Log in

View Full Version : Hidden Italian/Sicilian/Sardinian ancestry in "Broadly Southern European"?



jtoml3
08-11-2015, 11:59 AM
Hi all,

Quick question for those familiar with genetics of Italians, Sicilians, and Sardinians. I have Croatian ancestry and no known ancestry from these places, however in quite a few calculator oracles I get South Italian or a variant of it. Italy covers 1.6% of my genome on 4 grandparents @5cM.

http://i.imgur.com/f7ywHZW.png

Most of my matches are quite small (5-6cM), how I noticed once I plot the family locations of those people with public profiles, their locations are in Southern Italy, Sicily and Sardinia,

http://i.imgur.com/wPQ8NZQ.png

I don't know if its a coincidence, or if I may have some distant ancestor from that part of Italy/Sardinia. If so, could it be coming under Broadly Southern European?

I know 23andMe use Sardinian as example of that here:

"Some populations, like Sardinian, are just hard to tell apart from others. It's important to note that when Ancestry Composition fails to assign Sardinian DNA, this does not necessarily mean that it incorrectly assigns it to something else, like Italian. If it were, that would show up as poor precision for the Italian population. Instead, Ancestry Composition will label this Sardinian DNA "Broadly Southern European" or "Broadly European."

I know it could be the reverse, and these Italians have Balkan ancestors rather than me having an Italian ancestor, however the fact they are from a similar region has me curious regardless.

Any thoughts would be great.

Petalpusher
08-11-2015, 02:56 PM
If you don't get Italian at all in the AC, it could just be very ancient from the Croatian, hence why it would fall into broadly and not specifically Italian (and still get distant relatives in the CoA). This is how i interpret it at least.

How much West Med and East Med do you get on K13?

jtoml3
08-11-2015, 09:54 PM
If you don't get Italian at all in the AC, it could just be very ancient from the Croatian, hence why it would fall into broadly and not specifically Italian (and still get distant relatives in the CoA). This is how i interpret it at least.

How much West Med and East Med do you get on K13?

23andMe

West_Med 13.93%
East_Med 9.01%

FTDNA

West_Med 13.16%
East_Med 10.07%

My thinking is along those lines. Because my matches seem to be from around a similar area I suppose its possible an ancestor decided to migrate across to the Balkans. On FTDNA I get no Southern European however I get 8% Asia Minor. My East Med is probably slightly higher than the average Western European anyway.

Petalpusher
08-12-2015, 10:37 AM
23andMe

West_Med 13.93%
East_Med 9.01%

FTDNA

West_Med 13.16%
East_Med 10.07%

My thinking is along those lines. Because my matches seem to be from around a similar area I suppose its possible an ancestor decided to migrate across to the Balkans. On FTDNA I get no Southern European however I get 8% Asia Minor. My East Med is probably slightly higher than the average Western European anyway.

Yes it's a bit high, it's at Croatian level but since you are not only Croatian... every other ancestry you have are very low on it. It might be a clue, An Abruzzo f.e would have nearly 30% East med so it's a big addition with any relatively recent ancestry. For sure it's not Sardinian, you would have way more West med How much West Asian do you get, i forgot to ask.

jtoml3
08-12-2015, 11:30 AM
Yes it's a bit high, it's at Croatian level but since you are not only Croatian... every other ancestry you have are very low on it. It might be a clue, An Abruzzo f.e would have nearly 30% East med so it's a big addition with any relatively recent ancestry. For sure it's not Sardinian, you would have way more West med. How much West Asian do you get, i forgot to ask.

23andMe

West_Asian 5.43%

FTDNA

West_Asian 5.01%

From what I've seen I don't think they are above the norm. The Dodecad K7b calculator is what got my attention, even though its one of the older ones. The oracle in particular is quite interesting with regards to my above findings, i.e. the S_Italian_Sicilian. I don't think its a proxy for Balkan ancestry because I know Italians and other Balkan countries, besides Croatia, have different genetics.

