Abstract
In the Northeast U.S., many stock assessments have a history of problematic model diagnostics, with multiple age‐based assessments recently being rejected in the peer review process, and are not suitable for management advice. The role in which diverging signals in the coastwide bottom trawl survey may be contributing to assessment problems was explored here for 18 stocks in the region. Specifically, trends in total mortality ( Z ) estimated from catch curve analysis and a relative measure of the harvest rate (total catch/survey index; called relative F ) were evaluated. Across stocks, relative F has declined over time, on average, since the mid‐1990s, yet Z has not for many stocks. Weak positive or even negative correlations between relative F and Z resulted for 13 stocks. This diverging signal appears to be contributing to assessment model performance, as larger retrospective patterns (a measure of assessment uncertainty) occurred for stocks with negative correlations between relative F and Z . While a variety of mechanisms could be involved in these diverging signals, the available evidence suggests that unreported catch and/or increasing natural mortality likely play a role to varying degrees for each stock.