Largest Segment – Is it the best way to gauge the closeness of relationship?

In my earlier blog posts I have mentioned how significant the largest segment size is when determining a true 2nd to 3rd cousin relationship.  Polynesians can have a total shared amount that can easily exceed 100cM.   These totals tend to over-estimate the predicted relationships.

From the ISOGG Wiki’s page, you can see that the average for 2nd cousins once removed (2C1R) is 106cM while 2nd cousins are averaging around 212.50cM.

So we tend to get a lot of these 2nd – 3rd cousin matches, depending on the company you tested with.  This is why the largest segment size has become important.  Blaine Bettinger has a post entitled The Shared cM Project – Longest Shared Segment where people had submitted their longest segment size based on their known relationships.  You can compare 2nd and 3rd cousins there and see what the average is for the longest segment size for specific relationships.

A quick look at the type of numbers just by looking at my own ONE TO MANY from

My cousin Allen who is a 2C1R to me (his maternal grandmother & my mother are 1st cousins) has a large segment of 35.9cM.  You can see more comparisons of the largest segment for 2nd to 3rd cousins from Blaine’s Shared cM Project but I also have been keeping my own numbers from my known relatives.

Only one of those 2nd cousins shared a large segment of 21.8cM, pretty small, and then it gets even lower as you go more distant.  But normally 2nd cousins will share a rather large segment, which is why more than 20cM has always been advocated and also among the Ashkenazi Jewish community.  In fact, I thought they used 25cM, but I could be wrong.  I even mentioned 30cM would be good.

But is it a requirement?  Absolutely not.  However, if you cannot find a connection, or the same geographical origin i.e. New Zealand or Hawaii, then that would be a strong indicator that you are not as closely related as it was predicted.

I have been noticing how I do have a few Hawaiians whose largest segment is more than 30cM but have not been able to find a connection.  I also notice that these matches will not have the same geographical origins as I do.  So could it be that these large segments remain in our population for many generations?

Here’s an example of how it actually has remained for centuries by comparing my Hawaiian mother and a Maori.

Taking my mother’s ONE TO MANY matches, I sorted them by the largest segment size.  I indicated the known relatives in blue and the unknown in red.  My mother has a Hawaiian match as 44.6cM for the largest segment.  I still have not been able to find a connection, although one of that match’s branch goes back to the area of a few of my ancestors.  But even for us, that was more than 3 generations ago from my mother.  Another at 39.9cM, not sure if that person is a Hawaiian or Maori.  And there is a Maori match with the largest segment of 25.9cM.   At FTDNA, there is a Maori match whose largest segment is 23cM.

Here is the largest segment sized match with a couple of Hawaiians from MyHeritage.

37.2cM and 33.6cM.  They have pretty good trees but their ancestry goes back to totally different islands from my own ancestors.  And I saw in their trees the origins of the different islands is further back while the more recent ones were born in Honolulu where some of my more recent ancestors were born.  I did trace many of my ancestors’ descendants who remained in Honolulu but none are those connect to these matches.

Here is a Maori match from MyHeritage.

Notice that the largest segment is 34.2cM.  The highest I’ve seen with a Maori.  How can a large segment last that long after many centuries?

And while the focus here is utilizing the largest segment to get a more accurately find a true 2nd to 3rd cousin match, we know how in one generation a large segment can quickly be reduced.

Comparing with the largest segment that my mother shares with her 1/2 3C.  This is how they connect.  I outlined in yellow all testees in this particular comparison.

The largest segment that my mother & her 1/2 3C share is 49.6cM (FTDNA indicated 52cM) according to GEDmatch.  But that particular segment was not inherited entirely by my mother’s sister and seemed to have been broken up thanks to recombination and turned into a 10.7cM and a 25.1cM segment.

My mother’s deceased brother seemed to have received that same segment or maybe even slightly larger.  And while he is not alive to test, his son did, and he shares 50.2cM with this 1/2 3C of our parents, or our 1/2 3C1R.

