Rewriting the ­Anthropocene with ­citizen ­science

When I began my stay at the TUM-IAS in 2022, the plan for Citizen Scence Nature Watch looked ambitious yet deceptively straightforward. The idea was to bring together datasets collected by citizen observers of nature (citizen science or CS) to better understand seasonal “match-mismatch”: the growing discrepancy between key events in the lives of plants and animals as the climate changes. We wanted to know whether spatial diversity and complex food webs can soften the negative consequences of these shifts. A further, genuinely pioneering goal was to use CS data not only to monitor wildlife, but also to analyze human behavior in nature and build a kind of “digital twin” of the relationship between people and the natural world. Our initial plan was to combine classical records of seasonal events with social media data to show how people use natural areas, where conflicts arise, and how we might manage ecosystem services more sustainably, from recreation to biodiversity conservation.

Very early on, it became clear that the COVID-19 pandemic had left a distinct imprint in CS data. During lockdowns people went out into the field less often. For wildlife, this meant a brief “breathing space”; for our datasets, unusual patterns in reported observations. We examined this effect globally and related it to key features of species’ life cycles and conservation status. In practice, this meant looking beyond classical ecological tools and drawing on methods used to understand social behavior and changing patterns of nature use.

An AI revolution

Over the three years, our work evolved far more than we had anticipated, largely due to the AI revolution. When we started, large language models (LLMs) such as ChatGPT and Grok were still in their infancy. Within a short time, they became indispensable. Thanks to them, we could process and connect enormous, unstructured datasets from citizen science platforms (iNaturalist, eBird, and local portals) with posts from Instagram, Twitter/X and nature forums. Tasks that once required months of manual cleaning could now be done within days. As a result, we revised our vision several times, moving from classical seasonal ecology toward a hybrid approach: ecology + data science + the human dimensions of wildlife.

The power of long-term data

The key lesson is the decisive importance of long-term data series collected through CS. Only observations carried out systematically over decades, often by the same individuals or communities, allow us to detect subtle yet ecologically meaningful shifts:

• in the timing of seasonal events in plants and animals
   
• in anti-predator behavior
   
• in the structure of observer communities
   
• in how people use space and interact with wildlife.

Short-term projects, however intensive, remain snapshots. They rarely allow us to separate the true signal of change from the “noise” of seasonal variation and local context. Our capacity to recognize emerging trends before they become crises lies in these patiently assembled long-term datasets.

Challenges

CS data are extremely heterogeneous, and this diversity proved to be one of the project’s greatest strengths. In our analyses, we combined: 

• specialized platforms for birds and seasonal events (e.g., eBird, Tatarstan archives reaching back to 1988)
   
• global opportunistic databases (iNaturalist, Pl@ntNet) 
   
• national and regional monitoring schemes of butterflies, dragonflies, and roe deer (including German hunting databases)
   
• unstructured social media materials (geotagged Instagram photos, Facebook groups)
   
• analog archives digitized by volunteers (field notebooks from the 1970s onward) data from community-run camera traps and acoustic recorders. 

This mosaic required new processing pipelines, from standard statistics to deep learning and large language models that could automatically clean, classify, and integrate millions of records with different structures and levels of certainty. In practice, the main challenge was not a lack of data, but their abundance and fragmentation – and the need to fuse them into a coherent ecological and social picture.
 

What have we learned from this mosaic?

This multi-scale, multi-source strategy led to a series of studies that advanced ecological understanding and set new methodological standards for working with CS data:

Dragonfly adaptation to human presence. On wetlands and in city parks, dragonflies are gradually becoming “accustomed” to people. Where human presence is highest, their flight-initiation distance shortens, suggesting that behavioral flexibility may help invertebrates persist in urbanized landscapes.

Solving the puzzle of roe deer births. We helped clarify why birth timing shifts in roe deer: The growing season length in the year of conception determines when embryonic diapause ends, allowing this species to track changing spring rhythms.

Cuckoos, hosts, and climate-driven mismatch. Long-term series revealed growing asynchrony between two sympatric cuckoo species, their hosts, and insect prey – a subtle but potentially far-reaching cascade effect of climate change.

The observers themselves. Rare birds now attract dozens or hundreds of birders within hours. This trend is accelerating, with more women present in the crowd, signaling social change within the nature-observing community.

Shyness, cities, and counting birds properly. Urban bird detectability depends on individual and species-level “shyness” toward people, meaning standard city bird-count methods may need behavioral correction.

Each story began with different data – from roe deer anatomy and dragonfly escape distances to lots of geotagged photos – yet all share a common thread. Only by integrating them into a long-term, multidimensional mosaic can we see the full picture of change. This is where the future of ecology in the Anthropocene lies.

Looking ahead – toward a digital twin of our living planet

We believe we stand at the threshold of a genuine revolution in how humanity understands and co-exists with the rest of nature. Soon, a living, open, global digital twin of human-nature relations may emerge: a real-time “mirror of the Earth” weaving together millions of CS observations, social media images, camera streams, IoT data, and satellite imagery. Co-created by millions and powered by advanced AI, it will be a dynamic system rather than a static model. CS will no longer be a pleasant “add-on” to academic research. It will become a fully recognized, democratic pillar of science – the largest and most inclusive laboratory in human history. Through the marriage of long-term observation and modern technologies, we gain something priceless: foresight. We will be able to detect tensions between people and nature long before they erupt into crises, and to prevent rather than merely repair. This is no longer science fiction. It is the direction in which we are already moving, together with thousands of nature observers, with AI, helping to write a more harmonious chapter of the Anthropocene.  

Selected publications

• J. Kauffert, C. Ehrmantraut, P. Mikula, P. Tryjanowski, A. Menzel, A., and A. König, “Matching the green wave: growing season length determines embryonic diapause in roe deer,” Proc. Biol. Sci., vol. 292, no. 2047, pp. 20242903, 2025.
   
• P. Mikula, O. V. Askeyev, A. O. Askeyev, I. V. Askeyev, F. Morelli, A. Menzel, and P. Tryjanowski, “Climate change is associated with asynchrony in arrival between two sympatric cuckoos and both host arrival and prey emergence,” R. Soc. Open. Sci., vol. 11, no. 1, pp. 23169, 2024.
   
• P. Mikula, P. Czechowski, A. Dubicka-Czechowska, L. Jerzak, A. Menzel, A., and P. Tryjanowski, P., “Understanding antipredator strategies of insects: Human presence and escape behaviour in Odonata,” Ecological Entomology, vol. 50, pp. 682–692, 2025.
   
• P. Mikula, F. Morelli, A. Menzel, and P. Tryjanowski, “Urban birds’ detectability is affected by inter- and intraspecific variation in shyness,” Ibis, 2025, e-publication ahead of print.
   
• P. Tryjanowski, Ł. Jankowiak, P. Mikula, P. Czechowski, A. Menzel, and M. Polakowski, “What factors affect the ‘flocking’ of birdwatchers during bird rarity observations?” People and Nature, vol. 6, no. 6, pp. 2390–2398, 2024.

Full list of publications