Tweedie family vs Negative Binomial - which is best?

Hello everyone!
Newish in Rstudio.

I am working with abundance data (pelagic fish) and I did some models per area so I could use it to predict abundance at age 0.
I used a negative binomial family. One of my variable in the models uses a smooth.

Negative binomial is good to deal with overdispersion, but tweedie can also do that right?
I wanted to understand if it is worth to test tweedie for my models, using a GAM. My data does not have a lot of zeros. Some areas do not have 0's at all.

Is it risky to go with tweedie?

Thank you!