Determining half-relationships with Polynesians – Part II

In my last entry I demonstrated the difficulties of determining the half-relationships after receiving the DNA results of my half-first cousin.   Within an endogamous group, that could be even more difficult as we see larger amounts of DNA shared.

While the ISOGG Wiki Autosomal DNA Statistic page can list the average amount of centimorgans shared,  Blaine Bettinger’s The Shared cM Project  demonstrated that the minimum and maximum amounts shared can vary.  This becomes more evident as the distance of relationship increases.

Within an endogamous group it makes sense that having more than one pair of common ancestors may increase that amount.  The same would apply if you descend from the same common ancestor multiple times.  Both would produce higher amounts shared.

A few months ago I got the results of my aunt believed to be a full-sister of my mother.  My aunt suspected that her father was not her biological father.  And she was right.  But she was not the only one who knew of this, but the rest of the family, particularly the ones of my generations believed that this Aunt’s father was her biological father and did not suspect otherwise.

From my mother’s Family Finder (autosomal) match list at FTDNA:

Screen Shot 2016-05-03 at 5.21.34 PM

The top is my mother’s sister while the one right below it belongs to my half-1st cousin whose father George was mentioned in the last entry – Determining half-relationships with Polynesians.

Initially I was confused by the total amount since I knew it was more than what I shared with two of my half-brothers.  This is how two of my half-brothers compare to me and to each other.

Screen Shot 2016-05-03 at 5.47.08 PM

So my mother and her sister did share a bit on the high-end for half-siblings, but low end for full-siblings.  These are the predicted averages shared for siblings vs. half-siblings.

Screen Shot 2016-05-03 at 5.52.23 PM

The next step was to take a look at the X chromosome.  For half-sisters who had the same father, they would share an entire X chromosome based on how the X is inherited.  To my surprise, it looked like someone took a razor blade and sliced out some pieces of the image.


5+cM setting

5+cM setting




For half-sisters they share a lot compared to what I saw when comparing my half-brothers to each and to me.  Also, I decided to include both the default 5+cM setting and the 1+cM.  With my brothers, we hardly get anything when I lower it to 1+cM.  But with my mother and aunt, you can see a difference although chromosomes 4 and 18 are more likely to be IBS, but given the situation (endogamy, small communities, & isolation) it just may be IBD from a very long time ago.

So the X was not helping me one bit since I thought maybe they were areas on the chromosome that could not be read – no calls.

I immediately uploaded to GEDmatch for further analysis.  No surprise that when I looked at the X, it was the same exact thing.   Knowing that it wouldn’t be helpful, I turned to the other 22 pairs of chromosomes.

Screen Shot 2016-05-03 at 6.52.44 PM

What you would be looking for in full-siblings are full-identical regions (FIR) which are the green sections on the bar graph.  Here is an example of my 1st cousins, a brother and sister.

Screen Shot 2016-05-03 at 7.55.59 PMScreen Shot 2016-05-03 at 7.56.08 PM

About 25% will be fully identical.  You can read more about how much full versus half-identical regions siblings would share at ISOGG’s Wiki – Fully Identical Region page.

This is what my mother and aunt showed.

Screen Shot 2016-05-03 at 11.18.11 PMScreen Shot 2016-05-03 at 11.18.36 PM

There are only small chunks of  FIR rather than long segments of it that you would see in full-siblings.  So this confirms a half-sibling relationship.

The randomness of autosomal DNA

Now that Ancestry is able to show how many centimorgans and number of segments are shown, I was comparing my top two closest matches.  They are listed as “lkauhi” and “Frank”.  They are under the 2nd cousin category predicted in the 2nd – 3rd cousin range.

Screen Shot 2015-12-02 at 1.55.19 PM

Prior to my mother getting DNA tested, I had no idea exactly how close they would really be.   Now that my mother got DNA tested and I figured out who my mother’s biological parents were, I was able to construct a diagram.  “lkauhi” is on my grandfather’s side while Frank is on my grandmother’s side.

How Frank & “lkauhi” are related to me.

My mother Judy is a 2nd cousin to “lkauhi”.  That makes me and “lkauhi” 2nd cousins once removed.  While Frank and I are 2nd cousins, because his mother and my mother are 1st cousins.

Here is how much Frank and I share and how much “lkauhi” and I share.

