Katarina Zimmer

When you’re having a shot of tequila at your next fiesta, make sure to raise a glass to bats. Yes, bats, because they’re the animal that pollinates the blue agave plant needed to make tequila.

The small winged animals, which represent a whopping fifth of all mammals, do quite a lot of things for us. They pollinate many other plants – like wild bananas – and do a lot of insect control. Some species, like the little brown bat, can eat up to 1000 mosquitoes in a single hour. In fact, bats are one of our best defenses against the spread of mosquito-borne diseases such as Zika.

Knowing where bats live and how their populations are faring in response to our impact on the planet is clearly an important task. But they’re small, largely nocturnal and like to hide – so how do we know where the bats are?

One answer is through sound. Around 80% of bats emit series of acoustic pulses, which they use – along with their resounding echoes – to navigate the nocturnal world. This is called echolocation. Although these pulses are beyond the frequencies that humans can hear, they’re audible to a range of devices called ultrasonic detectors.

About a decade ago, an ambitious project was started in Europe to gather as many recordings of bats, known as the Indicator Bats (iBats) program. Today, volunteers drive through the countryside of 22 countries with bat detectors attached to the roofs of their car, collecting the acoustic information that bats leak to the world about their whereabouts. This has generated a vast amount of audio recordings – more than any human alone could ever listen to.

The big challenge is to develop automated ways of telling us just how many bats are in those recordings, and of which species. Though each species has its own signature echolocation ‘call,’ telling a computer how to distinguish between them is complicated. Let’s take a look at some calls. We can’t hear them, so plotting them on a spectrogram (a frequency-time plot) is one way of looking at them. For some species, the call ‘shapes’ look completely different, making them easy to tell apart, like the calls in the top image below. But within some groups of bats – like brown bats, shown in the bottom image – calls between species can look pretty alike. On top of this, there can be a lot of variation in calls between a species, so it’s hard to give a computer concrete instructions.

Screen Shot 2017-03-31 at 20.45.35.png

Screen Shot 2017-03-31 at 20.45.58.png

Spectrograms of calls of various bat species

This is the type of problem a new branch of computer science is cut out to solve: machine learning (ML). ML is a kind of artificial intelligence that allows algorithms to ‘learn’ from data you give them without being explicitly programmed to do so. In discriminating between bat species, they’ve proven to be more accurate than other computational methods and even well-trained experts.

Let’s build a simple classifier. We’ll pick a machine learning algorithm called Random Forest, which is easy to understand. To train the algorithm, we have a bunch of cleanly recorded bat calls from 33 European bats.

Before we do, we’ll need to package the bat calls somewhat differently – Random Forest doesn’t know what to do with raw audio. Using simple programming tools, we can extract some simple parameters from the calls. For instance, we can take the mean frequency for each millisecond time slice across a call – which you can see represented as black dots in the spectrogram below.

Screen Shot 2017-03-31 at 20.47.27.png

We can get crazier and fit a curve to that line, and calculate some fancier parameters from the curve itself, like the slope, or the steepest slope – stuff you might vaguely recall  from high school math. Using our imagination, we came up with 34 parameters in total.

Screen Shot 2017-03-31 at 20.48.57.png

We’re not going to tell Random Forest how it’s going to use these parameters to discriminate between species. The beauty of machine learning is that the algorithms are capable of “learning” for themselves the best way to order data. We simply feed Random Forest the parameters, and it will learn from these and will build us a classifier, i.e. an algorithm capable of assigning a species identification to a call it hasn’t seen before.

We set aside 20% of our data to test the algorithm on, and feed it the parameters from 80% of the bat calls. Random Forest works on the basis of decision trees. For each decision tree, it will take a handful of data – in this case, bat calls – and uses the parameters we extracted to create a pathway of decisions for the algorithm to decide which classification to make. For instance, a simple decision tree would classify every bat call with a start frequency above 100kHz as a horseshoe bat. If it’s below that frequency, it will, say, look at the steepest slope of the call. If it’s steep, it could be a brown bat, and so on. Random Forest computes many such decision trees and averages over all of them to create a final classifier for prediction.

Screen Shot 2017-03-31 at 20.49.45.png

After growing the classifier, we test it on the 20% of data we initially set aside. We see it correctly predicted the species in nearly 70% of cases! Notably, it did perform worse in some bat groups than others: for example in the group of brown bats, which – as we previously discussed – are tricky to distinguish.