Admix Results (sorted):

# Population Percent
1 Atlantic_Baltic 68.76
2 Southern 15.49
3 West_Asian 13.78
4 South_Asian 1.79
5 East_Asian 0.14
6 African 0.04

Single Population Sharing:

# Population (source) Distance
1 Hungarians (Behar) 2.43
2 French (HGDP) 5.59
3 French (Dodecad) 6.12
4 Mixed_Germanic (Dodecad) 6.28
5 German (Dodecad) 6.83
6 Cornwall (1000Genomes) 7.28
7 Dutch (Dodecad) 7.73
8 CEU30 (1000Genomes) 7.79
9 Kent (1000Genomes) 8
10 English (Dodecad) 8.39
11 Ukranians (Yunusbayev) 9.68
12 British (Dodecad) 9.68
13 British_Isles (Dodecad) 9.86
14 Irish (Dodecad) 9.93
15 Argyll (1000Genomes) 10.07
16 Orcadian (HGDP) 10.35
17 Mixed_Slav (Dodecad) 10.85
18 Orkney (1000Genomes) 11.09
19 Polish (Dodecad) 12.57
20 Cataluna (1000Genomes) 13.26

Mixed Mode Population Sharing:

# Primary Population (source) Secondary Population (source) Distance
1 88% British_Isles (Dodecad) + 12% Iranian_Jews (Behar) @ 0.73
2 88% British_Isles (Dodecad) + 12% Iraq_Jews (Behar) @ 0.74
3 88.2% British (Dodecad) + 11.8% Iranian_Jews (Behar) @ 0.8
4 90% Cornwall (1000Genomes) + 10% Uzbekistan_Jews (Behar) @ 0.8
5 88.2% British (Dodecad) + 11.8% Iraq_Jews (Behar) @ 0.82
6 86.4% Orcadian (HGDP) + 13.6% Druze (HGDP) @ 0.83
7 85.8% Orcadian (HGDP) + 14.2% Lebanese (Behar) @ 0.85
8 61.1% British_Isles (Dodecad) + 38.9% Romanians (Behar) @ 0.9
9 87.4% British_Isles (Dodecad) + 12.6% Assyrian (Dodecad) @ 0.91
10 90.1% Cornwall (1000Genomes) + 9.9% Iranians (Behar) @ 0.92
11 87.6% British (Dodecad) + 12.4% Assyrian (Dodecad) @ 0.94
12 87.4% British_Isles (Dodecad) + 12.6% Azerbaijan_Jews (Behar) @ 0.95
13 90.4% Cornwall (1000Genomes) + 9.6% Azerbaijan_Jews (Behar) @ 0.95
14 90.9% Cornwall (1000Genomes) + 9.1% Iranian_Jews (Behar) @ 0.96
15 88.7% English (Dodecad) + 11.3% Uzbekistan_Jews (Behar) @ 0.96
16 86.9% British_Isles (Dodecad) + 13.1% Uzbekistan_Jews (Behar) @ 0.97
17 87.6% British (Dodecad) + 12.4% Azerbaijan_Jews (Behar) @ 0.97
18 87.1% British (Dodecad) + 12.9% Uzbekistan_Jews (Behar) @ 0.98
19 89.3% CEU30 (1000Genomes) + 10.7% Syrians (Behar) @ 0.98
20 86.2% Orcadian (HGDP) + 13.8% Syrians (Behar) @ 0.98

Using 2 populations approximation:
1 50% Lithuanian +50% N_Italian @ 1.949568


Using 3 populations approximation:
1 50% British_Isles +25% Lithuanian +25% S_Italian_Sicilian @ 1.164295


Using 4 populations approximation:
1 British_Isles + Lithuanian + Orcadian + S_Italian_Sicilian @ 1.098230
2 British + Lithuanian + Orcadian + S_Italian_Sicilian @ 1.142910
3 Lithuanian + Orcadian + Orcadian + Sicilian @ 1.148125
4 Lithuanian + Orcadian + Orcadian + S_Italian_Sicilian @ 1.157268
5 British_Isles + British_Isles + Lithuanian + S_Italian_Sicilian @ 1.164295
6 British_Isles + Greek + Orcadian + Polish @ 1.171025
7 British_Isles + Lithuanian + Orcadian + Sicilian @ 1.184988
8 British_Isles + Irish + Lithuanian + S_Italian_Sicilian @ 1.205353
9 Ashkenazi + Lithuanian + Orcadian + Orcadian @ 1.211768
10 Irish + Lithuanian + Orcadian + S_Italian_Sicilian @ 1.212182
11 British + British_Isles + Lithuanian + S_Italian_Sicilian @ 1.214782
12 Greek + Orcadian + Orcadian + Polish @ 1.216809
13 British + Greek + Orcadian + Polish @ 1.218620
14 British + Lithuanian + Orcadian + Sicilian @ 1.232746
15 Irish + Lithuanian + Orcadian + Sicilian @ 1.253166
16 British + Irish + Lithuanian + S_Italian_Sicilian @ 1.254188
17 British_Isles + Lithuanian + Orkney + S_Italian_Sicilian @ 1.259488
18 British_Isles + British_Isles + Greek + Polish @ 1.259511
19 British_Isles + British_Isles + Lithuanian + Sicilian @ 1.260756
20 British_Isles + Lithuanian + Orkney + Sicilian @ 1.265599

