7Engagement with scientific manuscripts is frequently facilitated by Twitter and other social 8 media platforms. As such, the demographics of a paper's social media audience provide a 9 wealth of information about how scholarly research is transmitted, consumed, and interpreted by 10 online communities. By paying attention to public perceptions of their publications, scientists can 11 learn whether their research is stimulating positive scholarly and public thought. They can also 12 become aware of potentially negative patterns of interest from groups that misinterpret their 13 work in harmful ways, either willfully or unintentionally, and devise strategies for altering their 14 messaging to mitigate these impacts. In this study, we collected 331,696 Twitter posts 15 referencing 1,800 highly tweeted bioRxiv preprints and leveraged topic modeling to infer the 16 characteristics of various communities engaging with each preprint on Twitter. We agnostically 17 learned the characteristics of these audience sectors from keywords each user's followers 18 provide in their Twitter biographies. We estimate that 96% of the preprints analyzed are 19 dominated by academic audiences on Twitter, suggesting that social media attention does not 20 always correspond to greater public exposure. We further demonstrate how our audience 21 segmentation method can quantify the level of interest from non-specialist audience sectors 22 such as mental health advocates, dog lovers, video game developers, vegans, bitcoin investors, 23 conspiracy theorists, journalists, religious groups, and political constituencies. Surprisingly, we 24 also found that 10% of the highly tweeted preprints analyzed have sizable (>5%) audience 25 sectors that are associated with right-wing white nationalist communities. Although none of 26 these preprints intentionally espouse any right-wing extremist messages, cases exist where 27 extremist appropriation comprises more than 50% of the tweets referencing a given preprint. 28These results present unique opportunities for improving and contextualizing research 29 evaluation as well as shedding light on the unavoidable challenges of scientific discourse 30 afforded by social media. 31 their line of study will generate such impacts [5]. For example, the importance of societal impact 51 is codified in National Science Foundation grant application guidelines, where funding 52 applicants must dedicate a substantial section of their proposals to describing the "potential [for 53their research] to benefit society and contribute to the achievement of specific, desired societal 54 outcomes" [6]. This distinction between scientific and societal impacts is relatively new: as noted 55 by Bornmann [4], for most of the 20th century, policymakers believed that any investment in 56 scientific research naturally led to positive societal impact, so the only relevant impact to be 57 evaluated was the degree to which a particular research output or individual researcher 58 advanced high-level academic knowledge. 59 60 However, a co...