Hi everyone and many thanks for your responses.

I had already checked on differences between table schemas, indexes, data, and also on wait/locks/etc. but didn't find any differences. I ran a update statistics statement in both environments with sample of 100%. I checked index fragmentation, which is about 5% on both servers. Any other checks I can do?

One more question, based on the xml data I posted, it looks like the staging environment assigns a lot more RAM for the same query, isn't that correct? Could that be the reason why?