Petalpusher
08-12-2015, 12:03 PM
23andMe

West_Asian 5.43%

FTDNA

West_Asian 5.01%

From what I've seen I don't think they are above the norm. The Dodecad K7b calculator is what got my attention, even though its one of the older ones. The oracle in particular is quite interesting with regards to my above findings, i.e. the S_Italian_Sicilian. I don't think its a proxy for Balkan ancestry because I know Italians and other Balkan countries, besides Croatia, have different genetics.



It's all relative your East Med isn't high by all means, but Brit/Irish, Poland, etc.. get almost nothing, 1-3%, so it comes from somewhere else. You can have one outlier somewhere but it's extremely unlikely you had outliers in every countries of your ancestry.

The thing is Central Italian can be 15% West Asian in K13 which is enormous for Europe, it starts low in N.Italy like everywhere else in West Euro, peek in the center + South and goes down again in Sicily. On the opposite Sardinian have absolutely no West Asian it's zero (again K13 as ref), so it's an interesting bit to look for, since yours is low even for British Isles it could be a mix of Italy and Sardinian exactly like on your map but not Sardinian alone, or just Sicilian maybe. Now i don't know your entire exact tree and proportions and you have a lot of different ancestry but i think it's these components you should look at, they clearly set Italy apart from any other countries in the West.

jtoml3
08-12-2015, 12:13 PM
It's all relative your East Med isn't high by all means, but Brit/Irish, Poland, etc.. get almost nothing, 1-3%, so it comes from somewhere else. You can have one outlier somewhere but it's extremely unlikely you had outliers in every countries of your ancestry.

The thing is Central Italian can be 15% West Asian in K13 which is enormous for Europe, it starts low in N.Italy like everywhere else in West Euro, peek in the center + South and goes down again in Sicily. On the opposite Sardinian have absolutely no West Asian it's zero (again K13 as ref), so it's an interesting bit to look for, since yours is low even for British Isles it could be a mix of Italy and Sardinian exactly like on your map but not Sardinian alone, or just Sicilian maybe. Now i don't know your entire exact tree and proportions but i think it's these components you should look at, they clearly set Italy apart from any other countries in the West.

I have a fairly extensive Croatian tree, all of which is from Dalmatia. I can go back to the 1400s on some lines. That big Italian match I have is 15.6cM who states on his profile his ancestry is from Sardinia. I sent him a private message ages ago but he never replied. I know that doesn't suggest I have ancestry there, however its one of my bigger matches, and I don't know the chances of him having a Croatian ancestor, instead of me having an Italian/Sardinian ancestor.

I think its interesting that particular oracle picks up Sicilian often regardless.

Petalpusher
08-12-2015, 12:26 PM
I have a fairly extensive Croatian tree, all of which is from Dalmatia. I can go back to the 1400s on some lines. That big Italian match I have is 15.6cM who states on his profile his ancestry is from Sardinia. I sent him a private message ages ago but he never replied. I know that doesn't suggest I have ancestry there, however its one of my bigger matches, and I don't know the chances of him having a Croatian ancestor, instead of me having an Italian/Sardinian ancestor.

I think its interesting that particular oracle picks up Sicilian often regardless.

