Tag: remote sensing

Chasing the Ghost Dog of the Amazon: New Insights into a Mythical Canid

A 23-year camera-trap study across Bolivia & Peru recorded 594 events of the elusive short-eared dog (Atelocynus microtis), shedding light on its ecological behaviour and relative abundance.

For decades, the short-eared dog (Atelocynus microtis) has been considered one of the least-known carnivores in Latin America, and potentially one of the least-known canids in the world. Because of its highly secretive nature, acute hearing, and strong sense of smell, this cryptic creature has successfully avoided people, leaving biologists with very few direct sightings in the wild. However, a new study published in the open-access journal Neotropical Biology and Conservation now offers insights into the life of this mysterious Amazonian predator.

The Power of Remote Sensing

short-eared dog Atelocynus microtis
Photo of the short-eared dog caught on a camera trap from Bolivia. Credit to G. Ayala & M.E Viscarra.

According to the study‘s lead author Robert Wallace, the turning point in their understanding of the ghost dog came with the advent of remote sensing. When researchers first started capturing pictures of the species with camera traps back in 2001, they knew they were onto something special and envisioned a comprehensive publication.

Over the span of nearly 25 years, scientists systematized 500 distributional records across Bolivia and conducted 34 intensive camera-trap surveys. These surveys took place across the lowland regions of Bolivia and Peru, primarily in the Greater Madidi-Tambopata and Llanos de Moxos Biocultural Landscapes.

Accumulating 594 independent photographic events, this monumental effort represents the largest collection of confirmed short-eared dog records anywhere within the species’ distribution. The lead author highlighted that this research is a:

Wonderful example of how conservation technology and remote sensing – in this case the intensive use of camera traps – can provide substantial data on one of the least known species of the Amazonian rainforests.

Unveiling the Ghost

short-eared dog Atelocynus microtis
Short-eared dog from a trap camera in Bolivia. Credit to G. Ayala & M.E Viscarra.

What exactly does this ghost dog look like? The camera traps revealed a unique physical profile: a dark, dense coat ranging from blackish gray to reddish-brown, a large head with very small and rounded ears, short legs, and a long, bushy tail. Intriguingly, the species even boasts partially webbed paws, a trait that is completely unique among Amazonian canids.

The biggest surprise, however, wasn’t its appearance.

The most surprising aspect of the results was that despite being an almost mythical beast, short-eared dogs are much more abundant than we had imagined.

noted the researchers.

While they are by no means a common species, their camera-trap capture rates and an estimated density of 15 individuals per 100 square kilometers suggest they are not as rare as scientists once feared. The results allowed for the researchers to determine that the short-eared dogs are more abundant than larger carnivores like jaguars, though less abundant than medium-sized carnivores like ocelots. The study also highlighted a small aspect of these canids’ behavioral ecology: they are primarily diurnal, meaning they are most active during the day, with their activity peaking between 6:00 a.m. and 12:00 p.m.

A Forest Specialist in Need of Protection

So, where are these ghost dogs hiding? The data points to intact forests. The short-eared dog is a true forest specialist, showing a strong preference for terra firme – upland forests away from rivers. This specialized, dense habitat requirement is a primary reason for the species’ reduced visibility to humans.

Because their survival is intricately tied to continuous, intact forest cover, the conservation of the short-eared dog depends heavily on the creation and effective management of protected areas. The study found that the relative abundance of short-eared dogs was noticeably higher in national protected areas and Indigenous territories that overlap with these protected zones, compared to unprotected areas.

The most important management strategy is the protection of Amazonian forest canopy for which the creation and effective management of protected areas is the most important element, in combination with the sustainable management of Indigenous territories.

the researchers explained.

While we now know that this elusive canid is secretly thriving in the deep forests of Bolivia and Peru, its future relies entirely on our ability to protect its home.

Original source:

Wallace RB, Ayala G, Viscarra M, Porcel Z (2026) Unveiling the ghost: short-eared dog (Atelocynus microtis) distribution, activity patterns, habitat use, relative abundance, and occupancy in Bolivia. Neotropical Biology and Conservation 21(1): 49-66. https://doi.org/10.3897/neotropical.21.e183324

How can social media users help researchers with ecological or biodiversity studies?