70% is not bad for a start. When a previous research group used an artificial neural network – a different type of machine learning algorithm – for the same species of bats, it achieved 83.7% accuracy. But the trouble with artificial neural networks is that they are somewhat of a “black box” in terms of how they work. We can test if they are accurate, in the same way we tested the accuracy of our Random Forest agorithm. But what happens inside the box, i.e. why the algorithm produces a certain prediction, we can’t explain very well.

Random Forest can tell us which parameters exactly it found most useful in discriminating between species (in this case, this was the lowest frequency of the call). We know exactly what it’s doing. The downside is that it wasn’t quite as effective as the artificial neural network – at least for this particular data. When we tried our classifier on some real-world data collected on the island of Jersey a few years ago, it didn’t todo so well. For one, it kept categorizing calls to species not known to occur in that area. The program we wrote to extract features on the calls seemed to falter, too: in most cases, the curve was unable to align to the bat calls. The reason was simply that these new, real-world recordings were particularly noisy, unlike the clean data we had trained our algorithm on.

When it comes to machine learning algorithms, there’s often a trade-off between precision and transparency. Of course we want to know exactly how an algorithm works and why it makes the decisions it does, but sometimes we have to forsake transparency for precision.

The Jones group at the Center for Biodiversity and Environmental Research at University College London is making use of the latest trend in machine learning to classify bat calls: deep learning. These algorithms are particularly opaque and difficult to understand. They’re highly complex, but they’ve excelled in accuracy for certain tasks. The Jones group is working towards using deep learning techniques to create a new classifier for European bats.

Although they are less transparent, they are extremely promising. As long as they are vigorously tested to make sure that they’re drawing the right conclusion for each bat, such algorithms hold big promise for ecology.

The more we know about bats, the better. It seems that artificial intelligence can get us a good step of the way there.

And it might just help keep the tequila flowing.

Learning about learning – what mouse babies can tell us about the brain

Today I had the opportunity to speak to Jennifer Shiavo, a third-year PhD student and one of the many, many scientists collectively working to piece together one of the biggest puzzles of neuroscience: How the brain learns. It’s easy to take for granted what our brain does for us – constantly taking in everything we see, feel or hear, somehow making sense of it all and ultimately deciding on which bits we remember, which ones we forget, and which ones we learn for life. But what exactly the 100 billion nerve cells in our brain are doing to allow this, is something that we don’t quite yet understand.

Shiavo works in a lab led by Dr. Robert Froemke, which has been looking at mice to understand this. More specifically, she’s looking at how mother mice pick up a particular habit common to all mothers in the animal kingdom. Dutiful mouse mothers tend to change nests from time to time, carrying their pups along the way. Occasionally, one will become lost, and it’s up to the mother to respond to her pup’s distress cries and bring it safely to the nest. This is not an ingrained instinct that the mothers do naturally – it’s something they have to learn. What’s more, some females can pick up this behavior too, after being around the pups for a few days – without having gone through the hormonal and other bodily changes needed to become a real mother.

These virgin mice have become the focus of the lab’s work on learning. The team started by comparing the brains with the virgin mice that had learned to respond to the pup calls and rescue them, and those that hadn’t – and found that they actually have very different responses to hearing the pup calls. When playing recordings of the pup calls to the mice, there is always some electrical activity in the area of the brain that processes sounds, the auditory cortex. In the auditory cortices of those mice that knew to respond to the pup calls, activity spiked in a very precise way upon hearing the pup calls – looking a lot like the cortical responses of the mother mice themselves. In comparison, in the naïve mice that had not developed this rescue behavior, the spikes of activity appeared erratic, and had little correlation with the hearing of the pup calls.

But what actually transforms a naïve brain into a brain that has effectively ‘learned’ to react properly to pup calls? Knowing that a particular hormone, oxytocin, is involved in maternal and social behavior in general, the team investigated its effects on this process – with some remarkable findings. They found that they could prompt the brains of naïve mice to learn to react to pup calls a lot faster if they injected them with oxytocin whilst playing them the recordings. Shortly after this, the naïve mothers would run to the rescue of pups in distress. Blocking the activity of the left auditory cortex, which is particularly dense in receptors for the hormone, causes them to ignore crying pups.

Along with other experiments, this supports the idea that oxytocin is pretty important for mice to learn how to respond to pup calls. The complete picture, though, is likely to be more complicated, Shiavo explains: It’s probably not just oxytocin alone that facilitates this process, and there will be other brain regions involved, too.