This is what the comparisons look like.

My younger brother got nearly the entire segment as my mother got it but I got a very small portion of it, just 14.3cM.  That’s a huge difference from 49cM.  Had my mother nor my younger brother got tested, I would not have been able to find this good match and would have concentrated on matches with large segments more than 20cM or even 30cM.  My older brother got DNA tested however he does not share any of this same matching segment.  In fact, he shares 0cM on this particular chromosome.

This 1/2 3C was key in finding my mother’s biological parents.  At the time I did not know how we were related but I did concentrate on this match because of the large segment size.

So how do we really filter all of these matches?  By solely concentrating on the largest segment?  You should definitely not spend too much time on large segments that are less than 20cM and whose shared total is way over 200cM.  With those particular matches, if you compare trees and notice no common geographic area, that would be a big indicator that it is a distant match.

Remember that with a 2nd cousin you would share a pair of great-grandparents.  With a 3rd cousin you would share a pair of 2x great-grandparents.  By that generation or even a generation further back or two if you find that you do not share the same geographic location, then the match is a distant match.  The same applies for large segments greater than 30cM.  If no common geographic location, then it is probably a distant match.

MyHeritage Ethnicities

As of May 30, 2017, MyHeritage finally released their Ethnicity Estimate (beta) to those who uploaded their raw data.  So far this service is still free.  Not sure if they will discontinue that service.  Currently their tests are at a reduced price of $79.


Not only does MyHeritage (MH) have an Oceanian category but they included Polynesian along with Melanesian and Papuan.


Last year and probably the year before that, they reached out to people who had a tree at MH whose 4 grandparents were listed in a given geographic area confirming ties to that particular place or country.  And while they seem to have obtained more than Ancestry’s 18 Polynesian samples, they did not take into consideration that these people may be admixed.

A lot of admixed Polynesians who did test with MH are reporting to have lost a lot of their European while simultaneously having an increased percentage of Polynesian.  There seems to be about 10% difference.

Here are my mother, my maternal aunt and my own results.

Click for larger image

My mother and her sister are 85% Hawaiian while 15% is of European background.  My mother gets about 17% European at the varying DNA testing companies.

Several Polynesians have shared their Ancestry results with me.  Comparing it to MH it seems that the numerous samples that they used for the Polynesian category included some admixed Polynesians of European heritage.  I have been hearing the same situation for those with admixed Native American background reporting 20% to 30% more Native American while reducing the amount of European.

What is interesting about MH is that they did have other populations not covered by the other testing companies.  They separated the Melanesian and Papuan, commonly grouped together and labeled as “Oceanian” by other companies or at GEDmatch, and provided a separate Polynesian category.  They did something similar for the Asia group.

MH has specific groups within the Southeast Asia area, such as Filipino, Cambodian, Vietnamese, Thai, Malaysian and Indonesian.  While the thinking is that if you come from that particular background which they tested, you should score perfectly with that group.  This may not apply for some either due to the limited number of samples and/or where they got their samples from, such as taking samples from one specific area.  I have only seen a few Filipinos’ results where they score 100% Filipino/Indonesian/Malaysian.  A couple of Chinese people received majority Chinese and Vietnamese and a smaller percentage of Filipino/Indonesian/Malaysian.

My guess is that this breakdown of the various Southeast Asian groups helps separate Polynesians (and Micronesians) who also have some Southeast Asian background.  At Ancestry, Chinese people were reporting about 10% Polynesia, Vietnamese as high as 15% and Filipinos around 32%.  Ancestry has no Southeast Asian category, so those of Southeast Asian background will get some of the East Asian, or what Ancestry has as “Asia East” along with a small percentage of Polynesia.

