DNA and Barrow Lousada genealogy

Our first exposure to DNA technology was when our distant cousin Michael Waas suggested some male-line Lousadas be entered into a study of male-line Sephardic descent. This study was drawn upon in early 2020 by his collaborator Adam Brown when he covered the use of Y-DNA data in Jewish genealogy - paper uploaded here (see also ref 306). Allan and Jeremy Lousada duly proved to be extremely close Y matches with a rare haplotype E-FT 333503 (derived somehow from the North African haplogroup EM35) while Bob Leuzarder (see below) was not a match (his haplotype is R-BY 193173 of the R1b haplogroup which is common among European males; the genealogical significance of this result is discussed below). On 21 Oct 2021 Adam Brown advised us of the next closest Lousada Y match - from Finland. This match has 33 out of 37 STRs in common with Allan Lousada; this is 4 genetic steps or about 300 years (9 to 12 generations) from birth of a common ancestor but the genealogical picture is currently inadequate to explain the match other than that a Finnish great-grandfather was 'Mediterranean'. Adam Brown also, in the online discussion group Sephardic Diaspora during 2019, referred to the Ashkenasi founder effect (see here) and to the likelihood that Randy Schoenberg does not descend from the small number of original founders of the Ashkenasi community. But when Ed Barrow got his Y-DNA results back they placed him in the R-Y19847 haplogroup, and things got interesting. For as Michael Waas observed 'I briefly looked over the results and I can tell you that Ed belongs to one of our founding Ashkenazi lineages that also has been in the Jewish people probably since the beginning. I have brought Adam Brown in to the discussion for further information and detail. I didn't see any matches necessarily that descend from Simon Michael Pressburg; of course, if you know any direct Y descendants of him, we could test that theory..'. Ed's matches point to Myers and Wertheimer ancestors, and these families are present in the circle of Simon Michael Pressburg (see here).

 In the meantime, John Griffiths, descendant of Joseph Barrow, has proved inspirational in raising our interest in DNA. In late 2019 he compared his and Julian Land's autosomal DNA data on GEDmatch. It was exciting to experience this first match but it was insignificant (a single 3.4cM segment match on our 8th chromosome) and much less than might have been expected for 6th cousins (common ancestors Simon Barrow 1709-1802 and Bailah Montefiore 1720-73) but probably illustrates how drastically autosomal DNA can be discarded. Randy Schoenberg provided access to his DNA records, and a number of matches with Randy Schoenberg were found. Further work on our Schoenberg matches (see here) was necessary to clarify the situation as discussed below. The de Symons connection might be expected to give single and double descendants of Simon Barrow of Bath a greater link to Randy Schoenberg and his ancestor Simon Michael Pressburg compared with those Barrow descendants like John Griffiths who do not have this connection. However, the situation is not quite so simple (see here) and our work raises the possibility that the grandson of Simon Michael Pressburg - Baron Lyon de Symons - though the legal father of the wife of Simon Barrow of Bath may not be her biological father! Julian Land and Michael Waas compared their autosomal DNA and as Michael reported on 25 Jan 2020 'we do have some very small 3cM segments shared between us. But who knows if that's from the Portuguese or Ashkenasi sides of our families'. Indeed, we have no certain knowledge of common ancestry - corresponding to our increasing certainty that Abraham Israel Pereira (an ancestor of Michael) did not have a Baruch Lousada mother but rather became by marriage a nephew of Isaac Baruch Lousada #42.

Some Fischl descendants made contact after seeing our Dirsztay Family Tree (ref 31). Dena Jenkins, with Fischl ancestry via Lajos Fischl 1798-1856, found on FamilyTreeDNA that 'hooray we have the same 5th cousin. I found him in your kit and then went on my cousin's kit ... we both have him ... If we both have the same 5th cousin we must be related!' Though it is hard to see how this match with Dena Jenkins arose other than via Barrow ancestors, we were thankful that Scott McDougall's wife, with Fischl ancestry via Bela de Dirsztay 1861-1921, had her DNA results uploaded to GEDmatch. These became part of our set of data from probable relatives the analysis of which is discussed below. We are 7th cousins according to our Barrow/Fischl suggestion, with Simon Barrow's Baruch Lousada mother a common ancestor of her (and Dena Jenkins). A subsequent DNA match supports us here as it not only brings an additional Fischl link, but an additional Pressburg marriage link as well. In early 2020 Bob Leuzarder introduced some immediate relatives and also a DNA match also from the USA - Jeannine Wegmueller, a descendant of the Luzarders. It was not clear how the Leuzarders (see here) and the Luzarders (see ref 314) related to each other nor to the rest of us. With genealogical data so hard to find, autosomal DNA could not be dismissed, even though few people elsewhere bother with small segments of the size we were left with.