Screen Shot 2015-12-02 at 1.53.14 PM

Screen Shot 2015-11-26 at 10.00.35 PM

224 centimorgans is what “lkauhi” and I share

And although my mother shares 439cM with Frank while sharing 430cM with “lkauhi” (not shown in any diagram), the amount shared seems pretty high for a 2nd cousin.  However, we are talking about Hawaiians whose ancestors have gone through repeated founder’s effect which resulted in our high shared amounts.  She in return managed to pass unto me more of her father’s DNA so that when compared with “lkauhi” we end up sharing more compared to Frank who is in my generation and on my grandmother’s side of the family.  Unfortunately my grandparents are not alive to get them DNA tested for a true comparison.

Also, both of these people have not transferred over to GEDmatch so I am unable to get a better comparison.  This reminds me of what my cousins said about how much I look a lot like our uncle, and that both my mother and I really look like my grandfather Joseph Kaapuiki.  Maybe it is something genetic?  If my mother shared a lot with “lkauhi”, it could be because we inherited more from Elena Kauhi, my mother’s paternal grandmother.

The good thing about all of this is that it confirms that Joseph Akana fka Joseph Kaapuiki was my mother Judy’s biological father, since Joseph’s mother was Elena Kauhi.  And “lkauhi’s paternal grandfather Johnathan Kauhi was a brother to Elena Kauhi.

Both of these closest matches are from each of my maternal grandparents’ side.  Frank is from my grandmother Rose Kanae’s side while “lkauhi” is from my grandfather Joseph Kaapuiki’s mother Elena Kauhi’s side.

Tiny segments from the same common ancestors

Disclaimer: This post demonstrates the use of 1+cM when comparing specific groups of people in order to see patterns of multiple descent from a few ancestors.  It should not be used to validate connections with matches, particularly in this example where connections are beyond a genealogical time frame reaching at least up to 500 years.

Recently I have been comparing both western Polynesian (Tongan and Samoan) and eastern Polynesian (Hawaiian and Maori) matches.  I compared western Polynesians among themselves, and  did the same thing with eastern Polynesians comparing them among themselves.  Then I compared the two groups to each other.

To those who are not familiar with Polynesian origins and/or are new to reading my blog, I will recap.  The ancestors of Polynesians originated from the Melanesia area and thrived there for thousands of years. Thousands of years later a group of “Austronesians” originating from Southeast Asia moved into the area, intermingled briefly and continued to move into western Polynesia where Polynesian culture was born.  At least a couple of thousand of years would pass before they would continue to expand further eastward.  As Polynesians moved from west to east, their genome became less diverse due to repeated founder’s effects and bottle necking.


I analyzed my mother’s results and compared her to a Hawaiian (orange), and a Maori (blue) below.  The Hawaiian is her top match, sharing a total of 693.60cM, longest block 15.52cM, consisting of 158 segments.  The Maori is her 4th top match sharing a total of 517.90cM, longest block 18.08cM, consisting of 119 segments.  FTDNA counts all the tiny segments as low as 1cM once the criteria of a match is met, which is why the number of segments is high.


With the default at 5+cM I did not see anything unusual other than ordinary small segment matches.  But when I reduced the setting down to 1+cM (above), you can see a lot of tiny segments resembling a comb.  The slightly bigger gaps are just the missing teeth of a comb.  Some of these patterns begin to appear at 3+cM, although most do not appear until you reduce it down to 1+cM.  In my mother’s example above I show only chromosomes 1 – 20 since there were no segments that looked like a comb on the other chromosomes.

Then I looked at a Maori woman’s results (below) and compared hers to other Maoris and one Hawaiian.  She also shows the missing teeth at 1+cM, but only in a few areas.  Some areas have the comb pattern while other areas seem random.  The random segments could be IBS (Identical by State) or IBD (Identical by Descent).  Polynesians lack genetic diversity, particularly eastern Polynesians more than western Polynesians, so the random looking segments could be both IBS and IBD segments.


Then I looked at two Tongan men and compared them to other Tongans and Samoans.  With Tongans & Samoans there seem to be more randomness.  A few of the tiniest segments may be close to each other, but nothing resembling too much like my mother’s results, a definite comb-pattern.  Take the purple and green colors for example for this one Tongan man below.  Notice how on some chromosomes they seem to be closer together while on others it just looks random.  Again, these are only using the bare minimum 1+cM.


The other Tongan example.


As you can see, it is hard to look for patterns that resembles a comb, and instead you see random colors all over the chromosomes.  What was interesting to see was how little X these Tongans had.  Unlike with the Maoris and Hawaiians, many of them shared multiple segments with each other.

But what does all of this mean?  These are very small island populations.  They have had repeated emigration from these small islands that resulted in a series of founder’s population.  There there was also bottle necking that occurred a few times.  All of these combined would leave only a few closely related ancestors to populate and repopulate new areas every time.

