Modeling fish populations: Are we yet to reach to the bottom of the methodological shoal?

Frequentists are often at loggerheads with Bayesians. One believes in frame relay, that is one scenario leading the other as per universal norms, whereas the other does laser show, keeping everything except the sources of change flexible. Though there is a wealth of work in stock assessment of fisheries using the relatively fresh Bayesian methods, I have a serious poser as regards the priori Dn assumptions. Usually robustness of Bayesian inferences are often directly regressed upon the non-informative priors. Can we really have this comfort when we employ Bayesian techniques for Fisheries research? In my opinion opting for non informative priors for stock defining parameters like carrying capacity etc would be too much of a risk to start with. Please post your opinion...