Odd behaviour of outliers

[EveTC]

Hello,

I was wondering if you could help me. After successfully running lffm_ridge() and lfmm_test() on my dataset I receive two SNPs that are extremly high outliers for multiple environmental varaibles.

I took a look at these SNPs in the vcf and found that both are fixed 0/0 in the all but two populations where all individuals are 1/1. I then thought that perhaps these sites had similar environmental variables (where they are outliers) however they are, in fact, contrasting.

I do not know the ins-and-outs of the method, but I can't understand how these SNPs would look like outliers if the environmnetal variables in the two popualtions are so contrasting?

SNMF population coancestry suggests a K=6, which was then used in the lffm_ridge()

Cross-entropy	Ancestry barplot

The two sites I refer to above (that are fixed at 1/1 at the two outliers) are 30-CCP and 33-TAY, that cluster neatly as different populations. The rest of the sites are 0/0. Below is a screen shot of some of the env variables that these two SNPs show as large outliers.

The env variables (non-sclaed) for these popualtions are below and the range of the variable in the whole dataset...

Site	MinTemp	GDD5	CMI	hedge density	Broadleaf	Conif
30-CPP	1.8	20640.0	0.53	4.481674e-01	55	3
33-TAY	3.6	25015.5	0.68	1.610853e-13	60	0
overall range of var	-1.2 - 3.6	11817.0 - 26203.5	0.25 - 0.72	0.0 - 1.197232	0 - 60	0-100

Please can you help me understand this behaviour? To me it would make sense for these SNPs to be outliers if the env variables were similar/ the same but they are not? It makes me concerned about the other outliers identified... or are these just not behaving properly for some reason?

Thank you in advance