Wild counting in the future!

Imagining what ecological monitoring will be like in 50 years.

written Aug 4, 2018 • by Jon Sullivan • Category: Wild Soapbox

tangle banner

Grab some fresh banana skins for your flux capacitor, we’re going to the future! Not the mundane one-day-at-a-time kind of time travel either. Let’s project ourselves forward 50 years and imagine what it will be like to explore nature and count the wild in the future. What is the future likely to bring and how should it affect how we count the wild today?

It’s a safe bet that future technology will revolutionise nature watching and biodiversity monitoring (assuming the world hasn’t sunk into some Mad Max apocalypse). Fifty years ago was decades before the internet, personal computers, digital cameras, personal GPS units, and cellphones. Personal electronic calculators had just become available and the first digital watches were just a few years away. Compared with today, counting the wild took enormous dedication and a steady supply of notebooks and pencils.

Given the rapid progress in the past decades, it seems a certainty that the next decades will be at least as transformative. It’s easy to imagine many of today’s cutting edge technological advances becoming common place alongside surprising new devices and information technologies.

It’s already the case now that technology can both enhance and distract from our experiences with nature. It all depends on how you use it. Using a smartphone in the wild with the iNaturalist app can deepen your experience in the wild. Getting distract by Instagram notifications does the complete opposite. I can see that dichotomy and the lure of distraction growing over time. Still, I expect that the observation-enhancing power of new technologies is going to be phenomenal.

At risk of the future proving me wrong, I predict that the following advances will all become well established and widely used in several decades, if not sooner. All have the potential to revolutionise our monitoring of wild nature. Many also make us re-think what observations we should be collecting now.

Accurate automated species ID from audio recordings

It’s already possible to train neural network software like Kaleidoscope Pro to automatically recognise different animal species, populations, and even individuals, from audio recordings. Yes, it’s already possible, with a substantial amount of work, to train these systems to recognise individual birds. At the moment it takes a lot of expertise and work to pull together the right mix of carefully curated recordings to train the computers.

I can imagine a near future when all of the ground work has been completed and computers are be able to easily and accurately process massive amounts of audio recordings. They’ll be able to quickly extract and identify and count every individual animal from recordings. Audio recorders are also going to inevitably become much cheaper and much higher quality than they are now.

In that world I can imagine networked audio recorders on the top of every city bus or building automatically monitoring all of a city’s audible nature. Every bird that sings anywhere in a city would get recorded. Similarly, in important wild sites, I can imagine networks of audio recorders monitoring all audible animals at a site. Both of these scenarios are already entirely possible today, but expensive to implement. That will get much cheaper.

The future inevitability of widely available, accurate, automated audio processing of wild audio recordings has another important implication. We should be making lots of audio recordings right now, in places that matter to us, for future processing. All our original recordings of the wild should be properly curated and archived for future use. We are the time machine for future biodiversity scientists. We’re not recording nearly enough about what wild nature is like now.

Accurate automated species ID from images

It’s also inevitable, given the rapid recent progress, that computers will be able to identify species in photos and videos at least as accurately as the world’s taxonomic experts. Neural networks like iNaturalist’s computer vision and Google’s Lens are quickly digesting the expertise of the world’s biodiversity experts. Cameras are also getting better and cheaper all the time.

This means that alongside those networked audio recorders, on the top of each city bus or building, there could also be a network of cameras filming everything. The images and audio sent to computer systems would give a detailed real time account of the abundance of all large species seen and heard. Many of the world’s cities are already filled with closed-circuit surveillance cameras to reduce crime. Imagine what future networks could do for wildlife surveillance.

Cameras facing the sky on the top of buildings throughout a city could identify and monitor all flying birds. Cameras in pairs would give a 3D image that would both ID each passing bird and measure its distance and flight direction, allowing bird densities to be estimated. Such a system is probably possible now, but it would be expensive and take a lot of training.

Like audio, this inevitable future will place a high value on photos and videos of the wild that we make now, especially those made within well-structured monitoring projects. A strong case can be made for citizen scientists and our biodiversity institutions gathering much more audio and image data than we are currently archiving. Storing masses of audio and images for analysis by future artificial intelligence is much cheaper than storing masses of physical specimens in museums for future DNA analysis.

Cheap, accurate DNA analysis

Huge progress is being made in DNA sequencing technology making it increasing affordable to sequence DNA from environmental samples. Steady (but slow) progress is also being made on DNA barcode libraries that link sequences to species identifications. Without these barcode libraries, sequences are just sequences. With them, you can connect sequences to everything that is already known about a species, and you can add to that knowledge.

DNA can be sequenced from just about anything, including soil, water, and air. It’s possible to imagine future sensors in rivers, on farms, and air intakes on buildings, or buses, that identify everything, from microbes to tiny remains of larger animals. A dense network of such sensors, sampling continuously, would provide an extraordinarily rich record of nature and its changes.

It’s still a big question whether, or when, these eDNA sensors will make counting with brains, images and audio obsolete. As the affordability and power of DNA sequencing has increased, so too has the affordability and power of image and audio recording.