So the multiple, very small segments that represents a comb with missing teeth is the result of people descending from just a few ancestors who contributed that particular segment, but was inherited from multiple lines going back to the same ancestor over and over again.

Below is an image where I compare my mother with two Samoans (yellow & green) and three Tongans (orange, blue & purple).  There seems to be more randomness, however, there are a few of those comb patterns.


Notice how the X chromosome is much more full, unlike what we saw when comparing the western Polynesians (Tongans & Samoans) among themselves. The yellow color belongs to a Samoan woman. The fact that women have 2 X chromosomes may be the reason why there is a long match versus using two Tongan men whose matches included two women in their examples above.  But these are Polynesians, so you would expect more of a match on the X.  My observance of matches for the past 2 years was limited to only my mother being compared to others, which means I have seen a lot of X matches for her, and the same for myself and my brother.

From what I am noticing so far is that these patterns look like what is mentioned in research papers about Polynesian genome and the loss of heterozygosity going from west to east.  The last place in Polynesia to be settled was in the east, ending at the extreme points of the Polynesian triangle, namely Rapa Nui (Easter Island) in the south east, Aotearoa (New Zealand) to the south west, and the Hawaiian islands in the north.  This explains why my mother and the Maori woman have less random looking tiny segments compared to the Tongans and Samoans.  And if we compare western and eastern Polynesians to each other, we may see some randomness but not as much as we would see with western Polynesians alone.  Other types of Polynesians getting DNA tested would help to exhibit any other additional patterns that I cannot currently see with the majority of Hawaiians and Maoris getting tested.

Loss of heterozygosity – from Western Polynesia to Eastern Polynesia

Genetic research on Polynesians will frequently mention the loss of heterozygosity.  This is more noticeable when comparing eastern Polynesians to western Polynesians.


Map outlining migratory paths of Austronesian speaking populations, including estimated dates. Adapted from Bellwood et al., (2011) “Are ‘Cultures’ Inherited? Multidisciplinary Perspectives on the Origins and Migrations of Austronesian-Speaking Peoples Prior to 1000 BC.” [doi: 10.137/journal.pone.0035026.g001

Polynesian populations are relatively homogenous both phenotypically and genetically. Over a span of 3,200 years they moved throughout the Pacific, and unlike in Europe and other large continents, they did not mix with other populations due to isolation.  These small founder populations have experienced several bottleneck effects, which further caused this loss of heterozygosity ending with the settlement of eastern Polynesia.  Polynesians’ lack of genetic diversity is less evident in western  Polynesia where initial settlement began.  Hawai’i, New Zealand and Easter Island are considered to be eastern Polynesia, and these places were the last places of Polynesia to be settled.

Recently I have been able to look at the autosomal matches among Samoans and Tongans of western Polynesia.  Previously, I have been only studying Hawaiian matches and noticed that top matches were both Hawaiians and Maori people.  Looking at Samoans and Tongans was very interesting as I now could compare the two different regions.

My mother is 80% Hawaiian, while I am 40%.  And as admixed as I am, I still get 1st – 3rd cousin predictions on Family Tree DNA (FTDNA), while on 23andme I get 2nd cousin and 3rd to distant cousin predictions.  The centimorgan totals that I show with my matches reach as high as 369cM on FTDNA, and 161cM on 23andme.  For my mother, 693cM on FTDNA and 376cM on 23andme.  I see the same happening with Maoris, ranging between 300cM – 700cM (FTDNA) for the top 20 people.  And for a non-admixed Hawaiian, their top matches are in the 600 – 700cM range.   An admixed Polynesian would logically have lower totals. But even an admixed person can still have a fairly high amount of totals shared, as when I am comparing myself being less than half Hawaiian.

When comparing two Tongans, the highest that they shared was 335cM.   A Samoan compared to another Samoan was 366cM.  And both of these Tongans and Samoans had their remaining top matches in the range of 100cM to 200cM.  Many of their matches are the same Hawaiians and Maori that match each other at a much higher total.  It is amazing to see these autosomal matches and how diverse the western Polynesians are, or rather how Hawaiians and Maoris are not as diverse.  And even if it is an admixed Hawaiian or Maori, the matches to each other are still pretty high, and as high as what non-admixed western Polynesians would have to each other.

When comparing the longest block (largest segment) with Tongans and Samoans, they seem to rarely get close to 15cM, averaging around 10cM.  Anything more than that could indicate a possible closer relationship or perhaps a specific common geographic origin.  The Hawaiians and Maoris usually range between 10cM – 15cM for the largest segment, but can go as high as 28cM which is usually in admixed Hawaiians and Maoris compared to each other.  In other words, all Polynesians in general will have high totals exceeding 100cM, but whose largest segment rarely exceeding 10cM.