But it’s a start, and for now, Shiavo’s goal is to track the changes in the brains of naïve virgin mice as they become experienced virgins, step, for step: Nailing down exactly when and how oxytocin acts will lay the groundwork for understanding how this learning occurs. In the grand scheme of things, this will shine a light on how we humans, too, learn from social experience and grow as social creatures.

“A continuous series of corals, sponges, sea anemones, and other marine productions, of magnificent dimensions, varied forms, and brilliant colors. In and out moved numbers of blue and red and yellow fishes, spotted and banded and striped in the most striking manner, while great orange or rosy transparent jellyfish floated near the surface …The reality exceeded the most glowing accounts I had ever read of the wonders of a coral sea.”

Such were the words of British explorer Alfred Russel Wallace on his voyage through the Malay archipelago in 1869. The region is home to the most diverse ecosystems of the marine domain—coral reefs, created by thousands of years’ worth of work by minute organisms that extract calcium carbonate from the water and deposit this as solid limestone. By doing so, corals—and certain types of algae—have built colossal structures that give home to a diversity of organisms rivaling that of the Amazon rainforest.

This diversity sustains the lives of some 500 million people across more than 100 nations with coastlines fringed by coral reefs. These ecosystems are an indispensable source of fish and seafood and also critical breeding, feeding and nursery grounds for many open-ocean fish, notably tuna. They’re invaluable to medical research, harboring organisms with anti-cancer or AIDS-inhibiting properties, and to tourism, being the producers of the fine, white coral sand of tropical beaches. Together with mangrove forests they form fortresses that protect entire coastlines from waves, storms and floods.

But today, nearly 150 years after Wallace described them in their pristine state, the ‘rainforests of the ocean’ stand at the brink of ecological ruin. Increasingly crowded coastal populations are exhausting the once thriving fish stocks. According to a study published earlier this year in Nature, nearly nine in ten reefs have lost at least half their natural fish numbers. Particularly the most desirable fish—which tend to be large, predatory ones—have shown devastating population crashes, often crossing the threshold at which they are capable of regenerating.

Equally alarming are the knock-on effects of the decimation of these species. The coral reef food web is an ancient, carefully balanced network in which the loss of any one player can cause a domino effect that reverberates through the whole system.

For instance, the heavy exploitation of species that predate on sea urchins has had disastrous effects on Kenyan reefs: These particular sea urchins feed on crustose coralline algae, which—like corals—are critical to reef formation. In the absence of their predators, they were able to proliferate to extreme abundances, wreaking havoc over the reef by eroding away its foundation. This enabled other types of algae that are typically harmful to corals to thrive and dominate the reef. This effect was exacerbated by the overfishing of herbivorous fish species that usually graze on them. By comparison, non-fished reefs in marine parks remained healthy, where sea urchins and algae were kept in check by intact fish communities.

Screen Shot 2017-03-30 at 10.00.08

Kenya: Overfishing of invertebrate predators and herbivorous fish can lead to overgrowth of algae.

All unsustainable fishing has negative impacts on fish stocks and has the potential to throw the ecosystem out of kilter via such ‘trophic cascades’ – where the removal of a predator causes an increase in the numbers of its prey. However, because we don’t know exactly how reef food webs work—and each one may have evolved differently—, they are near impossible to predict.

Tragically, the last three to four decades have seen the rise of a number of destructive fishing practices that further exacerbate the situation. Desperate as ever to ensure their income despite dwindling stocks, many fishers have resorted to a variety of highly destructive fishing methods: Bottom trawling, spearguns, explosives and cyanide have had catastrophic impacts on coral reefs through their alarming potential to not only capture vast amounts of fish with little effort, but also to kill many other species and destroy the reef-building framework.

In Indonesia—the world-leading exporter of ‘ornamental’ fishes—cyanide is used to capture colorful reef fish for the aquarium trade or display in restaurants—supposedly stunning the fish without killing them. Many fish die regardless and surviving ones tend to have shorter lifespans. A non-selective respiratory poison, cyanide can severely damage any reef organisms it comes into contact with, including corals.

Also, bombs made from artificial fertilizers or illegally sourced dynamite are thrown upon schooling reef fish, which are killed and then collected by divers. A single beer bottle bomb can obliterate all life within an area of around 20 square meters and shatter the delicate skeletons of live corals, turning them into dead rubble.

Screen Shot 2017-03-30 at 10.00.19.pngIn Southeast Asia, nearly 95% of reefs are affected by overfishing and destructive fishing practices.

