Yves Vanderhaeghen speaks
to the founders of WildTrack, which is pioneering technology in South Africa to track small
mammals so they can tell us about disruptions to ecosystems.
Is it a mouse? A shrew? A rat? We can read a lot into the pawprints of small mammals, even
signs of climate change. But they tread so lightly it’s tricky to know from faint tracks in the
sand who’s passed by.
To solve the difficulty a team from WildTrack, a non-profit which uses AI, data analytics and
on-the-ground expertise to protect endangered species, is pioneering technology in South
Africa. It consists of laying down a patch of charcoal for an animal to step into and white
paper to pick up its prints, which are then analysed and recorded in a growing database.
Founders Zoë Jewell and Sky Alibhai were recently at Telperion Nature Reserve on the
border of Mpumalanga and Gauteng provinces to run a second round of trials together with
their South African colleagues Nico Avenant, head of the Department of Mammalogy at the
National Museum in Bloemfontein, and Marietjie Oosthuizen of the Department of Zoology
and Entomology at the University of Pretoria.
Jewell and Alibhai, together with OGRC head Duncan MacFadyen and Avenant, will do a
presentation on the technology at the Oppenheimer Research Conference (ORC), which
takes place in Johannesburg from October 9 to 11. The presentation is titled: “A
transformative tool for monitoring biodiversity using a combination of traditional ecological
knowledge with leading-edge analytics and AI”.
“While we're collecting small mammal footprints across a wide range of countries, South
Africa is the first place we've decided to start implementing it as a biodiversity survey,” says
Jewell.
“We did an initial run last year at Tswalu reserve in the Kalahari, and we repeated the
process at Telperion. The idea is to set up a system which enables the recording of all small
mammal species using their footprints to identify at the species level,” says Jewell.
The basic technology has already been proven in other parts of the world, but for large
animals. “WildTrack has got this project called FIT (footprint identification technology), which
we've been using over a long period of time,” says Jewell. “We've been working with tigers in
China; rhinos, leopard, lion in Africa. In Brazil, we work with puma and jaguar.” FIT can
identify species, individuals, and their age class which guides monitoring strategies.
“Now we're applying that to small mammals, and because it's a non-invasive technique we
don't need to catch or handle them,” says Jewell.
Biologists increasingly rely on non-invasive sampling techniques such as camera traps and
acoustic recorders (which are also being pioneered for small mammals by Dr Haemish
Melville at Telperion), as well as non-invasive genetic sampling from scat, for example.
“These non-invasive techniques don't disturb or change the natural behaviour of the animals
in any way, which can often provide better quality scientific data as well as enhance welfare,”
says Jewell.
“So we're not actually trapping animals, except when we set up the initial database. Once
that is done, we then just use the track plates to collect the footprint images from which we
can identify the species. Our aim is to develop a non-invasive and cost-effective metric for
monitoring biodiversity. And because small mammals are a very important part of the
integrity of most ecosystems, they make a great metric for assessing the health of
ecosystems.
“What we would love to do,” she says, “is have something like our track plates go out into the
field with little acoustic devices in them that will pick up sounds as well. So we'll have a
wealth of information coming through about these little animals without ever having to see
them or track them or disturb their natural environment.”
Computer technology is key to the process.
“To begin with,” says Jewell, “you have to develop a reference library to train the computer to
know which species is which. At the moment, with the small mammals, we're in the training
phase where we capture a range of species, and we have them walk over a little platform,
which we have in the lab, which is a little bit of charcoal on one end and some standard shelf
sticky paper. They walk over that, they get a little bit of charcoal on their feet, they walk over
the sticky paper, they leave their footprints, and then we pick them up at the other end, put
them back in the trap, and release them at the same locality they were collected.
“We then have a piece of sticky paper with footprints on it, which we can attribute to the
animal that we caught, so we know exactly which species it came from. And we repeat that
process with all the species that we catch, and we've got 20 species at the moment. So
that's a hot off the press figure,” says Jewell.
“When we've caught enough new species, and we're confident that we've got a good library,
we put those through the computer using two different systems. One is a morphometric
system, which allows us to take measurements from those footprints. We can identify the
species that made those footprints using a customised statistical program called JMP data
visualisation software. So that gives us a measure of which species is which.