Images and videos on social media platforms like Instagram can help sample species occurrence data. Here’s how

Guest blog post by Melanie Werner

Researchers studying the distribution of species need information about where the plant or animal species in question occurs. This usually requires a field study, which is costly, time-consuming and often CO2-intensive. In addition, not all locations where species occur are accessible to humans, such as high mountain areas. As a result, the full range of species is often not covered and the availability of species information is a major challenge for ecological research.

Social media and citizen science projects offer a great potential to increase this availability. Citizen science refers to the involvement of non-experts in scientific projects. In the case of ecological studies, for example, this is the sampling of species occurrences. Thanks to smartphones – capable of capturing high-quality photos and recording precise locations – the ability to gather and share species information has grown significantly. It’s quite simple: take a photo, save the coordinates, upload it, and researchers can use this data to map species distributions. User-friendly platforms or Apps like iNaturalist have become invaluable for this purpose, enabling researchers and citizen scientists to share and utilize occurrence data to advance ecological understanding.

Process of species occurrence data sampling by citizen science projects.

Although some people are interested in and help with such projects, they do not yet reach a broad mass of people. In our study, we explore the potential of the social media platform Instagram, which is one of the largest social media platforms with 2 billion users worldwide and millions of daily uploads. We hypothesize that even more species occurrence data is being generated on this platform, often without users realising its scientific value.

We searched for Instagram posts from 2021 to 2022 in which the tree species Nothofagus pumilio and the location of the photo can be identified. The deciduous species N. pumilio occurs at the treeline in the southern Andes. The species’ vibrant orange-red autumn colouring, in combination to spectacular landscapes, makes it a popular photo motif for tourists, hikers, and locals alike resulting in numerous Instagram posts. Because treelines are often sensitive to climate change, studying the distribution and dynamics of this species is of particular interest.

Nothofagus pumilio at Mount Fitz Roy, Patagonia (Marina Zvada,  2024 available at: https://unsplash.com/de/fotos/blick-auf-ein-gebirge-in-der-ferne-vMoLf1OrB-k?utm_content=creditCopyText&utm_medium=referral&utm_source=unsplash)

Our study began by searching Instagram for specific hashtags related to the species (#nothofaguspumilio and #lenga), as well as for locations through location tags, hashtags, and descriptions within posts. This approach allowed us to identify as many images with N. pumilio as possible, even if the users were unaware of its presence.

To select suitable posts, we focused on those where the species could be clearly identified, the location was described, and landscape elements (e.g. mountains, waterbodies, glaciers, urban and tourist area) are visible in the image, that could also be recognised in satellite imagery. For each post, we mapped at least one point at the location where the photo was taken. Additionally, we included occurrences of N. pumilio visible in the background, such as autumn-coloured treelines. This approach ensured a more comprehensive and evenly distributed sampling of the species’ range. The next image illustrates the sampling process of a post captured at Laguna Capri with Mount Fitz Roy in the background.

Suitable Instagram post (fernando.v.fotografia, 2022, left) and four transferred points with landscape elements (red boxes, right) that helped to identify the location, which is also described in the post with location tag and post caption.

In this way, we were able to generate 1238 Instagram ground truthing points. These points can be directly used in ecological studies. However, it is important to note that our Instagram ground truthing points, like most citizen science data, is spatially biased. This means that locations can only be captured where people have access, leaving remote or high-altitude areas unrepresented.

For this reason, we added a remote sensing method to our study. As this deciduous species forms mono-species forests at the treeline, it can also be identified in satellite imagery. The occurrence data obtained through supervised classification was then validated using the Instagram ground truthing points. This approach allows us to generate less-biased occurrence data for the species across its entire 2000 km latitudinal range.

All 1238 Instagram ground truthing points sampled by our approach (left) and the remote sensing result (right).

Participating in citizen science projects is an essential contribution to research. We encourage you to explore citizen science platforms such as iNaturalist, eBird, or naturgucker, and to include hashtags and detailed location descriptions – ideally with coordinates – when posting your holiday or everyday photos on Instagram. Every contribution helps!

Research article:

Werner M, Weidinger J, Böhner J, Schickhoff U, Bobrowski M (2024) Instagram data for validating Nothofagus pumilio distribution mapping in the Southern Andes: A novel ground truthing approach. Frontiers of Biogeography 17: e140606. https://doi.org/10.21425/fob.17.140606