Just because children prefer Tyrannosaurus, the process of biological evolution may be misunderstood!

Produced by: Science Popularization China

Author: Ye Shan (PhD in Earth Sciences, University of Wisconsin-Madison)

Producer: China Science Expo

Dinosaurs disappeared 66 million years ago, but driven by movies, comics and science fiction, rounds of "dinosaur fever" have continued around the world. There are many types of dinosaurs, and everyone who likes dinosaurs may have their own favorite. Someone conducted a rough survey of primary and secondary school students in the United States and Canada and found that the most popular dinosaur among local children is the ferocious Tyrannosaurus.

(Photo source: Veer Gallery)

But did you know that children’s love for Tyrannosaurus fossils may have affected scientists’ research on the changes in dinosaur diversity…

Seeing this, you may think that this article talks about paleontological issues, but you are totally wrong. In fact, this is a statistical issue. Let me explain why.

Statistical concepts in paleontological research: sampling bias

Studying the diversity of ancient organisms is an important task for paleontologists, as it can help us understand the process of biological evolution.

In 1860, John Phillips of Britain accomplished a "feat". He summarized all the fossils known in the world at that time, counted the age information and paleontological species of each fossil, and then drew a curve reflecting the changes in paleontological diversity during the entire Phanerozoic (541 million years ago to the present).

John Phillips Diversity Curve

(Image source: self-made by the author)

He found that the evolution of organisms during the Phanerozoic was not smooth, and experienced two major setbacks. These two setbacks divided the evolutionary history of the Phanerozoic into three stages. According to Mantell's (the first person to discover dinosaurs), Phillips named these three stages the Paleozoic, Mesozoic, and Cenozoic.

Following Phillips's pioneering work, many others have made similar attempts. As more and more fossils are unearthed around the world, this curve has become more and more detailed. In 1984, David Raup and Jack Sepkoski of the University of Chicago in the United States compiled global marine fossil data and redrawn a marine biodiversity curve throughout the Phanerozoic with a higher resolution.

Jack Sepkoski Diversity Curve

(Image source: self-made by the author)

In this new curve, Sepkoski found five very obvious periods of diversity decline, which were located in the late Ordovician, late Devonian, late Permian, late Triassic and late Cretaceous, which are the "five mass extinctions". Other scientists verified this work and obtained similar results, so the concept of the five mass extinctions was soon widely accepted.

However, as earth science research continues to deepen, a new generation of scientists has gradually raised some questions: there may be statistical sampling bias in the research of Sepkoski and others.

Are you afraid of lying statistics?

Statistics is a discipline with a wide range of applications. Whether in the field of natural sciences or in the field of socioeconomics, scholars need to use statistical methods to process data in order to discover the patterns hidden behind the numbers.

However, statistics is a double-edged sword. It can help us see the essence of things, but it can also tell us lies. Mark Twain once said jokingly: "There are three kinds of lies in the world: lies, damned lies, and statistics." As long as you are not careful, you will be deceived by the results of statistics and make a big mistake.

(Photo source: Veer Gallery)

Among the many lies associated with statistics, the most common one is sampling bias. There are many types of sampling bias, including generalization, healthy worker bias, and survivor bias. Most of them are caused by small samples or non-random sampling.

For example, some people use the fact that Bill Gates did not graduate from college as an excuse to promote the "theory of the uselessness of studying", which is a case of generalizing from a single example; some people want to study the physical condition of urban residents, so they choose the workers in the city as the research subjects, but they do not take into account that the workers are in better physical condition than most ordinary people, resulting in a "healthy worker bias"; some people count the purchase of tickets on trains that have already started, which is a "survivor bias" because people who have not bought tickets cannot get on the train at all.

(Photo source: Veer Gallery)

Huh? Paleontology can also be related to sampling bias?

Yes, sampling bias also exists in paleontological research. Peter Sheehan, a paleontologist at the Milwaukee Museum, analyzed the spatial distribution of fossils from different ages. He found that if sedimentary rocks formed in a certain geological age are exposed on the surface on a large scale, it is easier for us to find fossils from this age.

(Photo source: Veer Gallery)

This is actually easy to understand. Fossils are almost always dug out from sedimentary rock layers. If sedimentary rocks from a certain period are exposed on the surface, fossils from that period are easier to find. On the contrary, if sedimentary rocks from a certain period cannot be found on the surface, even the most skilled paleontologists will find it difficult to find fossils from that era. This shows that the amount of sedimentary rocks on the surface will affect the sampling of paleontological fossils.

So what determines the preservation of surface sediments of different ages? The biggest controlling factor is time.

You can imagine what a set of sedimentary rock layers will experience after they are formed. They will be exposed to the wind, sun, rain, and even destroyed and exploited by other geological events. Therefore, the older the rock layers, the less likely they are to be preserved to this day. Once the rock layers are destroyed, the fossils inside will be gone and will not be known to us.

The classic example is in the western United States. In the middle and late Cretaceous, the climate was hot and the sea level was much higher than it is today. At that time, the western United States was flooded by seawater, forming an intercontinental sea. The flooding seawater eroded the strata of the early Cretaceous period. In places such as Colorado and Wyoming in the western United States, all the strata formed in the early Cretaceous period disappeared.