John Griffiths suggested trying to see whether segment match correlations could show underlying connections between relatives since individual matches at 3cM were unreliable. (His work showed over 200 DNA segments arising from a study group of 11 possible family members; John's last report #4 is here, and his observations on this are here). A later inspection by Julian Land of John's work - namely of the multiple segment matches arising from a subset of 7 of the relatives (Ju, JG, SW, Ed - a Barrow 4th cousin, Je, MD - a Lousada 6th cousin, and MW - though the latter was later excluded as a relative as noted above) - is reported here with just 2 of 133 superficially intriguing multiple matches surviving scrutiny to ensure that these 2 were stronger than the types of multiples a random sample threw up. This is potentially a tough test in that there may be unknown genetic relationships within the random sample closer than the Baruch Lousada relationships we are seeking to confirm, and in any case the random sample threw up surprising multiples (including both types of what we refer to as strong triples - 3 people matching each other, and 1 person matching 3 others - neither conclusive in themselves). The chromosome 8 match passed easily, and the chromosome 2 match less so (but having both types of strong triple with a connected match it was stronger than the strongest random multiples in this study). In an inkling of what was to become a crucial part of our analysis, the Cr8 match provided our first encounter (at 52269392) with an improbable type of segment boundary coincidence (its statistical likelihood is discussed below). The matches linked us to the Fischls and to the New England Lousadas and thus simultaneously supported our Barrow/Fischl suggestion, and encouraged us to hypothesize the descent of both the Luzarders and Leuzarders from Jacob a NY chocolate merchant (with Jacob assumed to be a great-grandson of Amador de Lousada of Vinhais via Amador's son Fernando and Fernando's son Henrique who was in Vinhais until 1650). Bob and Jeannine now appear as 4th cousins once removed, but based on the genealogical situation of Bob's great-great-grandfather Benjamin (likely to have been the adopted son of a great-uncle also named Benjamin) we proposed that Bob was not a male-line Lousada despite his surname; this was remarkably confirmed in 2023 when Bob reported a Y DNA match with Ernest Lloyd Luzadder; this match was of the R haplotype (that is, not the Baruch Lousada E haplotype, as noted above). This match links Bob like Ernest Lloyd to the great number of Midwest Lousadas who all descend from 'Aaron2' the namesake but not the son of Jacob's son 'Aaron1'. Nevertheless we considered whether Ernest Lloyd might have Baruch Lousada ancestry transmitted through a female line.

We therefore updated our 7-relative study to include Bob and Ernest Lloyd while excluding MW (see here). This led us to increase the number of our candidate Lousada indicator segments from 2 to 4:

