remains to be formally tested.” (Laurent et al., 2023, p.8)

____________________

The quotations above are referring to another highly interesting finding from Laurent et al (2023). All the more so because it corroborates my own research findings from 2015, 2018 and 2021. As can be seen in figure 1 this study shows a minor yet clearly detactable genetic affinity between their Cape Verdean dataset and their samples from the eastern part of West Africa (“East Western Africa” = Lower Guinea). Especially Igbo ones from Nigeria. While also their (southwest) Central African samples (Kongo, Ovimbundu and Kimbundu) seemed to show shared ancestry in some cases. Especially for Barlavento islands but to a lesser degree also Maio and Santiago. In my own survey findings (see Table 1 above) I found somewhat similar patterns.

When I first discovered this occurence of unexpected non-Upper Guinean lineage in 2015 I was naturally very cautious. Because afterall admixture analysis has several limitations. Furthermore this is unexpected when going by Cape Verde’s geography and known history. Documented sources clearly describe the area in between Senegal and Sierra Leone (Upper Guinea) as practically the only provenance zone for African captives brought to Cape Verde safe for some individuals who came on atypical slave voyages from further away.⁵ See also this link.

However these minor Lower Guinean and Central African scores keep coming up for Cape Verdeans when they do personal DNA testing. Even after several updates on both Ancestry and 23andme. To be sure this still represents a small proportion of the African genepool for Cape Verdeans. But certainly worthy of follow-up investigation! This is something I have actually already done in 2018. Which is why I am inclined to say that in many cases even trace-amounts of Central African or “Nigerian” DNA might be the “real deal” for Cape Verdeans. Having these additional findings from Laurent et al (2023) makes for an even stronger case that this could be true. I have explored the various possibilities already to a great extent in previous blog posts. So for further discussion see:

• “Southeast Bantu”, artefact or genuine? (scroll down to section 5)
• Beyond Upper Guinea: valid outcomes or misreading by AncestryDNA? (scroll down to section 4)
• Regional admixture as corroboration for African matches outside of Upper Guinea? (scroll down to section 4)
• Substructure within Cape Verde (scroll down to section 2)

____________________

“Follow-up research focusing on associated DNA matches as well as dedicated family tree research may clarify things in individual cases. While possibly also a locally specific historical context may apply for the Barlavento islands. In particular in regards to (slightly) deviating slave trade patterns when compared with Santiago, the main hub of slave trade with Upper Guinea. Perhaps occasionally involving contraband Northwest European traders?” (Fonte Felipe, 2021).

“such scores (when truly genuine) might either be due to African captives outside of the expected Upper Guinea area being present (sporadically) in Cape Verde during the Slave Trade period. Or otherwise also caused by family histories involving back & forth migrations to and from Angola, Brazil, São Tomé & Principe and Mozambique. All fellow ex-Portuguese colonies. This latter scenario will usually be easier to investigate of course.” (Fonte Felipe, 2021).

____________________

Admixture scenario’s for each island

Figure 2 (click to enlarge)

____________________

“[…] scenario-choices indicate that multiple pulses of admixture from the European and African source populations (after the founding admixture pulse, two independent admixture pulses from both Africa and Europe […]: Scenario1), best explain the genetic history of individuals born on six of nine Cabo Verdean islands. Furthermore, we find that even more complex scenarios involving a period of recurring admixture from either source best explain the history of the remaining three islands.” (Laurent et al., p.19)

“We find that admixture from continental Europe and Africa occurred first early during the TAST history, concomitantly with the successive settlement of each Cabo Verdean island between the 15th and the early 17th centuries”. (Laurent et al., p.26).

“Finally, we find that recent European and African admixture in Cabo Verde occurred mainly […] in the 1800s“. (Laurent et al., p.27).