Aside from MH engulfing the European for admixed Polynesians, it seems fairly accurate at least for me being that I am half Filipino and 43% Hawaiian and about 12% European background.  I am going to assume that the West Asian below is part of my European background while the South Asian is part of my Filipino backgroundc

Confirming what could have been a NPE (non-paternal event) or misattributed parentage

Another useful tool for DNA testing is to answer those questionable paternity that either was brought up by a family member or documentation may not support what is known.  This was one of the main reasons why I got DNA tested in the first place.

Quite a bit of people getting DNA tested are finding what is known as an NPE (non-paternal event) or a misattributed parentage.  That is when the presumed or putative father was not the biological father.  This could have happened either recently, a generation ago, or way beyond that to where current living people may not be aware.

This is when people need to take the extra steps by testing other family members or also getting other specific tests, such as a Y-DNA test. Sometimes it can be a Y-DNA test that makes people realize that there was an NPE.

Back in July of 2015 I figured out who my mother’s biological mother was.  Her name was Rose Kanae, and Rose was married three times.  I found that one of her husbands — Joseph K. Akana  resided at the same address where my mother was born.  So the assumption was that he was probably my mother’s biological father.  The  Akana surname is of Chinese origin, and it is what initially made me believe that he was not the biological father.  My mother was told after having met Joseph Akana once as she was 5 years old, that he was a pure Hawaiian man.

Last October a cousin confirmed that Joseph indeed was my mother’s biological father.  It was explained to me by a couple of relatives that Joseph took the surname – Akana from his Aunt who married a Chinese man surnamed Akana.  Joseph’s original name was Joseph Kaapuiki, and later he went by Joseph Kaapuiki Akana.

This same cousin who confirmed that Joseph was my mother’s biological father did question Joseph’s paternity, suggesting that Joseph’s mother Elena Kauhi was not so faithful.  This is how I was able to confirm that Joseph’s father – John Kaapuiki was his biological father.

Below is my mother’s top 5 matches.Screen Shot 2015-12-17 at 5.18.58 PM

These all say “Possible range: 1st – 2nd cousins.”  Her first match is how I was able to figure out who her biological mother was.  This is how Frank is connected to my mother.

Screen Shot 2015-12-17 at 5.23.11 PM

Frank and my mother are actually 1st cousins once removed, making Frank & I second cousins.  With females there is less ambiguity whereas with men there can always be that questionable paternity.

The second top match was “lkauhi” and this is how that person actually is related to my mother once I was able to get my grandfather’s genealogy.

Screen Shot 2015-12-17 at 5.20.32 PM

“lkauhi” is off to the right, and she matches my grandfather Joseph Kaapuiki (Akana) via his mother’s side, through Elena Kauhi.  This would confirm that Joseph is the biological father of my mother since “lkauhi’s” grandfather Johnathan and Joseph’s mother Elena were brother and sister.

One of my cousins gave me the names of our grandfather Joseph Kaapuiki Akana’s ancestors going back as far as his grandparents.  His father John Kaapuiki‘s father was Kukahuna Kaapuiki.

Further research online revealed that the Akana-Kaapuiki family listed my ancestor Kukahuna and traced it a few more generations back.  But I was not confident at first to know that any of the names beyond Kukahuna were my own ancestors.  This is the same family that I was told my grandfather Joseph took his surname from, and that they were related.  Given that they listed Kaili Kaapuiki who married a Chinese man surnamed Akana as the sister to my ancestor Kukahuna Kaapuiki, I knew that was probably the connection but could not confirm it through documentation.

I looked for the genealogy of my mother’s 3rd match “milt17th.”  I contacted him and he confirmed his genealogy, that he was the grandson of Kaili Kaapuiki and Akana.

This confirms that John Kaapuiki was the biological father of my grandfather Joseph Kaapuiki Akana.

East Asian category for Polynesians

My mother told me today that she received from the First District Circuit Court that handled her adoption, the non-indentifying form, which is where it lists her biological parents’ ancestries.  They indicated that both parents are Hawaiian and Chinese.  I find that to be an error since my mother had her DNA tested at 3 companies.  Maybe that was based on an assumption or the biological parents may not have known too much about their ancestries.