The western United States had a good environment in the early Cretaceous period and was an ideal habitat for dinosaurs. However, due to the lack of strata, few dinosaur fossils from the early Cretaceous period can be found in those places today. From the diversity curve, the local dinosaur diversity reached a low point in the early Cretaceous period. Could it be that the dinosaurs in the western United States experienced a mass extinction in the early Cretaceous period? No, it's just sampling bias at work.

The impact of sampling bias goes beyond this. According to the "law of superposition" proposed by Nicholas Steno, the founder of geology, newly formed strata will cover the older strata. Therefore, without considering the impact of later geological events, the older the strata, the deeper they are usually buried. Compared with the sedimentary rock layers exposed on the surface, it is difficult for us to access those deeply buried rock layers. Even if these deeply buried sedimentary rock layers are well preserved, it is still difficult for fossil collectors to obtain fossils from these rock layers deep underground.

(Photo source: Veer Gallery)

Combining these circumstances, we can know that the older the fossils, the harder it is to find them. They have either been destroyed long ago in geological history, or are now buried deep underground where we cannot reach them. Therefore, the earlier the geological age, the more biodiversity may be underestimated.

The factors that affect fossil excavation are not only time, but also national borders...

In addition, socioeconomic factors also affect the sampling of paleontological fossils.

We assume that a long time ago, there was a large forest with a suitable climate, in which many ancient creatures lived. After they died, they were buried in various corners of the forest, and some of them formed fossils.

Tens of millions of years later, the forest no longer exists, leaving only a group of fossil-rich strata. Coincidentally, a national border runs right through the middle of this group of strata. One side of the border is a developed country, while the other side is relatively poor.

(Photo source: Veer Gallery)

The developed countries have a prosperous economy, and scholars working there have received ample research funding to support their large-scale field surveys, so fossils distributed on the developed countries side have been discovered one after another. When scientists draw the global paleontological diversity curve, the fossils on the developed countries side are taken into consideration.

But on the other side of the border, the situation is different - this country is relatively poor and cannot support scientists to conduct field surveys. The fossils in its territory may still be buried underground, waiting to be discovered. It was originally the same forest in ancient times, but because of the division of regions by humans, the fate of fossils in different regions is very different. Recent studies have shown that if the fossil sampling rate in the world reaches the level of developed countries, we can discover at least 640,000 new fossils. For scientific research, this is also a sampling bias that deserves vigilance.

Children: I love Tyrannosaurus ~ Other fossils: It’s not fair!

Now let's go back to the topic mentioned at the beginning of the article. When scientists restored the diversity curve of North American dinosaurs, they found that the diversity of dinosaurs reached a peak in the Campanian period of the Cretaceous period (about 83 to 72 million years ago), and then gradually declined until they were completely extinct in the asteroid impact event 66 million years ago. Based on this curve, some scientists believe that dinosaurs experienced a slow extinction process in the late Cretaceous period, and their ecological chain had long been shaky, and the asteroid impact was just the last straw that broke the camel's back for the dinosaurs.

But another group of scientists believe that this curve is unreliable because it is affected by sampling bias. When Americans built railways in the western wilderness, they cut open a lot of mountains, and the strata in those mountains happened to be the Campanian strata formed in the Campanian period. The railway construction exposed the Campanian strata that were originally buried deep underground to the surface, making the fossils contained in them easier to find. This is unfair to the fossils in other strata and will lead to sampling bias.

(Photo source: Veer Gallery)

In addition, because children love Tyrannosaurus the most, curators of museums around the world have sent people to focus on looking for Tyrannosaurus fossils to attract more young tourists. The Campanian period was home to Albertosaurus, Daspletosaurus, Gorgonosaurus, Deinodon, and other Tyrannosaurus dinosaurs. It was the era with the most Tyrannosaurus, so it became the focus of excavation. These strata have been searched repeatedly, and while Tyrannosaurus fossils were unearthed, many other dinosaur fossils were also unearthed, which also led to sampling bias.

Therefore, these scientists believe that the peak of the diversity curve in the Campanian period and the subsequent slow decline are just "false signals" caused by sampling bias. It does not mean that the diversity of dinosaurs experienced a long-term decline at the end of the Cretaceous period.

Conclusion

The examples mentioned above remind us that the trap of sampling bias is everywhere in paleontological research. Scientists in this field must be vigilant about this and have careful thinking and thorough consideration when using statistical tools, otherwise they will be "fooled" by the superficial results of statistics.

References:

【1】Chiarenza, A., et al. (2019) "Ecological niche modeling does not support climatically-driven dinosaur diversity decline before the mass extinction." Nature Communications

【2】Raja, N., et al. (2022) "Colonial history and global economics distort our understanding of deep-time biodiversity." Nature Ecology & Evolution

【3】Ye, S. and SE Peters. (2023) "Bedrock geological map predictions for Phanerozoic fossil occurrences." Paleobiology

Editor: Sun Chenyu

(Note: Latin text should be italicized.)