“our results highlight both the unity and diversity of the genetic peopling and admixture histories of Cabo Verde islands […].” (Laurent et al., p.27)

____________________

The main research efforts of Laurent et al. (2023) are directed towards the dating of African and European admixture pulses for each island. Making use of a new type of software specifically developed for this purpose (Methis). Naturally this is quite an ambitious endeavour, given all the complexties. But potentially very insightful! Many of the outcomes are actually quite plausible and also rather useful. But regrettably I find that the historical contextualization in this study has been misguided by exaggerated expectations of the genetic and social impact of Trans-Atlantic Slave Trade (TAST). Which often leads them to demonstrably inaccurate interpretations of their findings.⁶ Frankly I think therefore that their flawed historical framework should not be copied by any other research teams who might take this paper as a basis for similar investigations regarding Cape Verde or other parts of the Afro-Diaspora. I discuss this in much greater detail in the second section of this blogpost:

• Getting the historical context right (Fonte Felipe, 2024)

However this does not diminish their core finding which would be that across the Cape Verdean islands a substantial component of shared ancestry can be detected. Most likely dating back to foundational admixture events taking place during the 1500’s on both Santiago and Fogo, the earliest populated islands. Greatly indicative of a “stepping stone” settlement of the islands. This much is also supported by two other recent papers: Korunes et al. (2022) and Hamid et al. (2021), which use a similar methodology. See further below for more details. In fact this outcome was already established in an earlier paper (Beleza et al., 2012, discussed on this page). However back then the analysis was based solely on Y-DNA (male haplogroups). While this time genomewide autosomal DNA markers are being used.

At the same time Laurent et al. (2023) does also identify additional geneflow for each island, especially during the 1800’s. Which would be indicative of differentiation among (subsets) of the islands, at least to some degree. This has a profound impact for Cape Verdeans who have done personal DNA testing. Because it will determine their DNA matching patterns to a great extent. First of all when looking at close DNA relatives and their Cape Verdean island origins. These will usually closely mirror the cluster patterns shown in figure 4. As long as your family origins only involve 1 island or neighbouring islands.

But also the odds of finding mainland African as well as Portuguese DNA matches will naturally be highest when relating to the most recent timeperiods (1800’s). Often indicated by fitting genetic groups/country matches on 23andme and to a much lesser extent also by Ancestry’s tool of DNA communities. However this could possibly still be obscuring any earlier links with African and European heritage dating from the 1500’s/1600’s. As well as any unrecorded earliest island origins from either Fogo and/or Santiago for Barlavento islanders. Because due to dilution and matching thresholds these connections will often remain undetected. See also this blogpost series (part 2 is still to be written).

Finding the approximate time period(s) when your mainland African ancestors arrived in Cape Verde is crucial for Tracing African Roots. This is something I have always said from the start on my blog and also on my CVRAIZ website. So these new estimates for the timing of African and also European geneflow are potentially very significant! On the other hand much more research is needed. This type of analysis is not intended to be 100% accurate but rather indicative. The outcomes as displayed in figure 2 are heavily dependent on underlying assumptions and other inherent limitations of statistical modelling. This much is also explained within the study.⁷

Ancestry Timeline on 23andme for a Cape Verdean

Figure 3 (click to enlarge)

Source: 23andme. This screenshot features the Ancestry Timeline results for a Cape Verdean who has done extensive family tree research (into the 1700’s). He is indeed aware of (several) Portuguese ancestors within the estimated time frame of 1760 and 1850. But he does not know of any mainland African ancestors (yet). It is theoretically possible he might also have one or a few mainland African ancestors from the 1800’s. But generally speaking most of his Senegambian/Guinean ancestry should date back from much earlier. Just like for almost all other Cape Verdeans. 

***

On 23andme there is a similar tool to learn more about when approximately a particular ancestry entered your family tree. This is called Ancestry Timeline. Sometimes the estimates given are quite useful but other times they are completely off. Again mostly due to inherent modeling issues and assumptions not always being relevant. This is why I always recommend to judge plausibility by contrasting with the actual known historical context, personal genealogy as well as any other relevant clues.

In fact within Laurent et al. (2023) there is an extensive discussion of the admixture scenario’s per island, see appendix 5. They also provide a useful Excelfile which contains demographic information/dates which were used for their modeling efforts. However again a word of caution: the authors of these studies are not trained historians! The historical data they are using is incomplete. By default I would say, but actually I am also missing some important sources. Furthermore their interpretation of key aspects of Cape Verdean demography is sometimes questionable or even inaccurate.

