ILka,

I've had this same problem. I think this stems from the number of scans that are to be scrubbed for that participant. I've found that despiking sufficiently improves the QC-FC relationships and the distribution of participants FC values. It's become increasingly clear to me that there is not a one-size-fits-all approach to denoising. I found this review helpful: 

However, as Alfonso mentions above, it important to take a look at how many scans are being scrubbed for each subject as a large proportion of scrubbed scans indicates substantial movement according to your scrubbing parameters. Satterthwaite et al., 2019, Motion artifact in studies of functional connectivity: Characteristics and mitigation strategies, Human Brain Mapping.

Andrew