Students' beliefs about the extent to which meaningful others, including their peers, recognize them as a strong science student are correlated with their persistence in science courses and careers. Yet, prior work has found a gender bias in peer recognition, in which student nominations of strong peers disproportionately favor men over women, in some instructional science contexts. Researchers have hypothesized that such a gender bias diminishes over time, as determined by students' academic year: studies have found a gender bias in peer recognition in science courses aimed at first-year students, but not in science courses aimed at beyond first-year students. This hypothesis that patterns of peer recognition change over time, however, has yet to be tested with longitudinal data--previous studies only examine snapshots of different students in different science courses. In this study, we isolate the effect of time on peer recognition by analyzing student nominations of strong peers across a two-semester introductory physics course sequence, containing the same set of students and the same instructor in both semesters, at a mostly women institution. Using a combination of social network analysis and qualitative methods, we find that while many students receive similar levels of peer recognition over time, the four most highly nominated students--the recognition celebrities--exhibit some change between semesters even in this highly controlled setting. Furthermore, we observe that these changes in the celebrities track closely with changes in student outspokenness and that being outspoken is likely more important for gaining recognition than earning a high grade in the class. These findings lend support to prior work's hypothesis that peer recognition changes over time, but also challenge the generalizability of previous results (i.e., that patterns of recognition are related to students' academic year). Instead, peer recognition seems highly sensitive to variables such as individual students' participation and, therefore, may be course specific. We provide recommendations for both when and how instructors may intervene on peer recognition based on our results.