Most reefs today would appear unrecognizable to Wallace, their fish stocks drained, their food webs knocked out of balance, bombed, poisoned. These impacts make it all the more difficult for reefs to cope with the effects of coastal pollution, rising sea levels and global warming.

If current rates of destruction continue, it is estimated that 60% of reefs will be destroyed over the next 30 years, with devastating consequences for the lives depending on them for food, income and coastline protection.

According to the aforementioned study in Nature, completely banning fishing on reefs could restore the ecosystems to a robust state within 35 years. However, burgeoning populations and increasing poverty and hunger in developing reef nations make this strategy unfeasible.

Hope lies in communities that have succeeded for generations in maintaining a balance between fishing and keeping their reefs in good health. By enforcing seasonal fishing bans to give the fish time to spawn and breed, whilst condemning the use of destructive fishing practices, traditional peoples of Kenya and Raja Ampat, Indonesia, set an example that could lead the way into a sustainable future for reef fisheries. Perhaps one day, coral reefs will recover to their natural state – bursting with the color and life that Wallace so admired.


Carbon emissions, fossil fuels, that 2°C threshold we’re not supposed to cross—science is good at getting the facts straight about climate change, but less effective in communicating how it affects us. This is where art may come in.

In imagining art about changing climate, one might think of paintings of melting icecaps and expansive deserts, but there is a surprisingly diverse range of approaches artists use to talk about the phenomenon. “I believe that just looking at paintings in a gallery about the North Pole, for instance, don’t really do anything for consciousness of climate change, or awareness”, says Regina Cornwell, organizer and curator of inClimate, which comprises several artistic projects in New York about climate change, all under the umbrella of Franklin Furnace Archive Inc., an arts organization based in Brooklyn.

Cornwell started inClimate in the hope of addressing a gap in public awareness about climate change. “It’s hard to see when you walk around the streets of New York that there’s a kind of awareness about this,” she says. The main idea is that artists work together with specialists—scientists, engineers, ecologists—to explore new ways of getting people to think about the issues around climate change. And the work appears to be having an impact. The number of grants awarded by the National Endowments for the Arts to artistic projects like inClimate has increased significantly in recent years.

The projects that have come out of the program are extremely diverse. Lynn Cazabon is one of the artists of inClimate. Her project, entitled Uncultivated, is based at Hunts Point in the Bronx. Working together with a botanist, she held a number of workshops to introduce both young and old to the wild plants of their neighborhood, and plant new ones. A few months later she partnered with naturalist “Wildman” Steve Brill to harvest the plants and cook them into a meal. Cazabon’s main goal is to get people to recognize the nature around them, and how climate change is affecting it: “We think of nature a something that is far away, outside the city.  But we’re completely submerged in it, although we don’t think about it that way. I’m trying to change that kind of thinking.”

Part of Uncultivated – which has been done in other cities in the U.S. – also involves taking photographs of what are typically considered “weeds” in urban areas, enlarging these to billboard size and installing these around cities. By doing so, Cazabon hopes to reach a broader audience: “People who don’t go to art galleries are then exposed to something provocative – it’s a gesture of bringing attention to it”.

This kind of project may be on the fringe of what most people consider “art”, but Cazabon disagrees: “The whole thing is art”, she says: “The photographs, the community – I consider that all art.” When trying to get people to think differently about something, she says it’s important to explore new forms of artistic media. Chantal Bilodeau, a playwright known for creating The Arctic Cycle, a series of plays about climate change, agrees: “It’s important to get out of traditional venues.”

Though her work is conceptually quite different from Cazabon’s, both artists share the same sentiment in wanting to bring the topic of climate change closer to home. Bilodeau’s first play featured a variety of actors, each playing a different role of someone affected by warming in the Arctic, for instance a climate activist, a lobbyist advocating the drilling of a deep-sea oil rig, an Inuit and a polar bear. This approach made it possible to explain where different people are coming from, and that the picture is a lot more complicated than at first glance.

The main aim of these projects is not to explain about climate science per se, but to foster a sense of general awareness and in doing so addressing a certain apathy towards these issues in society. “A lot of my projects have to do with changing a paradigm”, says Cazabon: “Any kind of behavioral change can’t happen until a conceptual change in society.”

However, such projects remain a small niche in the art world – maybe because it’s not quite science, and not quite art in the traditional sense in the word. “There has been very little interest in producing this kind of work”, says Bilodeau: “The theaters haven’t caught up yet.”

Going on a bat walk with Paul Keim

This summer in Central Park, I had the opportunity to go on a ‘bat walk’ with naturalist Paul Keim. Have a listen!