I do somewhat question polling questions that showed mostly stable results followed by one sudden spike (without an obvious precipitating event). I have no doubt that the general trend has been in this direction, but the sudden spike in the last couple years doesn't feel accurate at all.
The pandemic did do a number on religious practice; perhaps that affected self-identification as well?
I don't think religion numbers were particularly down in immediate post-pandemic polling, though. Although it's difficult to say for sure, my best guess would be a Dobbs effect.
Look at the dates on the chart again. The most recent numbers are from 2021. If they were from 2022 then immediate-term
Dobbs backlash would be an even stronger explanation, yeah.