At the age of 5, she met her biological father and described him as a “pure Hawaiian.”  This made sense since she gets the following percentages from each company.

Screen Shot 2015-07-13 at 3.25.31 PM


So this meant that one parent was just Hawaiian while the other parent was admixed with some European.  Based on all the companies she has tested with and analysis [from Dr. McDonald], my mother gets the following averages.

European = 16%
Oceanian = 25%
East Asian = 55%

Those are based on 23andme, FTDNA’s old Population Finder and Dr. McDonald’s analysis.  FTDNA’s current myOrigin lumped their former Oceania category under Southeast Asia, or the more broader East Asian category.  AncestryDNA however created a Pacific Islander category with the subgroups Polynesia & Melanesia and between myOrigins and Ancestry, the average total is 83%.

In reality, the East Asia category is just one of two components that make up the genome of Polynesians.  The other is Melanesian/Papuan (Oceania).  For Polynesians, autosomally they are 79% East Asian and 21% Melanesian1.

In Population Genetic Structure and Origins of Native Hawaiians in the Multiethnic Cohort Study they applied ADMIXTURE on genome-wide SNP data to finely estimate the degree of admixture in Native Hawaiians.  They found that with Native Hawaiians, “an average of 32% and 68% of their genomes to be derived from Melanesian and Asian origins, respectively”.  But that “[r]ecently, Kayser et al. surveyed the nuclear genome with 377 microsatellite markers in 47 Pacific Islanders and identified 79% Asian and 21% Melanesian proportions of ancestry for Polynesians.”

So while other studies revealed that Polynesian genome consist of 79% Asian and 21% Melanesian components, the study with Hawaiians averaged 32% Melanesian and 68% East Asian2.  The higher amount of Melanesian could be attributed to the repeated bottleneck effects throughout the centuries specifically for eastern Polynesians, i.e. French Polynesians, Rapa Nui, Cook Island Maoris, Maoris from New Zealand and Hawaiians.

My mother averages 25% Oceanian and 55% East Asian.  The two combined equals 80%.  25% (Oceanian) is 31% of the entire Polynesian (80%) percentage.

Just over a year ago I was going through my mother’s matches on GEDmatch and began running their kits through various admixture tools to see their totals of Oceania versus East Asia.  I wanted to see if they fell within the 27% to 32% Melanesian/Papuan/Oceanian.

Screen Shot 2015-07-13 at 5.20.34 PM

This is only a partial list.  There are more lines on the bottom but I am just showing the first several.  I use various calculator admixture tools at GEDmatch that had the specific Oceania/Papuan/Melanesian category.  Dodecad World9 uses “Australian” instead. I created a column that totaled the average.  And the columns to the right of that shows those who tested at other companies and their Oceania percentages.  What is not shown in the list is the proportion of East Asian that would prove it does fall within the 27% – 32% Oceania.

The first row (in red) belongs to my mother, followed by my averages and then my brother’s.  The following lines in bold are for those at least 95% Polynesian.  I took the lowest and the highest percentages of Oceanian to see if it falls within the range consistent with the research.  Since these admixture calculators did not have just a single East Asia category alone, I listed the other categories that are known to split off from the East Asia category.

Eurogenes K9b
Oceania = 17.74%
Southeast Asian = 48.87%
Northeast Asian = 13.32%
Native American = 1.55%

Adding the Southeast Asian, Northeast Asian and Native American categories total 63.74%.  17.74% Oceanian makes up 27.8% of the total (East Asian compiled categories) of the Polynesian genome.  Consistent with the research.  Looking at the highest percentage.

MDLP World
Melanesian = 20.78%
East Asian = 56.31%
Artic Amerind = 1.65%
Mesoamerica = 0.13%

There was an Indian category showing 3.74%, but I did not add that in.  Adding it, changed the overall percentage to 33%, but leaving it out made it 35.7% of the Polynesian portion. For the others listed on that list, they also picked up a small 1 – 3% of the Indian category, and leaving it out made their average 30%.  If I look at the average column for all the admixture calculators for my mother, it comes out to 30%.