• Appendix 5: Genetic admixture histories of each Cabo Verdean island of birth (Laurent et al., 2023).
• Historical landmark chronology for the peopling history of Cabo Verde. (Laurent et al., 2023).
• Historical Demography of Cape Verde (Fonte Felipe, 2014)
• Getting the historical context right (Fonte Felipe, 2024)

***

Sex-biased admixture and assortative mating shape genetic variation and influence demographic inference in admixed Cabo Verdeans (2022)

link

Figure 4 (click to enlarge)

____________________

“All of the timing methods we used placed admixture timing for the different islands closer together than historical dates of settlement, consistent with historical expectations that the initial admixture in the southern islands was significant, and that many individuals that occupied the northern and eastern islands during the second and third settlement stages of Cabo Verde were already admixed. ” […]

“This observation, together with IBD and kinship patterns, support the serial founding of the groups of islands as the main model of settlement of Cabo Verde, as opposed to their independent settling as some historical data suggest (Correia e Silva 2002). This type of serial founding scenario is common throughout recent human migration, underscoring that settlement patterns, in addition to settlement timing, are critical components of accurately inferring human demographic history.” (Korunes et al., p.12).

____________________

This study from 2022 shows some degree of overlap with Laurent et al. (2023) with regards to modeling the timing of admixture among Cape Verdeans.⁷ The outcomes are actually quite similar and therefore mutually reinforcing. In particular supporting the hypothesis of a sequential “stepping stone” settlement of the islands. Starting in Santiago and Fogo and ending in the Barlavento islands. Again the same disclaimers for statistical modelling apply. I do find that the historical framework as decribed by Korunes et al. (2022) is much more in allignment with the historical sources they are referencing.⁸

Similar to Laurent et al. (2023) also this study performs an investigation of so-called ROH (runs of homozygosity). Basically this type of analysis is based on shared IBD (Identical by Descent) segments and intended to clarify population structure. Which is certainly very useful. However these preliminary outcomes are unfortunately quite ambigious and open to multiple interpretation. But in addition the authors also establish the genetic clustering of Cape Verdeans according to island origins. The researchers use the same extensive Cape Verdean dataset (n=563) which was already utilized by Beleza et al. (2012/2013). These older studies have already been discussed on this page.

Most admixture dates back from 1500’s?

____________________

“The dating of the socalled Creolization process/transition is fundamental for tracing back African ethnic roots“. (Fonte Felipe, 2015)

“Tracing back from Creole to African is going to take you back several centuries and many generations therefore. Everyone has individual family trees (I will continue stressing this) however generally speaking for many Cape Verdeans it might mean that most of their African mainland born ancestors arrived in Cape Verde more than 400 years ago […]”. (Fonte Felipe, 2015)

____________________

Genetic clustering of Cape Verdeans according to island origins

Figure 5 (click to enlarge)

***

Figure 6 (click to enlarge)

***

Rapid adaptation to malaria facilitated by admixture in the human population of Cabo Verde (2021)

link

Figure 7 (click to enlarge)

____________________

“Interestingly, this hypothesis goes back to a voyage in 1721 in which Captain Roberts, reported that a disease in Santiago is ‘dangerous to strangers’ during the rainy season. Consistent with ancestry-mediated protection from malaria, the record has been interpreted by medical historians to suggest that ‘foreign visitors and residents of European descent seem to have suffered more than the African and Afro-Portuguese majority’ from malaria in Santiago “ (Hamid et al, 2021, p.3)

“We hypothesized that selection at DARC may have increased the genome-wide West African ancestry proportions in the current population of Santiago.” (Hamid et al, 2021, p.7)

____________________

This study is not really relevant for uncovering the specific African lineage of Cape Verdeans. However the topic of how malaria-resistant genes were spread among Cape Verdeans is quite fascinating! Hamid et al. (2021) also provides an interesting but unconvincing take on why Santiago shows higher African admixture levels than other islands. Regrettably their reasoning is quite speculative, incomplete and therefore not compelling in the end. I fail to see why more mixed Cape Verdeans from other islands and also in Santiago would not have equally benefited from having malaria-resistant genes. This much is in fact also reported by historical sources such as mentioned in the quote above (Captain Roberts: “Afro-Portuguese”). Also the authors themselves as well as the peer reviewers of this study point out various limitations of their highly mathematically driven analysis.⁹ 

