Less than 5% of overgrown box gum forest once covered millions of square kilometres in south-eastern Australia, and the loss of old-growth trees poses a threat to many of the birds and other animals that live there.
Replacing this habitat won't be easy: there's no quick way to create centuries-old trees.
One thing we can do is create man-made structures that mimic the characteristics of large, old trees in degraded environments where trees cannot survive, or are too young and small. Working with the ACT Parks and Conservation Service, we are doing exactly this in the Morongo area of Canberra.
To build these man-made structures, we need to know what habitat looks good from an animal's perspective. To find out, we developed a way to use AI and machine learning to involve non-human stakeholders (in this case, birds and trees) in the design process. In essence, we enlisted big, old trees as lead designers and birds as discerning evaluators of the work.
Stanislav Rudavsky / Alex Holland
Trees, birds, electric poles
Morongo was once home to a thriving ecosystem that has now been fragmented and destroyed. Large, old-growth trees are becoming increasingly rare.
These trees, some of which are over 500 years old, form complex canopy structures that are essential for birds to nest, forage and roost in. As urban development increases and older trees die, the challenge is to fill the gaps left by these giant trees.
Stanislav Rudavsky / Alex Holland
In the past, alternative habitats have been introduced in this region using modified utility poles and transplanted dead trees (or branches). These structures can provide important habitat features, such as tall perches, nesting boxes, and bark, that are not found in planted saplings. However, it is very difficult to understand exactly which features of large, old trees are important to birds, limiting the value of artificial structures.
Careful analysis of images and other data allows us to identify these features: for example, we and our collaborators have found that birds prefer small, horizontal branches for perching and building nests.
By studying birds, we can learn about their preferences for certain features already engineered by trees. Our next challenge was to use this information to design better habitat structures.
Learning from trees
Stanislav Rudavsky / Alex Holland
We used a process that involved data capture, predictive modeling, and iterative design. AI and machine learning were essential to interpreting the complex spatial data.
First, we mapped each tree by bouncing millions of laser beams off every square centimetre of the tree's surface, capturing the canopy as a cloud of points. We then used algorithms to identify and measure important attributes, such as branch orientation, size and connectivity. A better understanding of birds' preferences for these attributes can help us design artificial replacements.
They then developed statistical models to predict bird behaviour. These models are based on long-term observations of bird interactions led by Philip Gibbons at the Australian National University. By simulating how birds use the artificial branches, they were able to refine their design to better meet the birds' needs.
Rethinking artificial habitats
Stanislav Rudavsky / Alex Holland
We further developed our algorithms to generate a range of artificial tree canopies: rather than judging the resulting designs by how much they resemble trees to the human eye, we used bird behaviour models to work out how these structures might benefit avian residents.
Our further goal was to create lightweight structures that are easy to install, reconfigure, and remove. Simulations showed that these structures could significantly improve habitat suitability compared to utility poles or dead trees.
Back on the scene
“We are currently building prototypes based on our design, but the final stage of this process will be field testing to find out what birds think. Through their interactions with the man-made structure, birds will be able to provide feedback on its properties. This testing will help us further improve our design.”
The design process for non-human stakeholders such as birds and trees is currently dominated by human perspectives and expertise. Our findings suggest that the design process can be improved by broadening the scope of creative contributions and judgments. The outcome of this design process can take the form of “continuous services” that provide shelter and other resources in a sustainable manner.
While we want to build better man-made structures, we must remember that there is no replacement for old, large trees. We must protect the ones we have now and plant more for the future.
Wide-ranging influences on design
The principles of Beyond Human Design used in Canberra also have broader application. Many environments around the world face similar challenges, and by rethinking current design and planning approaches, we can create more inclusive and resilient environments for a range of life forms.
The essential change is to treat other species as innovators and design experts. This approach expands on existing efforts to communicate with whales, bats, and bees, and uses AI to incorporate input from non-human lifeforms to produce new and better designs.
Our case studies show that participatory approaches that include non-human entities can avoid human biases, thereby opening up a much wider range of design possibilities.
Fair Design
The world faces many urgent environmental crises. Meeting this challenge requires innovative and inclusive design approaches. Trees are already good designers, and birds are good judges of their own work. With their input, we can produce better, “beyond human” design.
We believe that using AI to give non-human stakeholders a voice can lead to better solutions that allow many species to coexist. Our work in Canberra is an example of how participatory design can create a more equitable and sustainable future for all living things.
We recognise the work of Darren Le Roux, who has researched and installed artificial habitat structures to support biodiversity.