Thus our structure of the genealogy of the USA Lousadas places Ernest Lloyd Luzadder with most of the descendants of 'Aaron2' outside the Baruch Lousada genealogy (having none of the indicators), and on the other hand places Bob (indicators Cr2, Cr5 and Cr21), Jeannine (indicators Cr2. Cr5 and Cr8) and a Papacin-Lousada descendant (indicators Cr2, C5, Cr8 and Cr21) within the Baruch Lousada genealogy. Bob appears in both genealogies! The Papacin-Lousada descendant was only discovered by Bob Leuzarder after our 2023 8 by 8 study but because of her probable Lousada genealogy (descending from the older Benjamin - see above) her data was then included together with 2 other known Lousada descendants (Allan and Jeremy - respectively 1st and 4th cousins of Julian Land) in our consideration of the Schoenberg matches referred to above; matches appear especially at the Cr2 indicator (where using Qmatch at 3cM with P=3 we found 3 of the improbable segment boundary coincidences first discovered as recorded above and as discussed below), but also at the Cr8 and Cr21 indicators. This demonstrates that Randy Schoenberg has a Baruch Lousada genetic link and raises the question of whether it is discoverable genealogically. Further work with the complete set of our probable Baruch Lousada descendants (a total of 12 as at July 2024 now including Randy's mother and father) failed to discover any indicators distinguishing those with Barrow genes compared to those without. In Feb 2025, using the complete set of 12 probable Lousada descendants, and using Qmatch (3cM, P=7), we found that the increased number of single matches gave no enhancement of the inconclusive Ernest Lloyd multiple matches found with 7 Lousada descendants (Ju, JG, SW, B, Je, Ed and MD) on Cr2, Cr6, Cr12 and Cr13 as reported in our 2023 8 by 8 study, while creating 2 new inconclusive strong triples - on Cr1 (with none of the 5 new Lousada descendants) and on Cr21 (with 2 of the 5 new Lousada descendants). The 5 new Lousada descendants (RF, RM, Allan, J and TP) in fact generated only 19 additional single matches which is about half the rate of the older 7 who produced 49, which helps us understand their low effect on Ernest Lloyd's situation. The 3 strong ELL triples on Cr1, Cr12 and Cr21 showed inconsequential results when examined at P=3. This all confirms our exclusion of Ernest Lloyd from the Lousada genealogy. Using Qmatch (3cM, P=3) with all 12 probable relatives, we re-affirmed the significance of the Cr8 indicator via 19 matches and 2 of the improbable segment boundary coincidences (see below), but we found the Cr5 indicator much less compelling and note that it played no role in the Schoenberg study just discussed and that it has no improbable segment boundary coincidences (so it is merely a type of multiple one can get with a random sample and by no means the strongest at these settings - cf the 2nd Cr1 multiple shown here). That is, we now pay no attention to it, and observe that its absence does not suggest any change to our conclusions about Edward Lloyd Luzadder because the other 3 indicators remain functional. Naturally, we also reaffirmed the Cr21 indicator via 23 matches and 2 of the improbable segment boundary coincidences (see below). At the CR2, 8 and 21 indicator sites each Lousada descendant is present in matches with some of the other 11 (Julian 17 times, JG 12, SW 13, Ed 15, Allan 12, RF 4, RM 11, Bob 10, Je 13, MD 4, TP 14 and Jeremy 9). Our 7 improbable coincident segment boundaries (3 at Cr2, 2 at Cr8 and 2 at Cr21) involve 11 of the 12 (only Bob missing). Though useful in our work, the indicators are not straightforward to use - thus, the absence of an indicator does not imply the tester is not Lousada and neither does the presence at an indicator prove Lousada descent (see here for an example).

Some instructive methodological points emerged. In our 7-relative study we included a sample of 7 nominally unrelated people so that we could contrast cousin-cousin matches with unrelated-unrelated matches. For this work, with a 3cM threshold, GEDmatch advised that 'segment threshold size will be adjusted dynamically between 200 and 400 SNPs' whereas this changed in our later 2023 8 by 8 study to 'segment threshold size will be adjusted dynamically with an average of 200 SNPs. About 2/3 will occur between 185 and 214 SNPs'. Initially we noticed that the newer GEDmatch settings increased the number of matches between cousins as compared with matches between unrelated people, to the point where we got 5% more cousin-cousin single matches than unrelated-unrelated single matches (not fewer! - as with the prior GEDmatch setting). This 5% 'signal' is of course small amid the 'noise' of all those pre-genealogical matches and/or false positives and suggests why we should be very critical in accepting coincident multiple matches as indicating relatives and indeed why there are so few of value. But we noticed that, with the newer GEDmatch setting, the number of cousin-cousin coincident multiple matches increased strongly - by twice the increase in unrelated-unrelated coincident multiple matches (2 times cf 1.4 times as reported in our 2023 8 by 8 study) from which we can see why our version of John Griffiths' broad-brush analysis (of multiple coincident matches between relatives) had became more productive. GEDmatch advised that its Qmatch technique should give further improvement when looking at 3cM segments as we must with our 9-generation separations. But, even with the GEDmatch improvements, samples of unrelated people will continue to generate coincident multiple matches, and accordingly we continue to need ways of distinguishing useful multiple matches from pre-genealogical ones and ones based on any remaining false positives. We were lucky enough to find a specific type of overlapping  segment boundary coincidence - a boundary SNP shared by 2 distinct pairs of probable relatives (that is, comprised of 4 separate individuals) - which is statistically very unlikely (around 1/7,000 based on rule-of-thumb estimates for our circumstances - 200 SNPs per segment, 30 segments per match, 1000 segments in the genome); it thus shows probable genetic connections between the 4 relatives despite those connections not all meeting the criteria for being called matches at the settings chosen. We found such an improbable segment boundary coincidence 3 times with our Cr2 indicator after finding our very first one with the Cr8 indicator (a number now revised to 2 as indicated above - these are 2 new cases as the original finding at 52269392 is strictly only present at these Qmatch P=3 settings in the form of a much more likely type of segment boundary coincidence with probability 1/4 and which of course also occurs readily in samples of random testers - though the original segments have extremely adjacent boundaries at 52259764 and 52267567), and 2 with our Cr21 indicator. GEDmatch did not offer a comment on our pointing out the significance of these improbable segment boundary coincidences, perhaps not because its segment boundary definition has an arbitrary element (for a segment boundary can be defined by the last SNP in a segment or the next SNP outside the segment) - for if the definition is consistently applied our logic is unaltered. But GEDmatch's reticence is understandable, for as was seen, a change in settings had an effect not only on whether a match is called a match or not, but where the boundaries are shown! Our good fortune lies in the fact that our type of improbable segment boundary coincidence occurs at one or more of our indicator sites under any of the GEDmatch settings regimes we used, and this reinforcement is why we can be confident about our indicators.

