A simple method has been applied to extract 11-year running mean weighted δfoF211y and δfoF111y variations from June foF2 and foF1 monthly medians at Slough/Chilton and Juliusruh including observations in 2015. The geomagnetic control was shown to be valid. The dependence on geomagnetic activity has become more pronounced and explicit after 1990. The retrieved from routine foF1 ionosonde observations thermospheric parameters over the period of » 5 solar cycles were used for the first time to analyze the mechanism of foF1 and foF2 long-term variations. It was shown that the control was provided via two channels: [O] and [O]/[N2] variations. Geomagnetic activity controls the (O/N2 )11y ratio, while solar activity presented by (F10.7)11y controls [O]11y variations. Atomic oxygen manifests solar cycle and long-term (for some solar cycles) variations with the rising phase in (1965-1982) and the falling phase in (1982-2003). The empirical model MSIS-86 driven by Ap and F10.7 manifests [O]11y and (O/N2 )11y variations similar the retrieved ones including the period of deep solar minimum with a very low [O] in 2008. This confirms the basic idea that daytime mid-latitude foF1 and foF2 long-term variations had a natural (not anthropogenic) origin related to long-term variations in solar and geomagnetic activity.