Fig WGS lineage project · result note

Purple Tiger is not Martinenca Rimada, and not Panachée

A fig with no paper trail, bought from a vanished Etsy seller, and two experienced guesses at what it really is. Across 4,973,908 filtered SNPs in a 52-genome panel, both guesses are wrong — but the reason they were made is real, and the DNA shows what it is instead.

Whole-genome shotgun sequencing · 52 fig accessions · 4,973,908 SNPs after filtering · reference Ficus carica UNIPI FiCari 1.0 (GCA_009761775.1)

The question

Gary of Mazzei’s Nursery bought the cuttings years ago on Etsy, already labeled ‘Purple Tiger’. Very little has surfaced about the variety since. Gary later tried to track down the original Etsy seller and could not — too much time had passed — so there is no seller, no origin story and no published description behind the name.

Brian M visited Gary’s greenhouse and noticed that the tree was growing in the same area as his Martinenca Rimada, and that both varieties were fruiting at about the same time. The fruit is distinctive — Purple Tiger develops an unusually jet-black exterior when fully ripe — and because the two looked different, Brian did not think it likely they were the same variety. That was the initial call, stated on camera: confident that they were not.

As more growers saw the fruit, that confidence was challenged. Brian M, who obtained the variety, raised two specific alternatives: that ‘Purple Tiger’ is Martinenca Rimada under another name, or possibly Panachée. Both are real cultivars, both are in this panel, and both claims are testable. That is what this note does.

Two named cultivars, proposed by the person closest to the material, and a genome panel that happens to contain all three. This is the rare case where a collector-community argument has a decisive experiment available.

The answer

Each proposal was scored on four marker classes with different failure modes, so that no single kind of sequencing error can produce the result. Every threshold comes from reference pairs in the same panel: three leaves off one tree (the noise floor), the clone groups the study calls, the panel’s confirmed first-degree pair, and two genuinely different cultivars.

Measurement	Purple Tiger ↔ Purple Tiger-1	↔ Martinenca Rimada	↔ Panachée	Clone limit
Opposite-homozygote ratio (R0)the sharpest clone test here; exactly 0 for a clone, and it keeps working on shallow libraries	0 (exactly)	0.007645	0.004947	≤ 0.000028
Whole-genome differences per Mbcounted only over positions both libraries actually called	19,882	229,581	261,814	15,237–38,152
Chromosome windows sharing both haplotypesof 271 windows of 1 Mb, along the 13 chromosome-scale scaffolds	271	52	33	271
Longest run of windows that do not matchshared ancestry is contiguous along a chromosome; noise is not	0	60	74	0–1
Structural-variant disagreementlarge insertions and deletions, an entirely separate kind of evidence from SNPs	1.91%	25.73%	27.62%	1.49–1.91%

Every cell in the two right-hand columns is outside the clone limit, and the left-hand column — measured by the identical code on the pair this study does call a clone — is inside it on all five rows. That contrast is the control: a comparison that failed to find a match in every direction would have rejected the real clone too.

both haplotypes shared (same genotype)one haplotype shared (close kin)no haplotype shared

So what is it?

Two things, and the second is why the question was a good one.

It is its own genotype, confirmed from two samples off one tree. The Purple Tiger-1 column above is a second sample from the same tree, sequenced in a separate run as a control, and it is one clonal genotype with the original on every measurement in the table — so the identity is real, and a sample mix-up is excluded along with library artifacts. As far as this panel and 775 public fig sequencing runs can see, no other name matches Purple Tiger’s genotype.

It is a first-degree relative of the Black Madeira clone. That clone is the one holding Black Madeira, Figo Preto (CW), Craven’s Craving, Mario 50, Burjassot Negre — and Martinenca Rimada. Purple Tiger is first-degree kin to it — a parent, offspring or full sibling; the data cannot say which — sharing one haplotype across long contiguous stretches (Figure 1, third row). So growers comparing Purple Tiger with a Martinenca Rimada are comparing close relatives. The resemblance is inherited, not a relabeling.

Panachée is a different case: it belongs to another clone group in this panel (with CDD Roja and Blanca R), and Purple Tiger is not a member of that one either.

Method, in one paragraph

Both Purple Tiger samples sequenced here — the original accession and Purple Tiger-1 — are leaves off Brian M’s tree, which is what the ‘(BM)’ in the original’s label records — the source of the material, not a variety or a group. Gary’s tree traces to the same propagation line, so the result speaks to both; that shared line is testimony rather than something the sequencing establishes, and it is undocumented upstream of the Etsy purchase in any case.

DNA from leaf tissue, sequenced as whole genomes and aligned to the Ficus carica UNIPI FiCari 1.0 (GCA_009761775.1) reference, then genotyped jointly across all 52 accessions at once — jointly, because genotyping samples one at a time destroys exactly the statistic this rests on. After filtering, 4,973,908 SNPs remain. Comparisons are always expressed per position that both libraries actually called, since the libraries differ in completeness and a raw count would make an incomplete library look like everyone’s closest relative. Every threshold in the table is derived from reference pairs in the same run, not carried over. Full technical detail, all 52 accessions and the other clone groups are in the main report.