Jindong Zhang

Over billions of years, evolution has produced an immense diversity of biological structures, materials, mechanisms, and behaviours. Biomimetics draws on this diversity by analysing biological principles and transferring them into technical applications. Yet a comprehensive understanding of how biomimetic research accesses biological knowledge was still lacking. This thesis investigates which organisms biomimetic research draws on, who conducts such research, and which tools support it. It then developed methods to make the discovery and selection of biological models more systematic. The first part of the thesis diagnosed the field along three dimensions – biological, human, and methodological dimensions. An analysis of more than 74,000 biomimetic publications identified 31,776 biological models. The field relies mainly on animals and on a limited set of repeatedly used organisms; fewer than 23% of models are specified at species level; more than 90% of publications with an identifiable model refer to a single organism; and new taxa are added more slowly than the field grows. An analysis of 436,010 author-affiliation records across biomimetic publications showed that technical researchers outnumber biological researchers by about two to one. Interdisciplinary teams have grown more than twenty-fold since 2004 and are associated with higher citation performance; a broader, taxonomically more precise use of biological models is associated with biology expertise specifically. A total of 104 tools were identified, 24 of which were fully functional. Comparison of these 24 biomimetic tools against ten criteria revealed that few provide dedicated support for systematic biology-push discovery of unfamiliar biological models or for the explicit comparison of candidate models. The second part of the thesis developed methodologies to address these gaps. A data-driven, phylogeny-aware framework treats biological model selection as an explicit comparison across data sufficiency, prior biomimetic use, phylogenetic characteristics, and project-specific constraints. In a microplastic-filtration case study with 35 suspension-feeding taxa, the developed framework distinguished candidates that remained competitive under different priorities from those whose suitability depended on context. A complementary discovery framework organises biological literature into functional atlases that show which taxa use which mechanisms to perform a given function, how these occurrences are distributed across phylogenetic lineages, and whether biomimetic research has already engaged them. In two pilot domains, filtration (257 taxa across 21 phyla) and attachment (838 taxa across 24 phyla), only 6% and 15% of taxa, respectively, had been engaged by biomimetic research. Together, this thesis provides an empirical and methodological basis for engaging biological diversity more strategically.

Biomimetics is generally viewed as inherently interdisciplinary, yet the disciplinary composition of biomimetic research teams has not been systematically quantified. We analysed 74 407 biomimetics-related publications from the past two decades and classified 436 010 author–paper affiliation records by biological or technical expertise to reconstruct team composition. Our analysis reveals an asymmetric disciplinary landscape: technical authors outnumber biological researchers by roughly two to one. Team organization is more balanced—while purely technical teams still represent the largest share of publications, interdisciplinary Bio–Tech teams have expanded over 20-fold since 2004 and now nearly match purely technical teams in annual output. Pathways to interdisciplinarity vary geographically: countries achieve Bio–Tech collaboration through distinct organizational models, from domestic multi-institutional networks to international partnerships. Interdisciplinary teams also achieve higher citation performance than teams composed exclusively of biological or technical researchers. Team composition is further associated with biological models used for inspiration: purely biological teams draw on a broader range of taxa, whereas studies involving technical researchers rely more often on a narrower set of conventional organisms and typically report their biological models with lower taxonomic specificity. These findings map the team architecture of biomimetics and support more targeted approaches to interdisciplinary collaboration.

Selecting suitable biological models remains one of the most challenging and least formalised steps in biomimetic design. Existing tools support searching for biological strategies and transferring principles into engineering concepts, but provide limited guidance on which organism to prioritise once multiple plausible candidates are identified. Here we present a phylogeny-informed, data-driven framework that structures model selection as an explicit comparison. The framework computes four module scores per candidate model: data sufficiency, innovativeness, phylogenetic characteristics, and an open-ended contextual module for project-specific constraints. Users assign weights to the modules to reflect resources, timelines, and design requirements, yielding an overall compatible score and a ranked shortlist. We demonstrate the framework in a microplastic-filtration case study using 35 suspension-feeding taxa and three archetypal user scenarios. Rankings reveal a compact set of candidates that remains competitive across scenarios, while a smaller subset shifts in response to changes in priorities, thereby distinguishing robust starting points from context-dependent opportunities. Overall, the framework fills an under-supported gap in current biomimetic workflows by providing a structured decision-support layer for biological model selection.