To be sure it does have some merit to look into this aspect of genetic adaptation/selection to a disease environment whereby it might be beneficial to retain higher African ancestry levels. At least in order to obtain the malaria resistant genes related with African ancestry. But the historical context clearly indentifies other more decisive factors which resulted in the Santiago population showing higher levels of African admixture than on other islands (safe for possibly Maio). In particular the early formation of so-called Badiu communities in Santiago’s rural interior. Ex-run-away slaves who later on were joined by manumitted slaves. Already during census in 1731 they were the absolute majority of the island (67% “Forros“, see table 3 and 6 on this page). Mostly cut off from further admixture because of their relatively remote/isolated location. The reason other islands show greater admixture levels is firstmost to be explained by way of their initial settlement by already admixed individuals! While possibly other islands also absorbed greater numbers of Portuguese migrants/exiles during the late 1700’s and 1800’s. See also footnote 1 on this page for a Fogo related comparison.

• Cabo Verde certified malaria-free by WHO, marking a historic milestone in the fight against malaria (End Malaria, 2024)
• Cape Verde reaches malaria-free milestone (BBC, 2024)
• Director-geral da OMS destaca “bons progressos” de Cabo Verde no sector da saúde (Expresso das Ilhas, 2024)

___________________________________________________________________________

Notes

1) In Verdu et al. (2017) it is mentioned that many of their Santiago-based samples had either parents or grandparents who were born on different islands. In some cases even from abroad (Angola, São Tomé etc.). Furthermore most of their samples were from Praia instead of Santiago’s interior. This might be somewhat problematic.  Because it is known that Cape Verde’s capital has had a great influx of migrants from other islands (aside from the interior as well). Praia’s population grew from only 20,000 inhabitants in 1960 to a staggering 170,000 in 2020 (see this graph)! 

As mentioned in Laurent et al. (2023) it is also known that people from São Vicente, the second largest island by population, often have recent ties to neighbouring Santo Antão. This circumstance makes it all the more crucial that a “4-grandparents born on the same island” criterium is being applied. In particular when aiming to make statements about the historical population structure of each separate island. Of course populations are constantly changing and inter-island migration has been in existence for a longer time. Still the scale of modernday migration and hypermobility is clearly a new development which should be taken into account when striving for quality sampling.

In the interest of even more refined analysis for this research field I would also like to make the following suggestion. Perform dedicated sampling among older people living in the remotest part of the islands. I believe on Santiago the population from the interior is often referred to as Badiu di Fora. I know for Santo Antão people with the “thickest” accents are said to be talking “fund“. Usually stereotyped to be from the more remote ribeira’s (valleys). Given relative isolation I would assume that sampling these people will offer very precious testimony of both genetics and linguistics! A legacy which might now be under threat given the rapid modernization taking place in most islands.

• Discussion on undersampling from Santiago

2) Laurent et al. (2023) also provides a continuation of their earlier research into how linguistics and genetics may correlate among Kriolu speakers in Cape Verde (see Verdu et al. 2017, discussed on this page). This time making use of a much expanded sample database which covers persons from all 9 populated Cape Verdean islands. However they are still using the same Santiago-biased linguistic dataset. Hence the outcomes are quite similar as obtained in 2017 and do not reveal any major novelty. The correlation although positive actually appears to be rather weak (Spearman ρ=0.2070, p=0.0018, see this figure).

Nonetheless this researchfield of how linguistics may correlate with genetics remains very captivating! I personally think it is very striking to see both genetic and linguistic continuities between Barlavento and Sotavento islands, despite also some divergence. As they say in Cape Verde: Nos Tud Nos É Criol! Hopefully their intended follow-up efforts will be more in tune with the distinct island specific varities of Kriolu. I am looking forward to more insights. One of the involved researchers, Marlyse Baptista, is currently working on a new publication together with Ousmane Cisse.

• Definiteness and genericity in Cabo Verdean Creole and Mandinka (2023)

3) The earliest summary of my Cape Verdean survey findings was put online in 2015. Which was (as far as I know) the first ever demonstration that Cape Verdeans are overwhelmingly Upper Guinean in origins. When looking only at their African DNA and based on autosomal genotyping. Previous haplogroup studies (see this link) did also come to the same conclusion. However they were restricted in the sense that they didn’t measure the complete, genomewide ancestry of Cape Verdeans but only focused on maternal lineages.