It may be of value to comment on 2 further types of segment boundary coincidence which came to our attention. These 2 further types were discovered during a re-examination of Bob's strong triple (he matched ELL, SW and Je) on Cr2 which was one of ELL's few potentially interesting matches arising in our 8 by 8 study. As we report here the 2 further types are associated with contiguous segment matches - not overlapping segment matches - and as such do not appear to support multiple matches (and indicator segments). The first type at 8099955 shows a J/B segment match to the left and an abutting J/JG segment match to the right. Here J controls both the upper end of the J/B segment match and the lower end of the J/JG segment match. That is, JG extends below 8099955 and B above 8099955 - a 1 in 4 chance. Another occurrence has been observed. The second type at 8442248 shows an A/J segment match to the left and an abutting B/JG segment match to the right. It seems that Allan, Jeremy, Bob and John all retain in their DNA a record of an ancestral recombination at Cr2 8442248 - probably from procreative activity by Pedro de Lousada and Briatis Alvares. Alan and Jeremy show one side of the recombination with Bob and John show the other. These 4 descend from different sons of Amador de Lousada as follows:

• Allan - David #1584, Isaac #42 and/or Abraham #2419

• Jeremy - same possibilities

• Bob - Fernando #1785

• John - Isaac #42 and/or Abraham #2419

One of Allan/Jeremy and one of Bob/John each then had a subsequent recombination at 8442248 meaning that each of their segment matches terminated there - making for the Allan/Jeremy segment match an upper bound and for the Bob/John segment match a lower bound. This all seems unlikely - perhaps more so than our improbable segment boundary coincidence discussed above. However there is little doubt that Allan, Jeremy, Bob and John are related within the framework described.

In Nov 2021 MyHeritage notified Julian Land of an 8*GGF who had been in Recife namely Aaron Querido #3161. This ancestral contribution enters via Baron Moses d'Aguilar whose wife Simha da Fonseca was a great-grandaughter of Aaron Querido as shown on geni.com based on work by Jarrett Ross. However Bob Leuzarder shares a 7cm DNA match with a person who descends from Aaron Querido a great-nephew of the just-mentioned Aaron Querido. This match appears to arise from Esther #1768, one of the Den Haag Louzadas, and perhaps Bob Leuzarder's match is the 1st present-day descendant of the Den Haag Louzadas we have encountered. Esther cannot however have been a direct ancestor of Bob Leuzarder, who has a different descent from Amador de Lousada who was Esther's 2*GGF. Remarkably Bob's sister Christine does not share the match, so the relevant genes were discarded from her DNA in the previous generation. Julian Land does not share the match either, but here the relevant genes may have been lost many generations ago.

On 10 May 2021, FamilyTreeDNA reported an interesting match Julian Land has with 3 successive generations of the Nunes Vaz family. Using the obvious point of linkage evident in Nunes Vaz ancestry namely Simha Henriques Faro #1629, we could not resist suggesting an earlier linkage between the Henriques Faros and the Baruch Lousadas than was deducible from the paper records. This was not entirely impetuous - we had been looking at ways further Henriques Faro links could have arisen. Jarrett Ross on 26 Sep 21 cautioned not only that there may be other Sephardic elements in the match, but that an Ashkenasi element is important as well because his GM (the 1st of the 3 generations referred to) is Ashkenasi! We interpret the matches as follows. Julian's match with the GM has been halved (as judged by the largest segment being halved) by the GM's marriage with a Nunes Vaz descendant. However, because total match was similar in the next generation, the missing Ashkenasi matching genes have been more or less replaced by matching Nunes Vaz genes. That is, there is a Sephardic match with the Nunes Vaz GF roughly comparable to the Ashkenasi match with the GM. In the next generation, the combined match has been largely preserved, probably because the matching DNA has largely survived (as judged by the largest segment not being much reduced), with only a small introduction of new matching DNA. While our original interest was the Sephardic match, the original Ashkenasi match is as close or closer than Julian's other Ashkenasi matches, and we explore it by looking at 3 common Ashkenasi matches but as summarised here, we can see some patterns but nothing of statistical significance.