Biodiversity discourse and political transformation: A topic modeling approach to German Bundestag debates

Biodiversity is inherently embedded in complex relationships with other sustainability goals, particularly the SDGs (Allen et al., 2019), and must therefore be addressed within context to identify nuances. Although certain conceptualizations, such as framing biodiversity as natural capital, receive greater political and business acceptance (Bateman & Mace, 2020), tracing the development of biodiversity-related formulations in political discourse remains essential. As policy changes culture (Bau, 2021), decomposing political language is critical for understanding the frames that may accelerate public and institutional acceptance of biodiversity goals.Building on the expanding literature of discourse analyses in sustainability science (Damiens et al., 2021; Fischer et al., 2025; Kallis et al., 2024; Morrison et al., 2020), we advance Damiens et al. (2021) by examining biodiversity discourse across a broader corpus of political documents using machine-learning techniques. We treat political discussion as a stream of interacting topics and apply topic modeling and sentiment analysis (Chen & Mankad, 2025; Roberts et al., 2019) together with dynamic mode decomposition (Brunton et al., 2022; Schmid, 2010) to uncover the dynamics of biodiversity debate in the German Bundestag over the past 25 years. This approach reveals three temporal dimensions of biodiversity discourse: (1) oscillating, (2) steadily growing/decaying, and (3) non-stable growth/decay. It also allows detection of discursive silos, high-trending topics, and characterization of their associated sentiment.Our findings show that positively perceived topics typically involve mutually agreeable decisions supported by technological or economic innovation, aligning with political science theories on designing just transition policies (Bolet et al., 2024). We further align identified topics with the six modes of co-production for sustainability (Chambers et al., 2021) to assess how these modes appear in political debate. While technology-oriented climate actors often support less stringent climate policies (Burger et al., 2024; Caniglia et al., 2021), our analysis uncovers nuanced forms of welfare improvement, such as social innovation, that may more effectively advance biodiversity goals, consistent with pluralistic approaches to sustainable transformations (Caniglia et al., 2021).Finally, we examine politically contested “win-win” solutions, which remain difficult to agree upon despite their prevalence due to issue complexity (Hegwood et al., 2022).