While nothing can beat the taxonomic breadth of species detectable with DNA, I still find it hard to imagine that DNA-based monitoring will provide the affordable temporal and spatial resolution provided by image and audio methods. My bet is that these methods will continue to complement each other in biodiversity monitoring for many decades to come.

With cheap, powerful sequencing on the horizon, it becomes important to store samples now. That’s a common theme between audio, image, and DNA technologies. They should all be collected in monitoring projects today for future extraction of all their data.

Rapid personal data recording

It’s easy to imagine the vast improvements soon in store for Siri, Alexa, and their digital assistant colleagues. It’s much quicker to talk than type, and it’s clear that computers will soon be able to accurately understand human voices even in noisy outdoor environments.

I’m personally betting on this technological improvement to understand and accurately transcribe the few thousand geotagged and date-time-stamped audio observations I make each week. Current systems struggle to understand my Kiwi accent and my talking, especially when there’s outside background noise. There’s rapid progress though.

Looking further forward, in 50 years there could well be ubiquitous brain-computer interfaces where even talking will seem like a clunky way to enter observations into a computer. People are already natural observers of nature and there are a lot of us. Everything that makes it easier to count wild things of interest will generate many more counts.

Add on augmented reality glasses with high resolution cameras and audio recorders, plus a futuristic computer-brain interface, and future naturalists will be extraordinarily powerful recorders.

Living in a world of virtual and augmented realities

Perhaps the most unpredictable consequence of rapidly advancing technology is what increasingly powerful virtual and augmented realities will do to people’s engagement with nature. Many fear that it will draw society further away from meaningful interactions with the wild natural world. Instead, I see hope in the richer and more immediate interactions with nature offered by augmented reality. They may lead to more people spending more time with wild nature. They could make every garden as effortlessly rich in natural drama as an Attenborough documentary.

Technology in the past few decades has made watching and counting the wild much easier and more rewarding. Decades ago it took years of hard won expertise to recognise the species around you. Now, anyone with a smartphone can share photos and audio recordings of any species and get them identified quickly using iNaturalist.

Imagine what a fully immersive augmented reality technology could bring in another few decades (or less). Our exploration of the wild could be augmented by all manner of new data feeds and sensors, including greatly enhanced hearing, vision, touch, and smell.

How much more extraordinary would wild nature seem if you could see the world in ultraviolet like bees do, or see it in infrared like snakes. You could hear ultrasound like bats can, and understand the meaning of bird songs. You could detect and track the pheromones of moths or ants. You could taste the chemistry inside leaves. And, you could instantly identify any species you saw or heard and quickly learn everything known about it.

With artificially augmented senses and knowledge, the wild world would seem more alive and rich in meaning than ever before. That could be revolutionary for people’s engagement with nature, for advancing natural history, and for wild counting. I am hopeful.

Drones and nano probes

Automation plus augmented reality could mean that we’re no longer restricted to exploring the natural world at human spatial scales and the human pace of time. I can imagine augmented reality tech connecting you to miniaturised drones flying, crawling, digging, and swimming through all sorts of previously hard-to-reach habitats. It would feel like we were really in these places, running with ants or flying with bees. The new possibilities for exploration and discovery, and fun, would be enormous.

Perhaps many people will probably be quite happy staying inside virtual reality. After all, throughout history most people have been more engaged in human stories than the reality of nature. Still, for those people seeking the wild, future technology, when used right, could hugely enrich and expand those wild experiences.

Robots, AI, and living wages

Perhaps the biggest changes to wild counting may not be technological at all but result from the social changes brought about by technology. For example, it seems clear that many human jobs will get replaced with robots and artificial intelligence. What does a society do when there aren’t nearly enough jobs for everyone to work full time? One proposed solution is redistributing wealth to give everyone a living wage.

If the impending rethink of society includes a base living wage, there will be the potential for an explosion of creative and community-focused activities done by people with time to spare. Think of all the volunteer work done by retired people and expand that across most of today’s working younger people. People in the future could have much more time to engage with nature.

Combine all that extra time with technology-enhanced powers of observation and we could see a renaissance of natural history and biological discovery.

Back to the present

Coming back to now, things seem pretty humdrum with our dumb “smartphones”. Still, I still don’t think we’ve come anywhere close to realising the power of existing technology for counting the wild. Part of realising this potential is keeping an eye to the future and collecting the observations the future will need to make sense of its past. I can see a strong rationale for collecting and carefully archiving massively more images, audio, and eDNA samples than we do now.

When you’re next out in the wild, don’t restrict your thinking to what can be done now. Think also about what future tech could do with your observations. Perhaps you could have a smartphone set to record audio the whole time you’re at each site. You could take panoramic photos at every site you visit. If you have the right permits, you could collect a small sample of soil from each site for future DNA sequencing (if you have a spare -80°C freezer to store them in).

If we were organised about how we did this, we could offer the future a detailed view of what the wild is like right now, today. Everything points towards the wild of the future being different, and likely less diverse, than what we have today. What we observe now will tell future generations how different their wild is, and help them to predict and manage their changing wild.

Be part of tomrrow...today! Buy a brick.
Be part of tomorrow ... today! Instead of buying a brick (?), think about recording some audio of nature, or taking some geotagged photopoint photos, or perhaps, if you've got the facilities, start storing samples for future eDNA sequencing.