I look forward to more western Polynesians getting tested so we can see if there is any pattern to specific islands in their own island group, something I have been trying to do with Hawaiians with the few haplogroups that there are for Polynesians.  What also needs to be analyzed are people from Tahiti and the Marquesas being that they were key dispersal points for eastern Polynesians.  I managed to only see the results of one admixed Tahitian woman and her match totals are identical to mine when comparing totals.  I am curious to find out what non-admixed Tahitians will show, if it is more identical to eastern Polynesians, or to western Polynesians.

Small segments on the X; male vs. female

Kitty Cooper put out a blog post where she entitled it What Can the X Chromosome Tell Us About the Importance of Small Segments? by Kathy Johnson.   Kathy Johnson had gone through the males in her project and began analyzing and compared to females, determining how much of the females were producing false positives vs. the men.  Because not many men would get a lot of X-matches.  This seems to be an ongoing investigation with various people blogging about the validity of phasing, or rather how effective if not necessarily is it to weed out any false positive matches. It seems to be based on FamilyTreeDNA’s X-matches where they include many tiny segments as little as 1cM.  And the more substantial matches with 10cM or more tends to reduce the actual X-matches significantly, which would be due to the lack of phasing.  You can read more about it on Kitty’s blog, although most of the discussion about evaluating all of these matches took place outside of the blog and on Facebook’s “International Society of Genetic Genealogy” page.

That made me curious, because others have expressed how some men had little to no X-matches.  This was not my situation at all , and went through my list of 9 pages on FTDNA and counted 47 X-matches out of the total 89 matches that I have.  I noticed that one of them was actually an X match on my father’s side of matches, a Filipino.  I knew that was wrong.  So when I looked at it, no X match showed up in the chromosome browser until I reduced the threshold down to 1+cM where I saw a 1.9cM, a false match.

Aside from one woman mislabeled as a male in my matches, I actually have 20 men and 26 females as X-matches, not counting that Filipino false match.  That’s half of my matches.  My mother has 93 X-matches out of her 159 matches, so not that much more than me.  Could that indicate that my mother’s X-matches are more, or less of false matches?  It’s an interesting idea to see how men can have less false matches but we are looking at Polynesian matches which just adds something else to it.

I know that I do have a lot of my matches below 5cM on the X chromosome, so I used’s ADSA (autosomal DNA segment analyzer) to at least look at my ICW (in common with) matches on the X, but I had increased the threshold to 700SNPs and 10cM.

Screen Shot 2014-12-20 at 12.30.11 PM

I was thinking that not only would it be easier to use this tool by instantly seeing my X-matches above a specific threshold, but it would also compare me with others with whom we share the same segments, therefore decreasing the chances of false matches.  But taking into consideration that we are referring to Polynesians.  How would that affect it really?

I cannot determine from comparing my own to my mother’s X-matches if they would be false matches or not. Our problem, lack of documentation, lack of genetic diversity and the unpredictability of the X chromosome itself just to mention a few.

I have recently begun testing my first cousins on my non-Hawaiian side in order to take a closer look at the X chromosome and how that is passed on knowing the X path, that is how it is passed on unrecombined from father to daughter versus mother to children.  I also felt that knowing how it is passed on, it would be easier to distinguish which part of the chromosome was inherited from my grandfather versus my grandmother.  And not until I begin testing relatives from each of my grandparents’ side, I will not be able to fully distinguish all of them with the rest of the other 22 pairs of chromosomes.

Having said that, I cannot see how these X-matches, at least among Polynesians would be consist of a lot of false segments or not.  Especially when there are long segments with the more distant people, e.g., Maoris or Samoans and Tongans, of which I do have X matches with.  But the Samoans and Tongans are not included in the ICW due to the fact that I increased the threshold to exclude anything below 10cM.

I also used Gedmatch’s ONE TO MANY to get all my matches, sorted them by the largest segment on the X and just looked at how many were above 10cM.  There were only 20.

Screen Shot 2014-12-20 at 8.21.57 PM

I did the same for my brother, he got 17 above 10cM.  I also looked at other Polynesian men just to compare and the numbers varied, usually not exceeding 20 with 10cM minimum threshold.  It is still all interesting although it is hard to decipher how much of it is true for Polynesians.  Hopefully as more Polynesians get tested, we will start to notice more differences, or confirm that we just all have a high amount of X-matches.

