Scientific authorship norms vary dramatically across disciplines, from contribution-sensitive systems where first author is the greatest contributor and subsequent author order reflects relative input, to contribution-insensitive conventions like alphabetical ordering or senior-author-last. We develop evolutionary game-theoretic models to examine both how these divergent norms emerge and their subsequent effects on collaborative behavior. Our first model reveals that contribution-insensitive norms evolve when researchers who sacrifice positional advantage face the strongest adaptive pressure -- for example senior authors managing larger collaboration portfolios or bearing heavier reputational stakes. This "Red King" dynamic potentially explains why fields in which senior researchers command large labs, major grants, and extensive collaboration portfolios may paradoxically evolve conventions that favour junior-author positioning. Our second model demonstrates that established norms influence researchers' willingness to collaborate, with contribution-sensitive norms consistently outperforming insensitive alternatives in fostering successful partnerships. Contribution-insensitive norms create systematic coordination failures through two mechanisms: "main contributor resentment" when exceptional work goes unrecognized, and "second contributor resentment" when comparable efforts receive unequal credit. These findings suggest that widely adopted practices like senior-last positioning and alphabetical ordering may function as institutional frictions that impede valuable scientific collaborations rather than neutral organizational conventions, potentially reducing overall scientific productivity across affected disciplines.