To specify the overlap with my own research more clearly. See below for an overview of aspects about Cape Verdean genetics which I already established several years ago.

1) Cape Verdeans being mainly an Upper Guinean/Iberian mix: 2018, 2021
2) Interisland differentiation/substructure: 2015, 2018, 2021
3) Minor non-Upper Guinean lineage: 2015, 2018a, 2018b, 2021
4) Comparison with other parts of Afro-Diaspora: 2016, 2018a, 2018b, 2021
5) Recent ancestry (1800’s) from Portugal & Upper Guinea: 2018a, 2018b, 2022

Furthermore my own research has also gone beyond what has been published sofar by peer reviewed studies. In particular for these aspects:

1) IBD matching patterns (Ancestry, 23andme): 2018, 2022
2) Analysis of North African/Sephardi Jewish lineage: 2015, 2018a, 2018b
3) Analysis of South Asian and Amerindian lineage: 2018a, 2018b, 2021
4) Indications of Northwest European lineage: 2013, 2018, 2021

4) Within Laurent et al. (2023) two methods are being used to arrive at their “ethnicity estimates”. The first one involves the classical ADMIXTURE software. This method requires very careful interpretation because the outcomes are greatly dependent on the selection of  the reference populations. As well as the number of so-called K-runs. Potentially misleading when taken too literally! Quite comparable to the various admixturetools on GEDmatch which are known to be greatly variable and confusing! As stated by the authors themselves:

____________________

“The resulting ADMIXTURE patterns could be due either to admixture from populations represented in our dataset, to admixture from populations un-represented in our dataset, or to common origins and drift.” (Laurent et al., 2023, p.9).

____________________

Furthermore they also apply SOURCEFIND for their so-called local ancestry inference, as shown in figure 1. It is important to grasp that this type of analysis is based on quite a low level of granularity. When compared with what is provided by personal DNA testing companies such as 23andme and Ancestry. Given the limitations of only 4 or 6 source populations. But also apparently going by 100 possible admixture “slots” only. The way the results are presented appears to me to be quite similar to the Oracle predictions, also on GEDmatch. Certainly useful at times, but again very dependent on careful interpretation.

____________________

“We […] conducted two SOURCEFIND (Chacón-Duque et al., 2018) analyses using all other populations in the dataset as a possible source, separately for four or six possible source populations(‘surrogates’), to allow a priori for symmetric or asymmetric numbers of African and European source populations for each target admixed population. […]

Each individual genome was divided in 100 slots with possibly different ancestry, with an expected number of surrogates equal to 0.5 times the number of surrogates, for each SOURCEFIND analysis.” (Laurent et al., 2023, p.31).

____________________

5) Laurent et al. (2023) seem to not have been fully aware of how Cape Verde’s slave trade was overwhelmingly carried out with Upper Guinea during the 1500’s. Even when this is extensively discussed in the sources they list themselves in their references. This leads them to inaccurate and highly speculative interpretations of their minor Lower Guinean & Central African findings (see last paragraph p.69).

____________________

“Numerous enslaved-African populations from […] Central Africa were forcibly deported during the TAST to both Cabo Verde and the Americas, as shown by historical demographic records (1,25). There is still extensive debate about whether enslaved-Africans remained or more briefly transited in Cabo Verde during the most intense period of the TAST, in the 18th and 19th centuries, when the archipelago served as a slave-trade platform between continents.” (Laurent et al., p.23).

“However, historical records unquestionably demonstrated that numerous other populations from West Central and South Western Africa were enslaved and forcibly deported to Cabo Verde during this era [1700’s].” (Laurent et al., p.69).

____________________

The statements above are wrong for several reasons:

