AI reveals hidden traits about our planet's flora to help save species

In a world-first, scientists from UNSW and Botanic Gardens of Sydney, have trained AI to unlock data from millions of plant specimens kept in herbaria around the world, to study and combat the impacts of climate change on flora.

"Herbarium collections are amazing time capsules of plant specimens," says lead author on the study, Associate Professor Will Cornwell. "Each year over 8000 specimens are added to the National Herbarium of New South Wales alone, so it's not possible to go through things manually anymore."

Using a new machine learning algorithm to process over 3000 leaf samples, the team discovered that contrary to frequently observed interspecies patterns, leaf size doesn't increase in warmer climates within a single species.

Published in the American Journal of Botany, this research not only reveals that factors other than climate have a strong effect on leaf size within a plant species, but demonstrates how AI can be used to transform static specimen collections and to quickly and effectively document climate change effects.

Herbarium collections move to the digital world

Herbaria are scientific libraries of plant specimens that have existed since at least the 16th century.

"Historically, a valuable scientific effort was to go out, collect plants, and then keep them in a herbarium. Every record has a time and a place and a collector and a putative species ID," says A/Prof. Cornwell, a researcher at the School of BEES and a member of UNSW Data Science Hub.

A couple of years ago, to help facilitate scientific collaboration, there was a movement to transfer these collections online.

"The herbarium collections were locked in small boxes in particular places, but the world is very digital now. So to get the information about all of the incredible specimens to the scientists who are now scattered across the world, there was an effort to scan the specimens to produce high resolution digital copies of them."

The largest herbarium imaging project was undertaken at the Botanic Gardens of Sydney when over 1 million plant specimens at the National Herbarium of New South Wales were transformed into high-resolution digital images.

"The digitisation project took over two years and shortly after completion, one of the researchers -- Dr Jason Bragg -- contacted me from the Botanic Gardens of Sydney. He wanted to see how we could incorporate machine learning with some of these high-resolution digital images of the Herbarium specimens."

"I was excited to work with A/Prof. Cornwell in developing models to detect leaves in the plant images, and to then use those big datasets to study relationships between leaf size and climate," says Dr Bragg.

"Computer vision" measures leaf sizes

Together with Dr Bragg at the Botanic Gardens of Sydney and UNSW Honours student Brendan Wilde, A/Prof. Cornwell created an algorithm that could be automated to detect and measure the size of leaves of scanned herbarium samples for two plant genera -- Syzygium (generally known as lillipillies, brush cherries or satinas) and Ficus (a genus of about 850 species of woody trees, shrubs and vines).

"This is a type of AI is called a convolutional neural network, also known as Computer Vision," says A/Prof. Cornwell. The process essentially teaches the AI to see and identify the components of a plant in the same way a human would.

"We had to build a training data set to teach the computer, this is a leaf, this is a stem, this is a flower," says A/Prof. Cornwell. "So we basically taught the computer to locate the leaves and then measure the size of them.

"Measuring the size of leaves is not novel, because lots of people have done this. But the speed with which these specimens can be processed and their individual characteristics can be logged is a new development."

A break in frequently observed patterns

A general rule of thumb in the botanical world is that in wetter climates, like tropical rainforests, the leaves of plants are bigger compared to drier climates, such as deserts.

"And that's a very consistent pattern that we see in leaves between species all across the globe," says A/Prof. Cornwell. "The first test we did was to see if we could reconstruct that relationship from the machine learned data, which we could. But the second question was, because we now have so much more data than we had before, do we see the same thing within species?"

The machine learning algorithm was developed, validated, and applied to analyse the relationship between leaf size and climate within and among species for Syzygium and Ficus plants.

The results from this test were surprising -- the team discovered that while this pattern can be seen between different plant species, the same correlation isn't seen within a single species across the globe, likely because a different process, known as gene flow, is operating within species. That process weakens plant adaptation on a local scale and could be preventing the leaf size-climate relationship from developing within species.

Using AI to predict future climate change responses

The machine learning approach used here to detect and measure leaves, though not pixel perfect, provided levels of accuracy suitable for examining links between leaf traits and climate.

"But because the world is changing quite fast, and there is so much data, these kinds of machine learning methods can be used to effectively document climate change effects," says A/Prof. Cornwell.

Read more at Science Daily

Thousands of native plants are unphotographed, and citizen scientists can help fill the gaps

Scientists have documented plant species for centuries to help us understand and protect the incredible diversity of flora in our world. But according to new research, many have never actually been photographed in their natural habitats -- and that's a problem.

Researchers from UNSW Sydney and the Australian Institute of Botanical Science, part of the Royal Botanic Gardens and Domain Trust, surveyed 33 major online databases of plant photographs to examine the photographic record of Australian plant species. The findings, published in New Phytologist, reveal out of 21,077 native Australian vascular plant species, almost 20 per cent lack a verifiable photograph.

Lead author of the study and UNSW Science PhD student Thomas Mesaglio says Australia is one of the richest areas in the world for native species.

"It was surprising to see how many plant species had just line drawings, illustrations, paintings, or even no media at all," Mr Mesaglio says.