Not all endogamy is the same

Kitty Cooper’s recent blog post “using Ashkenazi Jewish DNA to find family” talks about how to look for key signs when it comes to finding a true connection to your matches.  I recently blogged about the problems with Polynesian matches and endogamy, just as Ashkenazi Jews will encounter.  I had two different blog entries.  One was “ADSA and Triangulation” where triangulation is used to possibly figure out if you and two other matches can help each other figuring out your common ancestor.  And “Endogamy and Multiple Smaller Segments” where I discovered the actual problem with Polynesian DNA and finding matches.

In Kitty’s entry, she did give tips for finding real AJ matches. Whenever people talk about endogamy, they always bring up AJ as the most prominent group, but the fact is any endogamous group will have its own peculiarity.  I noticed that with Colonial families they have to have what they refer to as “sticky segments”, or segments on a chromosome that basically lingers on for awhile, generation after generation.  I have seen how these segments can begin and end at exactly the same start/end points, which is very interesting.

AJ come from a founding population that started with a small number of people.  Many other endogamous groups started in the same fashion for the most part, particularly colonial families.  With colonial families and Acadians of French Canada, you can see the constant intermarrying within families generation after generation.

DNA Research says that Polynesians slowly moved eastward creating these series of founder effects.  By the time they reached central eastern Polynesia, they were getting genetically less diverse.  They thrived for centuries developing their culture, and then more emigration occurred to the farthest parts of the Polynesian triangle.  The Maori (New Zealand), Hawaiians and Rapa Nui people (Easter Island) are expected to be a subset of the genetic variability in eastern Polynesia, which in turn is expected to be a subset from western Polynesia, which itself is a subset of Melanesia.  A series of founder effects is what lead to this low genetic diversity.

Whenever people talk about endogamy and use AJ as an example when it comes to calculating relationships, or even reading what Kitty Cooper wrote for tips in working with AJ DNA, I keep reminding people that not all endogamous groups are the same, such as the case with Polynesian DNA.  So as Polynesians moved from the west towards the east, and then finally to the most farthest corners of the Polynesian triangle, where Hawaii is at its vertex angle, New Zealand and Easter Island at its base angles, genetic diversity diminishes.

Bottleneck is another feature enhancing the degree of endogamy with Polynesians.  There may have been several bottleneck effects that took place among various island group of Polynesia.  For Hawaii, the last known bottleneck occurred in the late 19th century.  When Captain Cook, a British arrived in the Hawaiian islands in 1778, he estimated the population to be around 300,000.  Scholars will mention anywhere from 800,000 to nearly a million in the islands at the time of European contact.  The 1890 Hawaiian Kingdom census counted 40,622 aboriginal Hawaiians.  The 2000 US Census had counted 401,000 Native Hawaiians.  So the current Kanaka (Native Hawaiian) population comes from the 40,622 that existed 124 years ago.

When thinking about the lack of genetic diversity given the entire history, it should not be surprising that, as in my case being Hawaiian, that not only will I have many close matches with other Hawaiians, but with other Polynesians too.  More specifically what we have been seeing, is a close genetic relationship with the Maoris.  That is understandable since both of our groups were the last places to be colonized by Polynesians.

So what does that mean for Polynesians working on their DNA matches?  It is okay to read about other methods that endogamous groups use to find their matches, but be aware that we have much closer matches unlike other groups.  If you are on FTDNA, you will find a lot of 1st – 3rd and 2nd – 3rd cousin matches.  I get 3 pages of 2nd – 3rd cousin matches while my mother has about a page and a half of 1st – 2nd cousins, just over two pages of 1st – 3rd cousin and five pages of 2nd – 3rd cousin matches.  At 23andme, I get 2nd cousins, and 2nd – 3rd cousins and two pages of 2nd – 4th cousins.  This is what to be expected, and again a true closer relationship will be distinguishable by looking at the number of segments.  We may share a lot when it comes to total centimorgans in order to get 2nd – 3rd cousins, but a real 2nd to 3rd cousin match should not have as many segments and these multiple segments will average anywhere between 8cM – 15cM.  This means that the match is endogamous.

When doing triangulations you will see that with your matches there is a fair amount of shared segments with other people of whom will share that matching segment.  With non-endogamous groups, you need to first verify that both of your matches are sharing that same segment with each other in order to determine that you all have a common ancestor.  For Polynesians, this is often the case, and probably descended multiple times from that same ancestor.  That may seem significant and on the right path for finding a connection however its extremely low genetic diversity coupled with the fact that many records did not exist until recently usually produces no results.