negative Fis values in every population

[Russell P]

I've sampled multiple populations of a species of frog (a typical outbreeding species) and for every population I'm getting observed heterozygosities (e.g 0.66) much larger than expected heterozygosities (e.g. 0.38) and correspondingly very negative Fis values (e.g. -0.65).

I can think of no biologically realistic explanations for this degree of heterozygosity excess. Could this be an indication that the RAD data is bad--that there's something wrong with the library/sequencing? Is there something I can do to evaluate this possibility and gain trust in my data?

I performed de novo assembly in dDocent followed by the recommended filtering steps.

Thanks for any insights.

[Russell P]

Also, I used a fairly conservative c-value of 0.94.

And it's not just a few loci causing the negative overall Fis--*all* of the loci are negative (some as high as -1).

[Russell P]

The lab that generated the dataset said that " We noticed some regions got enormous number of reads with coverage >30K, and these regions look like repetitive sequences. "

I read that excessive heterozygosity can be due to repetitive DNA, so I wonder if the majority of the reads are of repetitive DNA?