Scientists spent 25 years tracking the Amazon’s ‘ghost dog’ and what they found changed everything |
Somewhere in the dense, unbroken forests of Bolivia and Peru, a strange-looking dog is going about its day largely unbothered by the fact that science barely knew it existed. It has a large head, short rounded ears, partially webbed paws, a bushy tail, and a coat that shifts between blackish grey and reddish brown. It isn’t a fox, it isn’t a wolf, and it isn’t related to any domestic dog you’d recognise. It is Atelocynus microtis the short-eared dog and it has spent decades earning the nickname perro fantasma, the ghost dog, by being almost impossible to find. Now, a new study drawing on nearly 25 years of camera trap data has quietly dismantled everything scientists thought they knew about this species, starting with how rare it actually is.
Meet the ghost dog: the Amazon ‘s least-known canid and only endemic wild dog
The short-eared dog holds a genuinely unusual place in the animal kingdom. It is the only canid species entirely endemic to the Amazon rainforest, meaning it exists nowhere else on Earth. Phylogenetically, it sits in its own monotypic genus, Atelocynus and is most closely related to the bush dog (Speothos venaticus), another secretive Amazonian canid. Its known range runs across the lowland Amazon from southeast Colombia through Ecuador, Peru, Bolivia, and into Brazil, though confirmed sightings have always been scarce, particularly in the eastern reaches.For most of recorded natural history, almost nothing concrete was known about how the ghost dog lived, how many there were, or exactly what kind of habitat it needed. The species was assessed as Near Threatened on the IUCN Red List, but even that categorisation came with significant uncertainty because the data barely existed. A study published in Royal Society Open Science previously found that roughly 30% of the short-eared dog’s distribution could be lost within three generations due to Amazon deforestation but even that warning was built on a thin base of confirmed records. The ghost dog had simply avoided science for too long.
25 years of camera traps, 594 independent events, and the largest dataset ever assembled
The study that changed things was published in March 2026 in the open-access journal Neotropical Biology and Conservation, led by Robert Wallace of the Wildlife Conservation Society. What the researchers did was deceptively straightforward: they placed camera traps across key lowland forest areas in Bolivia and Peru, waited, and collected images across nearly a quarter of a century.Between 2001 and 2024, the team conducted 34 intensive camera trap surveys across two major landscapes the Greater Madidi-Tambopata Landscape in northwestern Bolivia and southeastern Peru, and the Llanos de Moxos Biocultural Landscape in northern Bolivia. Across those surveys, they systematised 500 distribution records for the species in Bolivia alone, and generated 4,635 photographs yielding 594 independent short-eared dog events. That is the largest confirmed record set ever compiled for this species anywhere in its known range. Short-eared dog detections showed up in 21 of the 34 surveys a detection rate that surprised even the researchers behind the project.
Ghost dog abundance, density estimates, and a surprising daytime behaviour pattern
The headline finding of the study is one that wildlife biologists didn’t necessarily expect: the short-eared dog is considerably more abundant than previously assumed. Using comparison data from ocelot (Leopardus pardalis) camera trap capture rates, the researchers estimated an average density of approximately 15 individuals per 100 square kilometres. That puts the ghost dog above jaguars in terms of relative density, more common than the Amazon’s apex big cat, though still less abundant than medium-sized carnivores like ocelots themselves.Just as significant was what the camera traps revealed about when the ghost dog is actually active. It had long been assumed, largely by default, that the species was nocturnal or crepuscular. The data from Bolivia overturned that assumption entirely. Short-eared dogs turned out to be primarily diurnal, with activity peaking between 6 am and noon. The reason they are so rarely seen, then, is not because they move under cover of darkness it is because they move deep inside intact forest, in terrain that humans rarely enter and cameras were rarely deployed in before this study. The ghost wasn’t hiding in the night. It was hiding in the trees.
Terra firme forest dependency and what it means for short-eared dog conservation
The study’s ecological findings carry real weight for conservation planning. The data made clear that the short-eared dog is a strict forest specialist not just an Amazon species broadly, but one with a specific preference for terra firme, the upland forest away from rivers and floodplains. It actively avoids transitional habitats and open areas. Where the forest canopy is intact and continuous, the ghost dog appears. Where it isn’t, the species essentially disappears.This tight habitat dependency explains the species’ invisibility and also its vulnerability. Deforestation in the Amazon does not need to reach the ghost dog’s exact location to threaten it fragmentation of the forest matrix, even at some distance, reduces the connectivity that the species depends on. Research published in Royal Society Open Science found that forest cover and forest edge density were both strong predictors of short-eared dog space use, with occupancy declining sharply as forest fragmentation increased. The ghost dog needs large, unbroken forest and in the Amazon, that is an increasingly contested resource.
Protected areas, Indigenous territories, and the conservation strategy that’s actually working
One of the more policy-relevant findings from the new research is where short-eared dogs were most reliably detected. The camera trap data showed higher relative abundance inside nationally protected areas and in Indigenous territories that overlap with those protected zones, compared to unprotected land. This matters because it confirms, at least for this species, that formal protection combined with Indigenous territorial management produces measurable conservation outcomes.The researchers were direct about what this implies. The most important management strategy, they noted, is protecting the Amazonian forest canopy through the creation and effective management of protected areas, in combination with the sustainable management of Indigenous territories. That is not a novel conservation argument, but it is rare to have it backed by 25 years of species-specific camera trap evidence from one of the world’s least-documented carnivores.
Deforestation threat, climate pressures, and what happens if the Amazon loses the ghost dog
Despite the more optimistic abundance estimate, the short-eared dog’s situation is not comfortable. The Amazon is losing forest at a rate that outpaces most conservation responses, and the ghost dog has no capacity to adapt to deforested or fragmented landscapes. It is not a generalist that can shift into agricultural zones or secondary scrub. It is an evolutionary product of deep, intact Amazonian forest, and that is the only place it can survive.Earlier modelling, based on deforestation projections for the Amazon basin, suggested that under a business-as-usual scenario a significant portion of suitable habitat for Atelocynus microtis could be gone within three canid generations. Given a generation length of roughly eight to nine years for medium-sized canids, that timeline is not distant. The ghost dog’s newly discovered relative abundance makes its conservation seem more tractable but only if the forests it depends on remain standing. What the 25-year camera trap study ultimately tells us is that the Amazon has been quietly protecting this species all along. The question is whether the Amazon itself will continue to receive the same protection in return.