Bad luck he doesn't answer, some have really extended trees, i made big leap forward thanks to one cousin on 23. The Oracle just see the same thing, no surprise, it gives you 3 northern pop that fits well Uk+Poland and then it sees a relatively high east med for these 3 pop, so it is throwing a 25% Sicilian because they are High East med but not as highly West asian than in the center. It gives a very close distance with this 4pop.

jtoml3
08-12-2015, 08:50 PM
Bad luck he doesn't answer, some have really extended trees, i made big leap forward thanks to one cousin on 23. The Oracle just see the same thing, no surprise, it gives you 3 northern pop that fits well Uk+Poland and then it sees a relatively high east med for these 3 pop, so it is throwing a 25% Sicilian because they are High East med but not as highly West asian than in the center. It gives a very close distance with this 4pop.

Interesting. So you're saying its likely I don't have Italian ancestry, but rather the oracle is simply fitting in a population because it has to? Would a Balkan population be a better fit than Sicilian, or do they have less East Med than Italians/Sicilians?

Sikeliot
08-12-2015, 09:03 PM
East Med rises in Sicily but their West Asian drops. Balkans including Greece have high West Asian like central and southern Italy but their East Med drops instead.

jtoml3
08-12-2015, 09:20 PM
East Med rises in Sicily but their West Asian drops. Balkans including Greece have high West Asian like central and southern Italy but their East Med drops instead.

Interesting. So my high East Med (in comparison to my known ancestry) is not necessarily due to my Croatian, but potentially in combination with something else? Theoretically, should I have more West Asian than East Med if that is the case?

My K7b oracle using 23andMe data out of interest. Is it acting as a proxy for Jewish?

Using 4 populations approximation:
1 Ashkenazy_Jews + Lithuanian + Orcadian + Orcadian @ 1.284827
2 Lithuanian + Orcadian + Orcadian + Sicilian @ 1.290056
3 Ashkenazi + Lithuanian + Orcadian + Orcadian @ 1.295971
4 Lithuanian + Orcadian + Orcadian + S_Italian_Sicilian @ 1.301442
5 Argyll + Lithuanian + Orcadian + S_Italian_Sicilian @ 1.318744
6 Argyll + Lithuanian + Orcadian + Sicilian @ 1.328874
7 Argyll + British_Isles + Lithuanian + S_Italian_Sicilian @ 1.329473
8 British_Isles + Lithuanian + Orcadian + S_Italian_Sicilian @ 1.344922
9 Argyll + Ashkenazi + Lithuanian + Orcadian @ 1.349870
10 Argyll + Ashkenazy_Jews + Lithuanian + Orcadian @ 1.354017
11 CEU30 + CEU30 + Hungarians + Romanians @ 1.364802
12 Ashkenazy_Jews + Lithuanian + Orcadian + Orkney @ 1.366374
13 Irish + Lithuanian + Orcadian + S_Italian_Sicilian @ 1.371590
14 Argyll + Irish + Lithuanian + S_Italian_Sicilian @ 1.372042
15 Argyll + British + Lithuanian + S_Italian_Sicilian @ 1.377455
16 Argyll + Greek + Orcadian + Polish @ 1.379692
17 British + Lithuanian + Orcadian + S_Italian_Sicilian @ 1.380318
18 Ashkenazi + Dutch + Lithuanians + Orcadian @ 1.380883
19 British_Isles + Lithuanian + Orcadian + Sicilian @ 1.386338
20 Ashkenazy_Jews + Dutch + Lithuanians + Orcadian @ 1.388380

Petalpusher
08-12-2015, 09:50 PM
Interesting. So you're saying its likely I don't have Italian ancestry, but rather the oracle is simply fitting in a population because it has to? Would a Balkan population be a better fit than Sicilian, or do they have less East Med than Italians/Sicilians? They have way less east med, unless you are going all they way down in Greece which is similar than Italy.

In my experience, 4 pop tends to polarize things. It's not necessarily S.Italian/Sicilian, but the data suggest something increased your east med compared to what you would get with all your ancestry because there s no reason you would get 9% with ancestries that are all 1-3%. There s a couple of population that can do this and they are mostly Italian in different proportions, it's just impossible to tell with an autosomal precisely where and how much exactly, since Croatians and Italians are very different genetically but close geographically, it only strongly points to somewhere in Italy.

I gave some thoughts to all this because im in almost the exact same case, in K13 and many Oracle. I get 3 NW Euro + 1 high west med, the difference is it mostly tries to increase my west med (Sardinian are just like N.Italian on East Med but have way more West med). And i know for sure im about 25% N.Italian, sometimes it is using it too, most of the time it takes the 4 most polarizing pop and of course im not 75% English, Irish, Austrian or whatever either.

