Simon Hedley
Oldest pre-I1 sample so far - BAL051 from skeleton found in Northeastern Iberia in the Serra de Busa Pre-Pyrenean range, described in a recent paper titled "Survival of Late Pleistocene Hunter-Gatherer Ancestry in the Iberian Peninsula" in Current Biology. The report states "BAL0051 could be assigned to haplogroup I1".
I've downloaded and analyzed the BAM file for BAL051 - like a lot of ancient samples, there are no calls at a lot of positions. Some of this is due to the analysis (1240K capture) and some may be due to the age of the sample. The age of BAL051 listed in the supplementary data as 11095±195; 10195±255 for lab radiocarbon date, 11,384–10,733; 10,681–9,263 Cal BCE (2σ), 13,380–12,660; 12,830–10,990 Cal BP (2σ) which would put it a bit before the TMRCA of modern I1.
BAL051 has reads for 33 of the 311 SNPs in the big I1 bottleneck block, so there is a lot of no calls where the position has no reading. However, those 33 SNPs (10.6% of the 311 I1 SNPs) have more read SNPs in BAL051 individually than the four RISE samples from Allentoft 2015 combined.
Derived calls for 13 SNPs: Z2699/FGC2430 (2C), Z2751/L841/YSC0000257, Z2885, Z2887, CTS7751/Z2813, Z2812/CTS7652, Z2860, L124/S64, CTS4532/Z2777, Z2724/V1771 (2G), FGC2441 (2G), CTS10140/Z2837.
Ancestral calls for 20 SNPs: Z2886, Z2679/CTS136, Z2727, Z2850, Y1962 (4T), P40, Z2747, FGC2422/Z2715, CTS3506/Z2765, FGC33327, CTS11534/Z2871, Y1863/S107/FGC2426, L848/Z2877 /YSC0000299 (2C), FGC2433, Y1950, FGC2427/Z2713, Y1932/S2007, Z2870/CTS11526, S22865/Z2845, Z2806/CTS6765.
All above one read unless otherwise noted.
Given the above data, I'd call BAL051 as pre-I1 or I* rather than I1. Given that he's ancestral for quite a few of the I1 SNPs, he could be an intermediate on the way from I to I1 whose descendants had those SNPs mutate from ancestral to derived in line with I1 or he could be a representative of a dead lineage that isn't an ancestor of modern I1. Difficult to say - still a lot of gaps in the coverage and quite a few one read results (like a lot of ancient samples) although BAL051 has a few in the 2-4 read range.