The world faces unprecedented biodiversity loss due to climate change, habitat loss, and overexploitation. Unsustainable wildlife trade contributes significantly, driving extinctions and population declines in species like pangolins. Effective management is crucial, particularly considering links between wildlife markets and disease outbreaks like COVID-19. However, current approaches often rely on historical data, lacking proactive foresight. Wildlife markets are dynamic, with constant innovation in commercializing new species or products and adapting to trade bans or regulatory changes. Businesses create, adapt to, and predict market shifts, offering valuable insights into future trade developments. Innovation helps extract value from established markets through technological advancements, new products, or business models. Patent data reveal innovations, as businesses seek patent protection for commercially viable innovations. Manual patent analysis has been used in some contexts; this research expands this approach using large-scale scraping and machine learning. The study examines patents (1970-2020) related to six commercially traded taxa: bears, rhinoceroses, caterpillar fungi, pangolins, horseshoe crabs, and sturgeon, encompassing diverse product forms, geographies, and regulations. The hypotheses explore how product innovation leads to market differentiation, how regulation shapes innovation (compliance or avoidance), and how changing patent patterns reveal business perspectives on wildlife markets. The study period includes the CITES implementation in 1975, a time of rapidly growing global interest in wildlife trade and increased patent filings globally.

Existing literature highlights the unsustainable nature of wildlife trade and its contribution to biodiversity loss. Studies have linked wildlife trade to extinctions and population declines of various species, including pangolins. The connection between wildlife markets and zoonotic disease outbreaks, like COVID-19, underscores the need for robust management. While some research offers insights into emerging trends in wildlife trade, the dynamic nature of these markets, driven by innovation and adaptation, necessitates proactive approaches. Previous studies have employed manual patent analysis to understand aspects of resource access and benefit-sharing and diversifying uses of certain species. However, a large-scale, machine learning-driven analysis of patent data is novel and offers the potential to provide more comprehensive understanding and early warnings.

The study used a multi-step methodology. First, a shortlist of six commercially traded taxa was selected, representing various product types, geographies, and regulatory statuses. Keywords related to these taxa in English and relevant languages were identified and refined through manual checking of a sample of patents. The Google Patents database was then scraped using the selenium library in R, extracting data on patent filings, including filing entities, authors, priority dates, abstracts, and full descriptions. Data cleaning removed duplicates and irrelevant results. To analyze patent filing trends, the year-on-year percentage change in patent filings from 1985 to 2020 was calculated, comparing the rates for the focal taxa against global patent filing rates. Bayesian changepoint analysis using Stan identified significant shifts in monthly patent filing rates for each taxon. Latent Dirichlet Allocation (LDA) topic modeling was employed to identify co-occurring themes in patent abstracts, providing insights into product diversification and innovation. Finally, a thematic coding scheme, integrating LDA topics and the Economic Botany Data Collection Standard, classified patents into broader topics (medicine, food, agriculture, cosmetics) and specific subcategories (production, preparation, detection). The manual classification was refined iteratively.

The analysis revealed substantial and increasing commercial innovation for both legal and illegal wildlife products. A total of 27,308 patents were identified for the six focal taxa, showing a mean annual increase of 104-149%, exceeding the global background rate of 104%. Patent filing rates for all taxa increased significantly over time, peaking between 2013 and 2017. Changepoint analysis identified taxon-specific shifts in patenting rates, often correlating with events like the legalization of traditional Chinese medicine patenting in 1993 or regulatory actions such as trade bans. Importantly, trade bans did not consistently lead to reduced patenting; for example, rhinoceros horn patenting continued to increase after a 1977 international trade ban. Innovation resulted in product diversification, including new uses of existing products (e.g., rhino horn snuff, pangolin scales in livestock feed) and the development of farming methods and synthetic alternatives for various taxa. While the number of patents related to established uses often exceeded those for novel ones, some novel products showed sudden increases in patenting. The proportion of patents focused on farming or synthetic alternatives varied across taxa.

The findings demonstrate substantial and persistent commercial interest in wildlife products, even for threatened and illegally traded species. The lack of correlation between trade bans and reduced patenting highlights the complexities of regulating wildlife trade. Businesses may bet on future market relaxations or aim to commercialize products using illegally acquired materials. Patent data provide a valuable early warning system, identifying emerging products years before they reach the market. This allows for proactive interventions to change consumer or industry behavior and inform enforcement efforts. The study's approach can be extended to broader sets of species or by focusing on specific product categories to improve identification of emerging trends.

This study demonstrates the value of patent data in anticipating and addressing trends in unsustainable wildlife harvesting. The continued high level of patenting of products from species subject to trade bans shows confidence in future market reopening, while the focus on farming and synthetic alternatives reflects commercial perceptions of the future of markets where wild products are illegal. Ongoing monitoring of patent data can identify emerging threats and inform proactive interventions.

The study's reliance on patent data means that not all traded wildlife will be captured. Further, not all patented products are commercialized. Triangulation with other data sources (e.g., market data, consumer behavior studies) would strengthen the analysis. The study's geographical scope is global but doesn't capture regional nuances within countries.