[AI for the Sustainable Future] Biodiversity Conservation
As conserving biodiversity on Planet Earth for a harmonious, optimal ecosystem is one of the sustainability goals (probably the ultimate one), AI can play a crucial role.
The primary reason we are concerned about climate change is its detrimental impact on Earth’s ecosystems, which pose both immediate and indirect threats to human life. Moreover, the damaged ecosystems further accelerate climate change. The process of ecosystem management has become an extremely complex mechanism that needs to take countless variables into account, which are challenging for the human mind to fully grasp.
AI can assist in developing conservation strategies to protect biodiversity and restore ecosystems, by monitoring environmental changes and analyzing vast amounts of data, eventually providing essential insights for effective ecosystem management.
Holistic biodiversity conservation aided by AI
Biological diversity, or biodiversity, refers to all life on Earth – from microbes and wildlife to rainforests – and their interactions with each other. Biodiversity plays a vital role in ecosystems, providing key functions such as buffering against climate extremes, regulating hydrological cycles, protecting soils, and moderating temperatures in urban areas. It also contributes to reducing food insecurity. Currently, biodiversity faces significant threats, primarily due to human activities. Furthermore, both climate change and biodiversity loss negatively affect each other, creating a vicious cycle. (For more about biodiversity, refer to the following article from UNFCCC.)
Why Biodiversity Matters | United Nations Framework Convention on Climate Change (UNFCCC)
Speaking of the UN, I recently discovered an amazing website operated by the organization. The UN Biodiversity Lab, an initiative powered by an interactive online platform, provides access to global data layers, analytics, and tools to support countries in understanding their biodiversity, leading to informed decisions for conservation. In partnership with UN agencies, Microsoft, and NASA, this joint initiative targets policymakers, researchers, and conservationists, facilitating collaboration and the sharing of insights for effective biodiversity management.
I explored some of the key features myself, and found the spatial data tools and custom mapping features particularly fascinating. This initiative and online platform exemplify and visualize the power of data and technology in biodiversity conservation.
While technologies have already been involved in efforts to protect biodiversity, AI can further enhance these practices. Here’s a noteworthy article.
How can we use AI to monitor biodiversity and support conservation actions? | AI for Good
The article discusses the innovative use of AI and machine learning in biodiversity science, particularly through Professor Andrew Gonzalez's proposed 'detection attribution framework,' a multi-step process designed to understand the impact of human activities on biodiversity and ecosystems. It comprises five key steps: 1) causal modeling, 2) observation, 3) estimation, 4) detection, and 5) attribution.
For 1) causal modeling, AI helps understand cause-and-effect relationships in ecosystems, through structural equation models1 and recurrent neural networks2. In the 2) observation step, machine learning algorithms analyze large datasets from satellites, sensors, and cameras to identify changes in species distribution, abundance, and traits. During 3) estimation process, AI technologies create predictive models of future biodiversity changes based on current observations and human activity scenarios, estimating impacts of climate change and other factors. For 4) detection, AI aids in understanding and detecting change signals across large areas. At 5) attribution stage, AI helps identify causes of observed biodiversity changes, including human activities like pollution, using statistical models and causal inference methods.
In a nutshell, biodiversity conservation and ecosystem management has become a near-impossible mission due to its extreme complexity, involving overwhelmingly numerous variables ranging from intrinsic factors to external ones like climate change. Simple math doesn’t work for this daunting task that requires a holistic approach. Even the best and brightest mathematicians and data scientists are not enough; we need a little(?) help from AI.
Specific practices: wildlife protection and forest management
While biodiversity is about all life on the planet — including microorganisms — and an ecosystem should be viewed as a whole, as discussed earlier in this post, we also need to pay attention to the unique challenges that fauna and flora are facing: wildlife poaching and deforestation in particular.
According to the DW video presented above and an article by the World Wide Fund for Nature (WWF), poachers kill over 20,000 African elephants for ivory every year. A Dutch nonprofit organization named Hack the Planet has developed an ‘AI camera trap’ system to detect poachers and sound an alarm in real time. Unlike legacy camera traps, which were not very effective, the smarter AI camera trap can immediately recognize whether a detected object is an elephant or a human, using machine learning algorithms, and send real-time alerts to rangers. Operating via satellite, the system can function anywhere on the globe without relying on GSM or WiFi networks. Countries like Gabon, Zambia, and Zimbabwe have already implemented these AI camera traps.
Meanwhile, the following article presents a case of an AI tool fighting against deforestation.
Could AI save the Amazon rainforest? | The Guardian
The Guardian article introduces PrevisIA, an AI platform created by an environmental nonprofit organization named Imazon, aimed at predicting and preventing deforestation in the Amazon. Triunfo do Xingu, the targeted area, has been a biodiversity hotspot under threat, which is also home to many endangered, vulnerable and near threatened species. PrevisIA developed short-term deforestation prediction models, based on geostatistics and historical data, and leverages AI algorithms to analyze satellite images. With support from Microsoft and Fundo Vale, Imazon has gained the computational power that PrevisIA needs.
PrevisIA's adoption by regional authorities marks a proactive step in forest conservation. For instance, it alerts monitoring agencies to high-risk areas, enabling closer surveillance and even prompting prosecutors to warn property owners that they will be held accountable for any potential deforestation. Isn't it interesting?
Ecosystem management as training for terraforming?
Before wrapping up, let me discuss a wild idea: could our current ecosystem management practices be a stepping stone to future terraforming?
Terraforming, as you know, is the hypothetical process of transforming another planet into a human-habitable environment. Some tech moguls seem to view Earth as nearing its expiry date, willing to move on to new space colonies. But here's a thought – if we've struggled to manage our own planet, what guarantees success on a new one?
Imagine this: our endeavors in ecosystem management might just be the 101 class in an interstellar terraforming curriculum. Concepts like smart farming and advanced urban planning (the topics I’ll discuss in the upcoming posts) aren't just for Earth – they might be rehearsals for space colonization.
In that sense, stay tuned for more to come in this series!
Structural Equation Model (SEM): a statistical technique that enables researchers to understand complex relationships between multiple variables, figuring out how different factors (like environment, behavior, or genetics) are interconnected and mutually influential. SEM is particularly useful in the fields where relationships between variables are complex and interrelated.
Recurrent Neural Network (RNN): Unlike other neural networks, RNN has a 'memory' of previous data in the sequence, which influences what they learn from the new data. This makes them especially good for tasks where understanding the order and context of data is important, like language translation or speech recognition.