As ecological collapse accelerates under the pressures of anthropogenic climate change, adaptation strategies increasingly include technological proxies for nature’s functions. But can ecosystem services (ES) be meaningfully replaced by technology? Revisiting this urgent question first posed by Fitter (2013), we assess the extent to which bio-inspired design—particularly biomimetics—has advanced the capacity to support, enhance, or replace natural ES. We convened an interdisciplinary team to synthesize and refine a comprehensive list of 22 ecosystem services, integrating often-overlooked cultural and relational dimensions. Using this framework, we conducted a large-scale analysis of over 68,000 peer-reviewed publications from the biomimetics and bio-inspired design literature between 2004 and 2025, applying AI-assisted classification to evaluate whether, and how, these technologies map onto specific ES functions and benefits. Our findings reveal both promise and profound limitations. Bio-inspired research engages with 20 of the 22 ES, but over 78% of this work concentrates on five technologically tractable functions—biochemicals, disease regulation, waste treatment, fibre/hide/wood, and fuel. Foundational supporting and regulating services such as pollination, soil formation, and nutrient cycling are almost entirely absent. Moreover, only 3% of technologies described in the academic literature aim to support existing systems; the overwhelming emphasis on enhancement (39%) and replacement (58%) suggests a design paradigm skewed toward substitution rather than coexistence. Intangible, co-produced services—particularly those related to culture, identity, and meaning—remain outside the current reach of biomimetic design. This skew reveals a dangerous imbalance: while certain ES can be technologically approximated, the relational, emergent, and systemic qualities of ecosystems elude replication. Technological replacement must not become a substitute for preservation. Instead, bio-inspired design should be mobilized as a tool for adaptation that amplifies and protects the living systems on which human and more-than-human futures depend.

Ecosystem services are crucial for animals, plants, the planet, and human well-being. Decreasing biodiversity and environmental destruction of ecosystems will have severe consequences. Designing technologies that could support, enhance, or even replace ecosystem services is a complex task that the Manufactured Ecosystems Project team considers to be only achievable with transdisciplinarity, as it unlocks new directions for designing research and development systems. One of these directions in the project is bio-inspiration, learning from natural systems as the foundation for manufacturing ecosystem services. Using soil formation as a case study, text-mining of existing scientific literature reveals a critical gap: fewer than 1% of studies in biomimetics address soil formation technological replacement, despite the rapid global decline in natural soil formation processes. The team sketches scenarios of ecosystem collapse, identifying how bio-inspired solutions for equitable and sustainable innovation can contribute to climate adaptation. The short communication opens the discussion for collaboration and aims to initiate future research.

Biomimetics draws inspiration from biological organisms, yet only a small fraction of Earth’s biodiversity has been explored for innovation. This study examines the biological models used in biomimetic research. Using GPT-4o, we analyzed 74,359 publications and identified 31,776 biological models, revealing distinct taxonomic distribution patterns. Our findings highlight a reliance on a narrow set of animal taxa; fewer than 23% of identified models were resolved at the species level—corresponding to 1,604 species; and broad taxonomic classifications (e.g., phylum, class) were more frequently cited. Despite the rapid growth of biomimetics, the exploration of new model taxa falters. Researchers tend to focus on one model per study, potentially limiting the field’s capacity to leverage evolutionary insights. To promote diversity and innovation in biomimetics, we advocate for stronger collaboration with biologists to integrate underutilized yet well-researched taxa, specify biological inspirations at the species level to enhance evolutionary insights, and—where appropriate—incorporate multiple models, enabling the use of comparative methods.

Biomimetics as the transdisciplinary field leveraging biologically inspired solutions for technical and practical challenges has gained traction in recent decades. Despite its potential for innovation, the complexity of its process requires a deeper understanding of underlying tasks, leading to the development of various tools to aid this process. This study identified an inventory of 104 tools used in biomimetics, of which 24 have been classified as fully accessible, functional, and ready-to-use biomimetic tools. Additionally, it provides definitions and evaluation criteria for biomimetic tools, offering a structured approach to tool assessment. The 24 tools have been assessed based on ten criteria in a qualitative and quantitative analysis yielding an overview of their typology, accessibility, stage of development, and other key characteristics. Patterns of the typology development of tools over time revealed a trend towards integrating computational methods and artificial intelligence, thereby enhancing the tool’s functionality and user engagement. However, gaps in tool functionality and maturity, such as the lack of tools designed to support technical processes, the absence of tools tailored for solution-based approaches, and insufficient evidence of successful tool application, highlight areas for future research. The study results underscore the need for empirical validation of tools, and research into the effectiveness of holistic tools covering multiple stages of the biomimetic process. By addressing these gaps and leveraging existing strengths, the field of biomimetics can continue to advance, providing innovative solutions inspired by biological models.