GISAID sequence phylogeny for the H5 gene

https://gisaid.org/fileadmin/c/gisaid/files/H5N1-HA-tree/H5N1_HA_subsampled_tree_US_20250302.pdf
This is a phylogeny of the HA gene segment.  It is for the H5 Asian allele that is found in both the B3.13 dairy genotype and the D1.1 wild bird genotype that has started to be found in dairy cattle.
The B3.13 H5 sequence is at the top of the phylogeny and the D1.1 H5 sequence is below and includes the Washington poultry worker (A/Washington/240/2024|2024-10-18).  A/Nevada/10/2025|2025-02-04 is a recent D1.1 dairy infection.  The H5 sequence came into the US in 2022, but the D1.1 and B3.13 alleles of the H5 gene may have separated before getting to the US.  They don't have any sequences from 2022 so I can't tell.
The bogus way that the USDA labels their sequences inhibits interpretation of the phylogeny.  The USDA only gives the collection date and sequencing date, and lists the location as USA.  GISAID seems to try to infer the location by the date of sequencing and press releases about those dates.  If they can't infer the location they leave it as USA which is useless for epidemiology within the US.
It looks like the Iowa and Wisconsin patients were infected with a D1.3 genotype closely related to D1.1.  The Washington, Oregon, Wyoming, and Louisiana patients seem to have been infected with different strains of D1.1.  But all the D1.1 and D1.3 H5 alleles seem to come from a clade that may have branched off in 2022-2023.
Something is strange about the B3.13 genotype H5 alleles.  It has been known that the first human patient in Texas with the B3.13 genotype was a sequence outlier compared to all the other dairy sequences. A/Texas/37/2024|2024-03-28 still falls outside of all the other dairy sequences, but it now groups with a mountain lion from Montana, and a wild bird from Wyoming.  For the NA gene sequence the Texas patient is within the dairy clade instead of out side of it and is most closely related to the Missouri and Michigan human patients.  The Texas H5 sequence may just have multiple mutations that occurred during the infection of that individual.  The mountain lion and wild bird N1 sequences continue to fall outside of the dairy clade.  My take is that this means that the Texas patient H5 sequence should not be trusted in trying to figure out the source of the dairy infection.
One thing to note is that all the California H5 sequence dairy infections are a separate clade that looks significantly different from other dairy infections.  Their sequences are also more variable than those more closely related to the Texas B3.13 H5 sequences.  It looks like the infection has been in California for a lot longer than when it was first detected late last year.  For some reason there seems to have been sequence evolution before California was infected or after the first few herds were infected, possibly, before Texas was detected in March 2024.  They need to redo their estimated time of divergence for the dairy B3.13 genotype.  A full genome sequence comparison is likely needed to figure out how California fits in with the spread of the dairy virus.  California commercial poultry farms started to go down in the central valley in October/November 2023 when it was estimated that the dairy virus entered into dairy cattle.  Because of the stupid way that the USDA labels their sequences I could not tell which ones came from California last year, and I was told that the USDA did not give out the names of the researchers involved in the sequencing so that I could ask them to identify the California samples from 2023.
The PB2 data may have some sample labeling issues or sample mix up because a sequence from one of the California human patients (California/192/2024) definitely groups with the D1.1 clade for the PB2 gene sequence, but the NA and H5 gene sequences group with the B3.13 genotype.
NA phylogeny:
https://gisaid.org/fileadmin/c/gisaid/files/H5N1-NA-tree/H5N1_NA_subsampled_tree_US_20250302.pdf
PB2 phylogeny:
https://gisaid.org/fileadmin/c/gisaid/files/H5N1-PB2-tree/H5N1_PB2_subsampled_tree_US_20250302.pdf
Ron Okimoto