But what happens when there is a higher percentage of East Asian?  In my case, it is higher because my father was Filipino. My portion could easily be verified simply by removing 50% (my father’s contribution) from my average total of 85% East Asian giving me 35% East Asian that would be my Hawaiian/Polynesian side.  My average (GEDmatch) showed 32%.

There are many Hawaiians admixed with Chinese, Japanese, Filipino or Korean, being that they were all immigrant groups to the Hawaiian islands.  Are you able to tell if they have an Asian admixture?  Can it be distinguished from the East Asian that is part of the Polynesian genome?  This is something I have been seeing more now particularly with adoptees.

Any excess East Asian percentage  compared to the Oceanian percentage [79% to 21%], would indicate that the person is admixed with some other Asian ancestry.  Since my mother’s genome does not indicate any more East Asian than what it should for Polynesians, it is clear that she does not have any additional Asian ancestry.


1. Genome-Wide Analysis indicates More Asian than Melanesian Ancestry of Polynesians
2. Population Genetic Structure and Origins of Native Hawaiians in the Multiethnic Cohort Study

Polynesia category – (part 2)

Last December I blogged an entry entitled Polynesia Category –  It was about their new Pacific Islander category and how I noticed Filipinos had been coming up with that category as well.  I have been noticing other people of mainly Southeast Asian ancestry having small percentages of Polynesia in their ancestry composition.

A Taiwanese aboriginal would have about the same amount as Filipinos, ranging around 25% – 29%.  A friend that I know whose mother is a Taiwanese aboriginal and he got DNA tested came up with 13% Polynesia and 37% Asia East.  That is nearly identical to what half Filipinos would get.  A Taiwanese person who has Fujian ties centuries ago and is not a Taiwanese aboriginal showed up having 8% Polynesia and 92% Asia East.  Another friend whose mother is from the Moluccas (Indonesia) and tested himself came up with 19% Melanesia, 12% Polynesia and 17% Asia East.  I have had three Chinese people who got tested with and told me that they also had Polynesia show up within the 10% – 11% range.

All of these are the results of picking up that Southeast Asian component that exists for these people including Polynesians, of which makes up about 80% of their genome.  A few weeks ago I spoke to Ross E. Curtis, Ph.D., a computational biologist, specializing in genetics and visual analytics, and has been with since 2012.  I informed him of the small sampling size that they had, just 18 samples (see Polynesia Category – and the problem that provided.  He was not too familiar with the details behind, mentioned a colleague that helped to put that category together but was unable to provide any real insight into that specific category.  So whatever I am learning about this comes from actual testees and especially those who are not of Polynesian ancestry.

My mother finally did test with back in March when I made an emergency trip to Wai’anae, O’ahu, Hawai’i and had my mother retake that test since it got rejected.  I had predicted that she would basically show up as 80% Polynesia and 20% European, but this is her actual results.Screen Shot 2015-06-30 at 7.33.11 AM


It is as almost that Ancestry tends to give us the lowest possible amount of European.  I got 8% while my mother got 11%.  And just as it did with FTDNA’s myOrigins, it gave my mother a bit more  Scandinavian.  Actually with myOrigins my mother gets 19% while I show 11% Scandinavia.

We will probably see more people of Southeast Asian origin come up with the Pacific Islander/Polynesia/Melanesia category as a proxy for Southeast Asian as more get tested.

No surprise however when I looked at her matches, a lot of close 2nd – 3rd cousin matches.