1. TAST (Trans-Atlantic Slave Trade) numbers for Cape Verde peaked during the 1500’s and not the 1700’s/1800’s (see this overview, taken from the Slavevoyages database)
2. During the 1400’s/1500’s it was explicitly forbidden for Cape Verde to have any slave trading activities beyond Sierra Leone (see these sources). Curiously Laurent et al. (2023) do mention this with extensive references even on page 27, but they do not process this information elsewhere.
3. During the late 1600’s and 1700’s a low-volume and sporadic slave trade was carried out clandestinely by Northern Europeans. Aside from also various Portuguese state monopolies. Most notoriously the Grão Pará and Maranhão Company. Purely to cover a (weak) domestic demand of enslaved labour and not for transit purposes (see these sources).
4. Only during the 1800’s there was a short-lived revival of Cape Verde acting as a transit hub for TAST again. But at that time enslaved people were less than 5% of the total population in Cape Verde. (see this table)
5. Most of these late arrivals from the 1700’s/1800’s would still be almost exclusively from Upper Guinea. During a very detailed slave census held in 1856 enslaved people from Lower Guinea (Costa da Mina) or Central Africa (Angola and Cabinda) were very rare and clearly exceptional. Also adding persons from Brazil and Cuba they merely represented 6 individuals out of a total of 5182 persons, a miniscule 0,1%! (see census 1856).
6. Also due to nautical reasons (trading winds, currents) it was not convenient to make a stop in Cape Verde for outgoing TAST voyages departing from Central Africa or even Lower Guinea (Windward Coast up to Bight of Biafra). At times this might still have happened but it seems only very sporadically so. After all these adverse nautical circumstances (such as the Guinea current) would have been widely known in this Age of Sail.

Naturally this forcible linking with generic TAST patterns is not beneficial for understanding the minor Lower Guinean and Central African admixture among Cape Verdeans! I was surprised that an arguably much more plausible scenario (among others) was not mentioned in Laurent al. (2023). That is to say African geneflow from Lower Guinea and Central Africa might have ended up in Cape Verde by way of detour via other parts of the Portuguese colonial empire.

Not only restricted to intermingling with enslaved persons but also in fact free and/or mixed-race persons from either Brazil, Angola, Mozambique or São Tomé & Principe. Back & forth migrations as well as more fleeting stop-overs by sailors are all realistic scenario’s I suppose.  A strong increase of either Lower Guinean or Central African admixture could very well be indicative of a more recent intermingling.  Aside from solid family tree research also to be corroborated by IBD matching patterns.

6) Despite their undoubtedly well-intentioned efforts I do have to point out that I have mixed feelings about the accuracy of the historical framework, as laid out in Laurent et al. (2023). This study seems to have overlooked an essential aspect of Cape Verde’s historical demography. Because most likely already in the late 1600’s a majority of the Cape Verdean population (both black and mixed-race) was no longer enslaved (see this table). Cape Verde is indeed part of the Atlantic Afro-Diaspora. And especially in its early history (1460-1660) Cape Verde was greatly impacted by the TAST (Trans-Atlantic Slave Trade). But several singular aspects about the Cape Verdean experience do require very careful and context-dependent analysis.

Incomplete knowledge of Cape Verde’s history can result in exaggerated expectations of the genetic and social impact of Trans-Atlantic Slave Trade (TAST). However not everything about Cape Verde is TAST related! A lack of much needed nuance leads to an overly deterministic interpretation within Laurent et al. (2023).

____________________

“Be careful to respect the localized context and different historical trajectories across the Afro-Diaspora. Instead of just letting one single perspective on inter-racial relationships overcloud things. This goes especially for Cape Verdeans who despite being part of the Afro-Diaspora in many aspects have a unique history of their own. And as a consequence their family trees & genealogy will usually not fit in well with simplistic generalizations. And even less so with ideologically charged assumptions arising from a specific USA context or an overly Americanized mindset.” (Fonte Felipe, 2018)

“For Cape Verdeans it is vital to be aware that despite being the earliest hub of Trans-Atlantic Slavery in the 1500’s Cape Verde was actually also one of the first creolized societies in the Atlantic world! In fact the economic importance of slave trade quickly declined after the 1600’s because other slave ports on the mainland became more significant. Due to both racial mixing and a greatly diminished need of slave imports Cape Verde had a locally born population with a clear majority consisting of free-status Afro-descendants already in the early 1700’s (as confirmed by the 1731 census)!

