The AI analyzed 1,974 unclassified tracks and raised some very troubling questions about the “bird tracks”

How do we know that the first evidence for birds' existence did not be found in bones with a fragile skeleton or even from something simpler and more commonly used like a footprint that was pressed into the earliest clay? The results of a new study challenge traditional beliefs regarding the ancestry of birds, focusing on artificial intelligence as well as an undiscovered discovery that is fossilized tracks.

The study, entitled "Unsupervised Machine Learning Reveals Hidden patterns of Dinosaur as well as Bird Footprints," was directed by Gregor Hartmann of the Helmholtz-Zentrum Berlin fur Materialien und Energie in Germany along with colleagues with The Open University and the University of Edinburgh in the United Kingdom. In analyzing more than 2000 fossil footprints that were not given names, the researchers have revealed a whole new perspective of the evolution of birds as well as when they first came into existence.

Footprints that are historical time capsules for behavior


In contrast to bones, which are the bones of animals from long ago footprints are referred to in the field of trace fossils. They record a specific instant in time, the way animals moved about, the direction it moved, as well as the way it divided its weight. They are often more numerous than bones and remains of skeletons, making an important but under-appreciated sources of paleontological information.

But footprints aren't the only evidence. One foot may make radically different impressions based on the amount of water in the soil and the speed at which the animal moves and also the surface texture of the soil. Anyone who has seen the footprints blur and change shape when sand is wet has experienced the problem in person. This variability for scientists is a constant problem.

Another kind Artificial Intelligence

In order to tackle the complexity of this they used unsupervised machine learning which can let computers recognize patterns and not be instructed on what they should look for. In contrast to traditional methods that depend on the human labeled data it is not possible to embed biased opinions of experts into the data analysis.

The model was based on 1,974 footprints that represent an array of dinosaurs and modern bird species. In order to make the model more durable, the researchers added variations to the dataset by imagining how footprints could alter under various environmental situations. This was a step to ensure that the AI was capable of handling the complex nature that fossilized footprints have.

Instead of assigning labels instead, the system reduced every footprint to just a few fundamental features, and then rebuilt the footprint. By doing this it identified eight important features that differentiate different tracks kinds, such as the spread of toes, depth of impact, heel form and the way that the weight is distributed throughout the feet.

When the model's results were evaluated to existing classifications by experts, the degree of agreement ranged between 80 and 93 percent. The consistency of the results suggests that the model has captured significant biological patterns, not random noisy.

Unpopular information regarding early bird spotting

The most striking result of this study is a collection of three-toed, tiny footprints dating to in the Late Triassic and Early Jurassic period. They have always puzzled researchers due to their resemblance to those of birds that are modern, but they are tens of million of years earlier than the first recognized birds' skeletons.

The AI study found that a lot of the footprints that are disputed cluster more closely with well-known footprints of birds than those from non-bird dinosaurs. If this is the case this could suggest that similar bird-like relatives existed 60 million years before was previously believed.

A shift like this could be profoundly impacting our understanding of evolution and especially the earliest ancestors of the flight. This suggests that the key adaptions associated with birds could occur earlier, and in different environmental environments than the theories currently proposed.

Caution in the mud

In spite of the enthusiasm, researchers insist that the footprints alone will not settle the issue. They are not only influenced through anatomy, but also the environment in that they formed. The possibility of convergence in evolution is still a possibility as dinosaurs that are not related might have had the same foot structure and left similar footprints.

The previous research shows that environmental influences can stretch or compress the toe's impressions, making shapes similar to the appearance of birds. The recent findings support the argument for a pre-historic origins of bird species, they are not able to offer definitive proof.

The Scottish mystery Rediscovered

The research also analyzed the footprints of a controversial group found on Middle Jurassic rocks on the Isle of Skye in Scotland. The footprints, made around 170 million years back within an environment of mud and lagoons are subject to debate over the years.

The AI put a lot of the prints more closely to theropods which is the meat-eating dinosaur group which include the ancestors to birds. However the tracks of some appeared to be like ornithopods the group of dinosaurs that eat plants. The mixed results highlight the difficulties of reading trace fossils, and indicates that several species could contribute to the track site.

The DinoTracker project and future fossil analyses

In order to make their method more accessible, the group created an open-source program known as DinoTracker. It allows users to examine footprint outline against a vast database. It also provides a fast method of evaluating new findings.

Although the software isn't designed to replace the expertise of experts however, it is a step towards more open and transparent methodologically driven methods for paleontology. Recent studies of Earth-Science Reviews and Integrative Organismal Biology are revealing a growing trend towards using machine learning for studying biomorphological patterns This study is directly into that trend.

A footprint that can alter the history of mankind

The possibility that birds may have left traces on Earth longer before their bones indicate is fascinating as well as unsettling. When analyzing footprints from the past using advanced technologies, scientists are beginning to discover patterns that challenge traditional time lines.

If these unpopular tracks are the work of early animals or smart evolutionary mimics is a question that remains an open issue. However, what is certain it is that the earth under our feet could provide answers to questions that bones can't give.