Using AI, we asked birds what qualities make a good tree

Using AI, we asked birds what qualities make a good tree

The southeastern region of Australia once had vast expanses of grassy box gum woodlands, but now less than 5% of this habitat remains. This loss of large old trees has had a devastating impact on the numerous bird and animal species that rely on them for shelter.

While it is difficult to replace these centuries-old trees, one solution is to create artificial structures that mimic the features of large old trees in areas where trees cannot grow or are too young and small. In collaboration with the Australian Capital Territory Parks and Conservation Service, we have been working on this in the Molonglo region of Canberra.

To design these artificial structures, it is important to understand what makes good habitat from an animal’s perspective. To achieve this, we developed ways to use AI and machine learning to involve non-human stakeholders, such as birds and trees, in the design process. Essentially, we treated large old trees as lead designers and birds as discerning evaluators of their work.

The Molonglo region, which was once a thriving ecosystem, has become fragmented and damaged. Large old trees are now rare, posing a challenge to fill the gap left by their absence. Previous attempts to introduce artificial habitat structures, such as modified utility poles and relocated dead trees, have been made, but they fail to replicate the complex canopy structures provided by large old trees.

By carefully analyzing imagery and data, we were able to identify the features preferred by birds. For example, birds showed a preference for small horizontal branches for perching and nesting. Armed with this knowledge, our next step was to use AI and machine learning to design better habitat structures.

Our process involved capturing data, creating predictive models, and engaging in iterative design. AI and machine learning played a crucial role in interpreting complex spatial data. We mapped each tree using laser scanning technology and algorithms to measure attributes like branch orientation and size. By understanding bird preferences for these attributes, we could inform the design of artificial replacements.

We also developed statistical models based on long-term observations of bird interactions to predict bird behavior. Simulating how birds might use artificial branches allowed us to refine our designs to better meet their needs.

To generate a variety of artificial tree crowns, we developed additional algorithms. Instead of relying on human judgment, we used our bird behavior model to determine how these structures would serve avian inhabitants. Our goal was to create lightweight structures that are easy to install, reconfigure, and remove. Our simulations showed that these structures could significantly increase habitat suitability compared to utility poles and snags.

Currently, we are building prototypes based on our designs, but the final step will be field testing to gather feedback from the birds themselves. Birds can provide valuable insights on the characteristics of artificial structures through their interactions with them. This testing will help us further improve the designs.

Our findings highlight the importance of including non-human stakeholders in the design process. Currently, design processes are dominated by human perspectives and expertise. By broadening the scope and incorporating input from other species, we can enhance the design process and create more inclusive and resilient environments for all lifeforms.

It is crucial to remember that while artificial structures can help, there is no true substitute for large old trees. We must continue to preserve existing trees and plant more for the future.

The principles of more-than-human design we applied in Canberra have broader implications. Many environments worldwide face similar challenges, and by rethinking current approaches to design and planning, we can create more inclusive and resilient habitats for various species.

By treating other species as innovators and expert participants in design, we can unlock a wider range of possible designs. This participatory approach, which incorporates AI and input from non-human lifeforms, can help overcome human biases and lead to better solutions.

In a world grappling with urgent environmental crises, innovative and inclusive design approaches are essential. Trees and birds are already excellent designers and judges, respectively. By including their input, we can create better “more-than-human” designs that promote equity and sustainability for all beings.

We would like to acknowledge the contribution of Darren Le Roux in researching and installing artificial habitat structures to support biodiversity.