Dr Hervé Sauquet, co-author of the study and Senior Research Scientist at the Australian Institute of Botanical Science, is based at the National Herbarium of New South Wales.

"All species of plants ultimately rely on specimens in herbarium collections for their identification," Dr Sauquet says. "Yet, even in this digital age where most herbarium specimens have been scanned and are accessible on the web, photos of live plants in the wild remain in critical need."

Senior author of the study from UNSW Science Associate Professor Will Cornwell says a lack of detailed photos can have real consequences. Many plant species that are difficult to identify in the wild may go extinct if scientists cannot properly identify them with the help of photos.

"We had assumed every plant species would have simply been photographed by someone, somewhere, throughout history. But it turns out this isn't the case," says A/Prof. Cornwell.

"This is where citizen scientists can come in and help us fill this gap with their photos."

Gaps in the photographic record

Photographs can help botanists and taxonomists who work with plant specimens by preserving characteristics like flower colour that get lost over time in their samples. They can also show additional features, such as the orientation of leaves or bark appearance, and add ecological context.

"Having a comprehensive photographic set helps us to be confident in our identifications," Mr Mesaglio says. "Particularly when it is practically challenging to collect and preserve the entire plant, photos complement the physical voucher by showing the soil type, the habitat it's growing in, and other species growing alongside it."

But it turns out not all plant groups are photographed equally. Just as some animals receive less attention than others, there might also be a bias against less charismatic plants.

The study found the most well-photographed plant groups tend to be shrubs or trees with more noticeable or spectacular features, such as colourful flowers. Banksia, for example, is one of only two Australian plant genera with more than 40 species to have a complete photographic record. Meanwhile, the family with the most significant photo deficit was Poaceae -- commonly known as grasses -- with 343 unphotographed species.

"We noticed a charisma deficit, so the species that tend to be harder to see are the ones missing out," Mr Mesaglio says. "They may have innocuous or pale-looking flowers or be smaller and harder to spot grasses, sedges and herbs."

Geography also affected the photographic record. While most species across the south-eastern states of Australia have comprehensive records, Western Australia had the largest void, with 52 per cent of all unphotographed species found there.

"The primary 'hotspots' for unphotographed Australian plants are areas with high plant diversity, but the environments are rugged and often difficult to access, particularly by road," Mr Mesaglio says. "But it means there's an exciting opportunity to visit these locations because we might capture something that has never before been photographed."

Activating citizen scientist snaps

It's one thing to have comprehensive photographic records for professional scientists to use in identification guides. But when the plant world is under threat from multiple fronts, including habitat clearing and climate change, photos can help engage the public in plant science.

"People can engage with, sympathise with, and get much more excited about plants with photographs, which is vital when our natural environments are more at risk than ever," Mr Mesaglio says.

"Because digital photography is so accessible now, anyone can also help make a meaningful contribution to science using the camera in their pocket."

Using a platform like iNaturalist, keen citizen scientists can have their snaps identified by experts and share the data with aggregators like the Atlas of Living Australia and the Global Biodiversity Information Facility to be used in research and conservation.

"Since April last year, we've identified nearly 10 per cent of those previously unphotographed species thanks to members of the public uploading their photographs and experts who've kindly identified them," Mr Mesaglio says. "There could be many more in personal collections or behind paywalls just waiting to be shared."

The researchers recommend a standardised system for scientific plant photography be developed, starting with a requirement in the International Code of Nomenclature for Plants to include at least one field photograph where possible in new species descriptions. They also suggest all new species descriptions be published as Open Access in searchable databases with Creative Commons licensing to maximise their usage.

"We also suspect more photos exist, but they're hidden away on social media or behind scientific paywalls that aren't accessible, discoverable, or searchable," Mr Mesaglio says.

Read more at Science Daily

Case study of rare, endangered tortoise highlights conservation priorities for present, future World Wildlife Days

Though wildlife trafficking has been effectively disrupted since the first World Wildlife Day -- established 50 years ago today via the 1973 Convention on International Trade in Endangered Species (CITES) of Wild Fauna and Flora -- a newly published case study on one of the world's rarest tortoise species, the ploughshare tortoise, highlights how much room for improvement still exists.

In a new paper published in the Proceedings of the National Academies of the Sciences, University of Maryland Associate Professor Meredith Gore and her coauthors -- Babson College's Emily Griffin, Bistra Dilkina and Aaron Ferber from the University of Southern California, Michigan State University's Stanley E. Griffis, the University of Alabama's Burcu B. Keskin, and John Macdonald from Colorado State University -- detail a 2018 effort to map ploughshare tortoises' location within and around Soalala, Madagascar; nearby villages; known trafficking pathways and transit routes; and the amount of trafficking risk associated with each of those areas. The group of approximately 50 stakeholders also shared more qualitative information that might play a role in poachers' trafficking process, such as paths of cultural and spiritual significance, tides' influence on decision-making; and where poachers met to plan their activities.

