Your genind file seems to be fine. It would be more informative to see the eigenvalues of the principal components (if you use dudi.pca: barplot(pca.object$eig) ) rather than the cumulative variance and the pca plot itself.

It looks to me like you simply have no strong structure in your data set. Are you expecting clusters or is it rather an exploratory analysis? The pca plot should give you an idea how "clustered" your data is. By the way, from your description it seems like you did not account for LD, but you should when performing PCA (see Zou et al., 2010). You can also calculate the cluster tendency using Hopkins statistic (Lawson and Jurs 1990) with the get_cluster_tendency() command from the factoextra package. There are a bunch of methods for cluster validation like silhouette width or dunn index, so try a few and check first if your data is clustered at all and then decide on an appropriate number of cluster.

I suggest to check out the practical guide to cluster analysis in R from Datanovia. It is very well made and will give you theoretical understanding and practical tools to perform a solid cluster analysis.

https://www.datanovia.com/en/blog/cluster-analysis-in-r-practical-guide/

Lawson and Jurs, 1990

Article New Index for Clustering Tendency and Its Application to Che...

Zou et al., 2010; doi: 10.1159/000288706

Article Quantification of Population Structure Using Correlated SNPs...

I think Timo Hellwig nailed it: there just might not be any genetic structure among these populations. Eigenvalues would be informative, as would a general PCA plot. Also, the developers of adegenet have developed a new method for identifying genetic clusters that may be more useful. The user manual is pretty good.

Article A fast likelihood solution to the genetic clustering problem

I have never tried this package. You may try ADMIXTURE which applies cross validation technique to define the appropriate K. You may find more quick options; Pcadapt in R is among others.