Screen Shot 2015-06-30 at 3.45.00 PM Screen Shot 2015-06-30 at 3.45.11 PM


My mother has three 1st – 2nd cousin matches, Extremely High confidence.  But for the 2nd – 3rd cousin range, she has sixty-seven matches, Extremely High confidence and a total of two-hundred sixty-seven matches for 3rd – 4th cousins, all Extremely High confidence as well.  Others not familiar with Polynesian endogamy, or endogamy in general insisted that these close matches have to be actual close matches.

I managed to match two of my mother’s 2nd – 3rd cousin matches on  Assuming that the matches are listed in order of closest to the least closest, these are how the two matches up with my mother.

The 7th 2nd – 3rd cousin match:
510cM – total shared
26.9cM – largest segment

The 40th 2nd – 3rd cousin match (out of 67):
331.9cM – total shared
22.2cM – largest segment

These are one of the few larger segment matches that she has while most rarely go up to 15cM.  Something common with endogamous groups.  Only uploading to GEDmatch can it be clear if the 2nd – 3rd cousin matches are real 2nd – 3rd cousin matches, or an endogamous ones that has numerous segments rarely going higher than 15cM for the largest segment.  Other than that, any Polynesian testing at Ancestry cannot assume that the 1st – 2nd or 2nd – 3rd cousin matches are actually that close as estimated.

Polynesia Category –

Earlier this year I tested with (or since I’ve been noticing non-Polynesians coming up with this new category.  This is way after the fact the research does not specify a Polynesia component, but rather a Melanesian and Asian or East Asian or Southeast Asian component.  I have seen other Asians, specifically Filipinos coming up with decent amount of this Polynesia category, as well as those of European descent coming up with small traces of Polynesia.

Under their Polynesia category, it mentions the sampling size was 18, and that one of the samples showed 11% Scandinavian.  A larger sampling size would yield better results especially in this case where one of the 18 samples had some European admixture.  This was enough to cause those with Scandinavian ancestry to come up with small traces of Polynesia, and in return cause people to wonder how they could have ever had such ancestry in their lineage to a point where some people create possible scenarios how they could have inherited this less than 0.1% Polynesia.

Screen Shot 2014-12-15 at 9.07.24 AM


Their Polynesia category was one of those categories where they had the least amount of samples.

Screen Shot 2014-12-15 at 9.42.16 AM

After receiving my results, as I suspected due to the fact that I am half Filipino, my percentage of the Polynesia category was pretty inflated.  It showed that I had 57% Polynesia versus 34% Asia East.  Knowing that my mother is 80% Hawaiian, and that my father was pure Filipino, I figured the amount of Asia that I showed 34% was missing 16% that was thrown into the Polynesia category.  That would in turn leave me with 41% Polynesia.  My mother is 20% European, and according to Ancestry I am 8% Europe, which seems to be about right.  The other DNA companies I tested at showed more than 10% Europe.  But adding the 41% plus the 8% comes out about right, 49%.

Recently I had a cousin on my father’s side of the family test, and she got her results.  She too is half Filipino, while her other half is completely Europe.  I expected her to show some Polynesia but I did not even guess how much that would be.  I was surprised to see 16% Polynesia for her, which is the same amount I had deducted from my own.  In fact, she shows 33% Asia while I show 34% Asia, and more specifically we both share 31% Asia East.  So they both are consistent.

Screen Shot 2014-12-15 at 9.03.53 AM

Although my mother was given an AncestryDNA kit, she has yet to take it.  But I can easily guess that she will easily show 20% Europe and 80% Polynesia.  Any other person who is Polynesian but admixed with some other Asian it may include part of their Asian component into Polynesia.  Maybe the fact that we are Filipinos and they have ancestral ties is why some of it is classified as such.  I did have another paternal cousin tested, she is half Filipino and half Japanese so not sure what type of results that will yield with the Polynesia category.  Will it be the same and show her as 16% Polynesia?  Or will it give her more due to her Japanese ancestry, or is that different enough to not be classified under the Polynesia category?

To find out more about AncestryDNA’s ethnicity/ancestry categories, you can read through their Ethnicity Estimate White Paper.