Slavery did continue up till 1878 for a minor part of the population. However the resulting gene flow from mainland Africa must have been much more subdued in later time periods, on average. Given that the enslaved portion of Cape Verde’s population was […] below 20% throughout the 1700’s. This implies that generally speaking when tracing back to mainland Africa Cape Verdeans will often have to go back to the 1500’s and/or 1600’s instead of the 1700’s/1800’s as is more usual among Trans-Atlantic Afro-Diasporans.” (Fonte Felipe, 2021)

____________________

For a continued and more thoroughly referenced critique see:

• Getting the historical context right (Fonte Felipe, 2024)

7) Interestingly both Korunes et al. (2022) and Laurent et al. (2023) provide timing estimates of admixture within the Cape Verdean population. However there is one key difference between these two studies. Which is the crucial assumption that mating would be either random or assortative (seeking partners who resemble you in either pheno- or genotype). This is a fascinating topic indeed. But also rather technical. However Lauren et al. (2023, p. 25/26) seem to admit that the approach taken by Korunes et al. (2022) would result in more realistic estimates.

One aspect to be taken into consideration then also being the role of mulatto/mixed-race men in the dispersion of European genes among the Cape Verdean population. For example the spread of European Y-haplogroups is not per se a reflection of the reproduction rate of European men but also their mixed-race male descendants! Especially when being privileged by social status of any kind they may have had disproportionate offspring, incl. also with unmixed “black” Cape Verdean women. Just going by anecdotal evidence many Cape Verdean men of older generations tend to be boastful of the number of children they have conceived. Polygamy is also being mentioned by several historians as a frequent aspect of Cape Verdean society. Hopefully future research will be able to shed more light on this topic.

Recurring famines in Cape Verde are another significant factor likely to influence the timing estimates of African/European geneflow. Often severely decimating the island populations. And known to have caused major bottleneck effects. On the other hand due to high birth rates perhaps at times there might still have been a quick recovery in numbers. Census reports often appear to be strongly fluctuating. Still in Laurent et al. (2023) it seems this factor was not (yet) taken into account:

____________________

“Note, however, that while we explored and found, a posteriori, a different demographic regime for each Cabo Verde island separately, with constant, hyperbolic, or linear increases of reproductive sizes, we did not consider possible demographic bottlenecks which may also have occurred as a result of the difficult settlement history of Cabo Verde described above. Such possible bottleneck events will need to be explored in the future, a particularly challenging task given the extensive number of competing models to be considered and given that bottleneck intensities and duration parameters have to be co-estimated with admixture parameters over a very short history of 21 generations.” (Laurent et al, 2023, p.25).

____________________

8) All 3 DNA studies reviewed on this page make reference to the most quoted works on Cape Verdean history. However I find that only Korunes et al. (2022, p.2) provides a decent summary of Cape Verdean history:

• “História Geral de Cabo Verde” (HGCV), various authors (1991, 1995, 2002) (Available Online)
• “Cabo Verde: formação e extinção de uma sociedade escravocrata (1460-1878)”, António Carreira (first edition 1972; second edition 1983; third edition 2000).

I am not quite sure why but Laurent et al. (2023) appears to have seriously misinterpreted these historians. Perhaps because these books are only available in Portuguese? Either way these crucial quotations below clearly contradict several of the key statements/assumptions being made in Laurent et al. (2023). Translated from Portuguese:

____________________

“All this means that Santiago, as a slave depot for export, only operated from 1462 to 1647, in other words, only for 185 years.” (Carreira, 2000, p.174).

“A century later (1727), the population of the island of Santiago is composed of 2% whites, 80% mixed-race and free people, and 16% slaves. The first slave society in the Atlantic world is also the first in which the vast majority of slaves move to the condition of free people.” (HGCV III, p.408).

“Are we seeing a slave-owning [“escravocrata”] society without slaves? It seems to us that the idea of ​​having a slave-owning enclave in the middle of a ocean of free peasants [black & mixed-race] reflects well the social configuration then existing on the largest Cape Verdean island.” (HGCV III, p.18).

____________________

First quote is referring to a royal Portuguese decree from 1647 by which the mainland port of Cacheu in present day Guiné Bissau was given fiscal preference. This practically put an end to most slave trading activities in Santiago/Cape Verde. Which by then had already been in steep economic decline, due to other factors. See also HGCV II, p.273.