“The other system that we're developing is artificial intelligence, where we'll simply take an
image of the whole track plate with all the footprints on it, and we'll put that through the
computer, and train it how to identify the species just from looking at those images of the
footprints without having to take any measurements or any other manual intervention. Those
two approaches together will then allow us to put out these very simple little tunnels into the
field with some sticky paper and charcoal, pick those up every day, and get a measure of
which species are out there without any trapping or handling, or any real need for a lot of
expertise in the field.”
It's easier with large animals which leave clear indentations in sand, “but the challenge with
small mammals has been they're very light, and they don't leave very clear footprints in
natural substrate,” says Jewell.
However, they do leave “absolutely perfect footprints on these little pieces of sticky paper,
and you can put them down anywhere in the field. You can nestle them in the grass, near
the small mammal runs, or in the sand.”
She shows a short video to illustrate the clarity of impressions.
Alibhai explains the footprints we’re seeing: “This is a Micaelamys namaquensis, a Namaqua
rock mouse,” he says. “And you can see this lovely looking hindfoot here. I can tell you that
that's a left hind foot just by looking at the arrangement of the digits: 1, 2, 3, 4, 5 – they're the
tips of the fingers, and then these are the interdigital pads. And from the arrangement, you
can actually tell that's a hind foot.
“And this one over here, which is a little bit more faint, is a front foot. And once you get your
eye in, you begin to understand exactly the arrangement of the left and right. And then what
we basically do with morphometrics, which is taking measurements from the footprint, is to
assign each of those points that I've just mentioned a number, and we put that into our
software, and we extract distances, angles, areas, and then we use statistics, which allows
us to identify which species it is.”
While morphometrics have been used widely with other species, Alibhai says that “the
actual implementation of this as a biometric survey technique is being pioneered in South
Africa. It's our intention that it'll help large companies and organisations who are required to
do environmental impact surveys. It'll help them do it much more cheaply and quickly than
they're currently able to. At the moment, if you want to go and survey small mammals, which
everybody agrees is an important group to survey, you need to have experts in the field who
can catch them, handle them, identify them, count them, and repeat this over and over and
over again.”
Small mammals are excellent indicators of biodiversity disruption, says Alibhai, “because
their replacement rate is very, very fast”. This means being able to measure their response
to environmental change, “be it natural or man-made. For example, if a company is trying to
set up a huge factory somewhere, and they want to do an environmental impact
assessment, we can use small mammals as indicators. Go in there and do a before study, a
during study, and then follow it up with an after study. It's a very simple technique, very
cheap, very easy to deploy, doesn't require a great deal of expertise, doesn't require a great
deal of equipment.
Adds Jewell: 'It's not only the number of species that's important. We also need to know
which species. Many are indicators for different types of environmental disturbance, for
example showing up if there's a negative impact or a change in predator balance,” which
might show why a grassland is being destroyed by mice when their numbers explode due to
a lack of predation.
Once the AI has produced a list of species, it will indicate the health of the environment.
“When you start looking, you realise what incredible variety there is. And small mammals
actually account for up to 40% of all mammal species in some areas,” says Jewell.
“But people don't see them,” interjects Alibhai. “They're almost invisible. But if you set up a
long term study, there would be great indicators of climate change. If you look at their
species distribution, and look at their numbers, that would be an indicator of climate change.
Jewell stresses that local collaborators are critical in this research. “We're not from South
Africa, but a lot of our team members are. One of our objectives is to make sure that when
this technique is ready for implementation, people on the ground are able to use it.”
This WildTrack programme has another local connection, Oppenheimer Generations
Research and Conservation, whose head, Duncan MacFadyen offered that they work at one
of the OGRC sites. “So that's how we ended up at Tswalu last year and Telperion just
recently. And we'll be back at Telperion in October for the 3rd phase of the project.”
The JRS Biodiversity Foundation (https://jrsbiodiversity.org/) funds this project.
“We're very lucky that we've got good colleagues in South Africa who are helping us develop
it because they'll be key to making it work in South Africa. And then, hopefully, South Africa
will pioneer it and then spearhead development for other regions in Africa.”
Yves Vanderhaeghen writes for Jive Media Africa, communications and research partner of
Oppenheimer Generations Research and Conservation.
