If I understand correctly, you are saying that Y-STR NMDS plots is a weakened or filtered version of autosomal PCA. But the two methods measure entirely different things. Autosomal PCA reflects genome-wide ancestry shaped by recombination and admixture across both sexes. Y-STR NMDS reflects only a single non-recombining paternal lineage and is dominated by drift, founder effects, mutation-rate variance, and sampling structure. Any visual similarity between the two is incidental and occurs only where male-line history happens to follow the same broad geographic patterns as overall ancestry. This similarity breaks down immediately in cases of isolation, endogamy, or sex-biased history, which is why groups like Maniots or Jews look normal in autosomal analyses but “odd” in STR plots.
The idea that certain haplogroups “carry more weight” or that small percentages of lineages can “pull” populations misunderstand how distance-based ordinations work. STR NMDS does not weight lineages by age, frequency, or autosomal contribution. Apparent influence comes from homogeneity, drift, or overrepresentation in the sample, not from real ancestry proportions. These are mathematical artifacts, not biological signals.
STR plots cannot undo autosomal mixing or recover a population’s “original” position. They remove information rather than refine it by ignoring nearly all ancestry outside the paternal line.
Concepts like “ethnic Y-STR makeup” or “autosomal (cultural) history” are not meaningful in population genetics. Y-STRs describe paternal lineage structure, not ethnicity.
The idea that populations like Portuguese, Basques, Spaniards, or Italians have “5000 years of ethnic (STR) continuity” is not correct. Their Y-chromosome histories include multiple turnovers, bottlenecks, and expansions, especially during the Bronze Age and later periods. The Basques, in particular, are a counterexample, they have strong autosomal continuity relative to Neolithic farmers, but near-complete Y-chromosome replacement by Bronze Age R1b lineages. That is the opposite of uninterrupted “ethnic STR continuity.”
The arguments about Mycenaean continuity, enhanced Caucasian ancestry, or ethnogenesis cannot be inferred from Y-STR alone. Such conclusions require ancient DNA, dated Y-SNP phylogenies, and demographic modeling. At most, the Y-STR NMDS plot shows that Deep Maniots are a highly drifted, J2-dominated paternal population clustering with other J2-heavy, drifted groups in STR space, nothing more.
So, what they did by mixing all the Y-STRs makes absolutely no sense when there is already a terminal SNP like J2>L26>L923.
STR mixes should only be done within the same SNP clade, not mixing J1 with J2 and R1b as they did in this case.
By simply looking at the SNP, the date, and the bottleneck size, you can already deduce what happened.
Since they mixed clades, the projection ends up resembling, to some degree, an autosomal PCA because different SNP clades were combined. That’s why I say it looks similar to an autosomal PCA, but without the autosomal SNP IBD segments that mtDNA contributes; they have nothing to do with each other since they’re only measuring, so to speak, the last ~1,000 years of Y-SNP variation.
NMDS STR graphs are only useful as an illustrative narrative to visually show recent patterns or anomalies, but if you understand what an STR and an SNP are, these kinds of projections are unnecessary.
STR PCAs are “confirmatory.”
Autosomal PCAs are “representative” based on pre-selected SNP sets; that’s why a sample with 5% coverage gives the same result as a WGS.
Y-17 STRs only reliably reflect roughly the last ~1,000 years.
Each SNP is a solid time marker; STRs show variation within that SNP but cannot be traced back thousands of years without direct archaeological evidence.
This is why mixing STRs from different lineages without considering the terminal SNP produces “absurd” results.
Example of misinterpretation:
In NMDS or PCA STR analyses, you see Sephardic Jews clustering between Spaniards and Italians, even though genetically they barely share ZZ11.
This is because:
• ZZ11> is very common in Spain and Italy (>50%), whereas Jews have very little (<10%).
• Each ZZ11 reaching Jewish populations comes from different lineages depending on the region, while P58 (representative of Jews) is much smaller and therefore has less STR variability.
Thus, STRs reflect recent homogeneity, not real ancestry or deep continuity.
This is why mixing all STRs from different SNPs pulls them toward the Western cline.
About the Basques and what I meant when I said their STR profile hasn’t “changed” but “evolved” over the last 5,000 years:
Considering SNPs and approximate dates:
• P312> 3500–3000 BCE → 90%
• P312>DF27> 2800–2700 BCE → 70%
• P312>L21> 2700–2600 BCE → 20%
• P312>DF27>Z195 2650 BCE → 40%
• P312>DF27>ZZ12 2650 BCE → 30%
• G2> pre-5000 BCE → 5%
• I2> pre-8000 BCE → 5%
STRs change slowly (~1,000 years per significant change). That’s why, if you take an average of Y-17 STRs over the last 1,000 years, they appear relatively homogeneous in a population with such SNP continuity.
This structure confirms 5,000 years of continuity because what certifies continuity are the SNPs, validated with ancient samples: EHU002, of non-steppe and non-Central European origin, confirmed by strontium analysis (P312>DF27>ZZ12 from 2600 BCE), shares direct continuity with modern ZZ12 in the Iberian Peninsula. While its exact STR profile isn’t identical, using multiple ancient specimens spaced every 500 years can demonstrate SNP continuity and also certify STR continuity.
P312> did not arrive from any external source; it is an internal Atlantic/Mediterranean expansion.
The correct method is to use SNPs as a base and STRs to analyze millennial evolution between consecutive SNPs. This allows reconstruction of 10,000 years of genetic history step by step, always validated with ancient samples.
In the autosomal context:
If you take any Iberian individual (Lusitania, Tartessos, Cogotas, Gallaecia, Asturias, or Argar), whether I2 from the Chalcolithic or P312 from Chalcolithic/Bronze Age, you will see:
• 60% Atlantic/Mediterranean ancestry, and
• From 2000–1000 BCE onward, ~40% derived from minor absorption of Amazigh, Phoenicians, Greeks, Romans, and Goths.
The Basques only absorbed ~10% foreign input.
Internal drift is minimal: over 5,000 years it does not drastically alter the autosomal landscape.
NMDS analyses of Y-STRs are only illustrative of recent male structure.
They cannot replace or refine autosomal DNA or deep lineage history.
In both types of PCA, Basques would appear in the same relative position to other groups because they have had very few changes and very high continuity. Portuguese, Italians, and Spaniards would also meet similar criteria.
Any attempt to extrapolate “ethnic continuity” or “cultural ancestry” from STRs alone is mathematically and biologically incorrect; SNPs and ancient samples of related lineages are required.
To interpret where this STR mixture projects, you must consider the SNPs used to generate that “projection mix.”
This explains the “pull” of Sephardic Jews toward the Western cluster or the “pull” of Maniots toward the Chechen STR cline.
Their positional fit in the Caucasian region in PCA is not coincidental; it is causal, because Y-STRs of J2 originated in the Caucasus, so they will always tend to reflect that cline, since that’s where the most basal J2 lineages exist.
Thus, seemingly unrelated phenomena are in fact related.
Both Maniots and Sephardic Jews appear in “odd” positions compared to an autosomal PCA, but for two completely different reasons: because different SNPs of different haplogroups are being mixed in the same NMDS STR.
In this case, the STR projection resembles autosomal PCA because of STR mixing:
For Jews, it is due to haplogroup mixing.
For Maniots, it is due to isolation.