During the 1800’s there was a brief return of Cape Verde being used as transit station for TAST. However the scale of this illegal slave trade in the 1800’s was nothing like in the 1500’s. See this overview and also last section of this blogpost for more details:

• Illegal slave trading in the 1800’s: Cape Verde’s involvement (Fonte Felipe, 2015)

The second and third quotes are referring to the situation circa 1731 when according to census the share of enslaved persons on the main island of Santiago was reduced to 16%. See also this overview. The overwhelming majority of the population being involved in selfsustaining rural activities. This because of widespread manumission, fleeing of slaves, socially accepted miscegenation and a collapse of most slave trade orientated activities during the 1600’s. During the 1600’s/1700’s these free peasants would also increasingly migrate to other islands in search of a better life. Not much of this movement was historically documented because it relates to the “common people”…

Two very recent books deal exactly with this topic of how Cape Verde’s society/economy evolved beyond TAST related activities during the 1700’s/1800’s. During the late 1600’s and onwards Cape Verde’s main economic function was arguably more so providing passing trading ships with fresh water, vegetables & fruit. But a clear majority of the population was involved with selfreliant agriculture. This became the most dominant production mode aside from pockets of export orientated sectors (mostly cotton, pano textiles and salt) which did rely on slave labour till the mid 1800’s. But they were insignificant populationwise. Written resp. by a Cape Verdean historian and a Cape Verdean anthropologist:

• “Noite escravocrata, Madrugada camponesa” (A. Correia e Silva, 2022)
• “Cabo Verde: Ilhas Crioulas – Da Cidade-Porto ao Porto-Cidade” (M. Brito-Semedo, 2024)

9) Again I have to emphasize I do think all these DNA studies offer great insights. Which is also exactly the reason I am sharing their main findings on this website. Because I know for the casual reader it’s not always easy to see the forest through the trees of seemingly endless mathematical modelling and fancy graphics ;-). Actually there are also a few other additional things which in my opinion could help to improve future DNA studies in this field. And also widen their readership among potentially many other interested people seeking to Trace African Roots!

• “All models are wrong, but some are useful”. This is a common warning against the unavoidable abstractions made by mathematical modelling. I certainly believe that especially the new approaches to infer the timing of admixture hold great promise. And in fact they are already providing good insights! However generally speaking I do think that DNA studies should be a means to an end and not the other way around.

• Outdated methodologies should preferably be disregarded. And “best practice” should prevail. However generally speaking what I notice is that the tools and techniques applied by some personal DNA testing companies often lead to more robust/detailed results than what is offered in DNA studies. Especially 23andme and Ancestry providemore adequate analysis for regional admixture and IBD matching patterns. At least in my opinion and based on my numerous survey results (see following pages). Obviously not without any shortcomings. But I do suspect that due to source snobbery personal DNA testing results are not always fully appreciated for their informational value.

• In line with the peer reviewers I find that there is a lack of meaningful discussion on how the new DNA studies fit in with previous research. Integration or synthesis of knowledge is naturally very useful for everyone interested in these topics! I was hoping for more elaborate discussion especially regarding any overlap or progress being made regarding the pioneering study of Beleza et al. (2012). I was also missing a recognition of the work done by Stefflova et al. (2011): Dissecting the Within-Africa Ancestry of Populations of African Descent in the Americas. Although based on maternal haplogroups it does also feature a relevant comparison with samples from Cape Verde

The leading authors for Hamid et al. (2021) and Korunes et al. (2022) are relatively young American researchers who started out as computational/evolutionary biologists, investigating fruit flies among other things. Both these studies are based on the Cape Verdean dataset which was collected by the research team for Beleza et al. (2012/2013).

The research team behind Laurent et al. (2023) is more varied. And also includes a linguist of Cape Verdean descent: Marlyse Baptista. Also involved is Cesar Fortes-Lima a very productive Cape Verdean descended researcher specializing in Afro-Diaspora related genetics. As well as African genetic diversity. Much recommended to have a closer look into his other work (see this link). I have actually already reviewed one of his co-authored studies (see this blogpost). 

The main authors behind Laurent et al. (2023) are however from France. In many ways a continuation of their earlier study (Verdu et al., 2017). Apparently much of their research was funded in order to support the Methis software they developed for analyzing admixed populations. Cape Verdeans being used because they provide a good case-study. But in fact this software is also to be applied on a plant species.

• Parallel Trajectories of Genetic and Linguistic Admixture in a Genetically Admixed Creole Population (Verdu et al., 2017)
• Influence of admixture histories on the genetic evolution of hybrid populations – METHIS