The mosaic of science: Disciplinary diversity and scientific prestige in research groups in Colombia.

Collaboration among science teams is essential for addressing complex global challenges. A key feature of such collaboration is disciplinary diversity; however, its relationship with team performance remains debated. Existing research has focused primarily on high-income countries and has relied on proprietary databases, often overlooking the distinctive scientific ecosystems of middle- and low-income nations. This geographical and methodological bias has created a gap in understanding how team composition affects scientific outcomes in these underrepresented contexts. This study examines a ten-year period using publicly available data from all Colombian research groups maintained by the Ministry of Science, Technology, and Innovation (MinCiencias). Disciplinary diversity was measured using the DIV indicator proposed by Leydesdorff et al. We show that the relationship between disciplinary diversity and scientific prestige is non-linear and moderated by both group size and broad disciplinary area. Our analysis identifies two main findings: low diversity consistently characterizes research groups with a declining performance trajectory, and groups that advance in national rank exhibit a statistically similar diversity structure to those following a volatile trajectory in the national ranking. These results challenge the assumption that increasing diversity necessarily leads to better performance. Instead, they indicate that the functional role of diversity is not monotonic and that an optimal, context-specific level may exist. This nationwide study contributes to science policy by demonstrating that fostering field-dependent diversity structures, rather than maximizing diversity indiscriminately, may be critical for strengthening integrative and transformative research systems in emerging economies.