Using 4 populations approximation:
1 Austrian+Irish+Irish+Sardinian @ 1,658905
2 East_German+Irish+Sardinian+Southwest_English @ 1,681795
3 Austrian+Irish+Sardinian+Southwest_English @ 1,710424
4 Austrian+Norwegian+Sardinian+Southwest_English @ 1,728833
5 Austrian+Irish+Orcadian+Sardinian @ 1,74162

jtoml3
08-12-2015, 10:08 PM
They have way less east med, unless you are going all they way down in Greece which is similar than Italy.

In my experience, 4 pop tends to polarize things. It's not necessarily S.Italian/Sicilian, but the data suggest something increased your east med compared to what you would get with all your ancestry because there s no reason you would get 9% with ancestries that are all 1-3%. There s a couple of population that can do this and they are mostly Italian in different proportions, it's just impossible to tell with an autosomal precisely where and how much exactly, since Croatians and Italians are very different genetically but close geographically, it only strongly points to somewhere in Italy.

I give some thoughts to all this because im in almost the exact same case, in K13 and many Oracle. I get 3 NW Euro + 1 high west med, the difference is it mostly tries to increase my west med (Sardinian are just like N.Italian on East Med but have way more West med). And i know for sure im about 25% N.Italian, sometimes it is using it too, most of the time it takes the 4 most polarizing pop and of course im not 75% English, Irish, Austrian or whatever either.

Using 4 populations approximation:
1 Austrian+Irish+Irish+Sardinian @ 1,658905
2 East_German+Irish+Sardinian+Southwest_English @ 1,681795
3 Austrian+Irish+Sardinian+Southwest_English @ 1,710424
4 Austrian+Norwegian+Sardinian+Southwest_English @ 1,728833
5 Austrian+Irish+Orcadian+Sardinian @ 1,74162

I see what you mean. For me, if I'm 1/4 Croatian on paper and Croatia covers 1.7% of my genome, then I shouldn't rule out Italy if it covers 1.6% of my genome, even if I don't have a known ancestor. All I have to go on at the moment is autosomal data and matches locations. If my matches locations are in and around southern Italy and, even though not reliable, the oracle gives me S.Italian/Sicilian, it seems like a possible area if I do have a distant ancestor and an interesting pattern regardless. It may just be a coincidence.

I see what you mean in your case. At least you know you have Italian ancestry so it makes it slightly easier to interpret.

Not a Cop
08-13-2015, 02:52 PM
I see what you mean. For me, if I'm 1/4 Croatian on paper and Croatia covers 1.7% of my genome, then I shouldn't rule out Italy if it covers 1.6% of my genome, even if I don't have a known ancestor. All I have to go on at the moment is autosomal data and matches locations. If my matches locations are in and around southern Italy and, even though not reliable, the oracle gives me S.Italian/Sicilian, it seems like a possible area if I do have a distant ancestor and an interesting pattern regardless. It may just be a coincidence.

I see what you mean in your case. At least you know you have Italian ancestry so it makes it slightly easier to interpret.

You get south-Italian matches for simple reason- s. Italians moved to America, intermarried with Brits and now list their ancestry as Calbaria+Essex. You match them on Essex side, but 23andme still shows both locations since it doesn't know on which side do you match them.

jtoml3
08-15-2015, 01:20 AM
You get south-Italian matches for simple reason- s. Italians moved to America, intermarried with Brits and now list their ancestry as Calbaria+Essex. You match them on Essex side, but 23andme still shows both locations since it doesn't know on which side do you match them.

In that case, why would they list all 4 grandparents being from Italy?

Sikeliot
08-15-2015, 01:35 AM
"Southern Italian" on Oracle could come up for anyone with European origin with substantial Neolithic Near Eastern. I think it came from your Jewish side and has nothing to do with Croatia.

jtoml3
08-15-2015, 10:13 PM
"Southern Italian" on Oracle could come up for anyone with European origin with substantial Neolithic Near Eastern. I think it came from your Jewish side and has nothing to do with Croatia.

This is what I'm thinking. In that K7b oracle Ashkenazi and Sicilian never appear in the same combination. Its one or the other so I assume they are being interchanged within the algorithm to represent the same thing.