This information was drawn onto a clear, plastic sheet that was laid across a color-based map of the region. That information was then digitized into a geographic information system, creating what the researchers' called a "mess" that nevertheless revealed novel information for the effective targeting of those ploughshare tortoise trafficking networks.

"Our science team used a cross-disciplinary and cross-sectoral approach to think about, measure, and analyze data," explains Gore. "Not only were we able to shift the data landscape to clarify how important water routes are to the resilience of the illicit supply chain, we were able to normalize technical spatial data with insights from traditionally marginalized voices -- women."

Gore and her co-authors argue that if a process like this were to be paired with the latest advancements in computational science, operations engineering, and supply chain management, together, researchers could dramatically disrupt wildlife trafficking networks, and thereby conserve more animals, like the ploughshare tortoise, who are already on the brink of extinction.

"As we celebrate World Wildlife Day, our recent work highlights the urgent need for interdisciplinary collaboration to address the complex global issue of wildlife trafficking," says Bistra Dilkina, an associate professor of computer science and industrial and systems engineering at the University of Southern California. "I feel privileged to have the opportunity to work with a trans-disciplinary team to synthesize a roadmap of how our different disciplines can work together to fight illegal wildlife trafficking and trade. In particular, I am excited to think deeply about the advantages that data-driven approaches in machine learning and optimization can bring to this important endeavor."

Looking to the future, the researchers believe that with increased interdisciplinary collaborations, conservationists may one day be able to predict which path a trafficker will take, target areas where locals could undergo trainings and be empowered to play a part in preventing wildlife trafficking, better allocate limited resources to have the greatest interventional impact, and more.

Read more at Science Daily

Siberian tundra could virtually disappear by mid-millennium

Due to global warming, temperatures in the Arctic are climbing rapidly. As a result, the treeline for Siberian larch forests is steadily advancing to the north, gradually supplanting the broad expanses of tundra which are home to a unique mix of flora and fauna. Experts from the Alfred Wegener Institute have now prepared a computer simulation of how these woods could spread in the future, at the tundra's expense. Their conclusion: only consistent climate protection measures will allow roughly 30 percent of the Siberian tundra to survive to mid-millennium. In all other, less favourable scenarios, the unique habitat is projected to disappear entirely. The study was just released in the journal eLife.

The climate crisis can especially be felt in the Arctic: in the High North, the average air temperature has risen by more than two degrees Celsius over the past 50 years -- far more than anywhere else. And this trend will only continue. If ambitious greenhouse-gas reduction measures (Emissions Scenario RCP 2.6) are taken, the further warming of the Arctic through the end of the century could be limited to just below two degrees. According to model-based forecasts, if the emissions remain high (Scenario RCP 8.5), we could see a dramatic rise in the average summer temperatures in the Arctic -- by up to 14 degrees Celsius over today's norm by 2100.

"For the Arctic Ocean and the sea ice, the current and future warming will have serious consequences," says Prof Ulrike Herzschuh, Head of the Polar Terrestrial Environmental Systems Division at the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI). "But the environment on land will also change drastically. The broad expanses of tundra in Siberia and North America will be massively reduced, as the treeline, which is already slowly changing, rapidly advances northward in the near future. In the worst-case scenario, there will be virtually no tundra left by the middle of the millennium. In the course of our study, we simulated this process for the tundra in northeast Russia. The central question that concerned us was: which emissions path does humanity have to follow in order to preserve the tundra as a refuge for flora and fauna, as well its role for the cultures of indigenous peoples and their traditional ties to the environment?"

The tundra is home to a unique community of plants, roughly five percent of which are endemic, i.e., can only be found in the Arctic. Typical species include the mountain avens, Arctic poppy and prostrate shrubs like willows and birches, all of which have adapted to the harsh local conditions: brief summers and long, arduous winters. It also offers a home for rare species like reindeer, lemmings and insects like the Arctic bumblebee.

For their simulation, Ulrike Herzschuh and AWI modeller Dr Stefan Kruse employed the AWI vegetation model LAVESI. "What sets LAVESI apart is that it allows us to display the entire treeline at the level of individual trees," Kruse explains. "The model portrays the entire lifecycle of Siberian larches in the transition zone to the tundra -- from seed production and distribution, to germination, to fully grown trees. In this way, we can very realistically depict the advancing treeline in a warming climate."

The findings speak for themselves: the larch forests could spread northward at a rate of up to 30 kilometres per decade. The tundra expanses, which can't shift to colder regions due to the adjacent Arctic Ocean, would increasingly dwindle. Since trees aren't mobile and each one's seeds can only reach a limited distribution radius, initially the vegetation would significantly lag behind the warming, but then catch up to it again. In the majority of scenarios, by mid-millennium less than six percent of today's tundra would remain; saving roughly 30 percent would only be possible with the aid of ambitious greenhouse-gas reduction measures. Otherwise, Siberia's once 4,000-kilometre-long, unbroken tundra belt would shrink to two patches, 2,500 kilometres apart, on the Taimyr Peninsula to the west and Chukotka Peninsula to the east. Interestingly, even if the atmosphere cooled again in the course of the millennium, the forests would not completely release the former tundra areas.

Read more at Science Daily