Is it true that birds are "obviously" not dinosaurs?

Jeff Poling

INTRODUCTION

"But it clarifies nothing to say that birds are dinosaurs. Any child can see that birds are not dinosaurs. Likewise, any child can see that whales are not condylarths, although they are members of the same clade. We have always discussed evolutionary lineages, and the dinosaur-bird connection has been discussed since the late 1800's. In this sense, phylogenetic taxonomy has added nothing. But making birds into dinosaurs--that's a new way of thinking about things that has its own shortcomings. Cladistics clarifies the obscure (phylogeny). Yet it attempts to obscure what is obvious even to a child (morphological distance). Hence, children are not likely to be impressed by cladists who are 'cladistic hyper-reductionists.'"¹

The above paragraph was posted by Norm King to the Dinosaur Mailing List. It was posted in opposition to one of the rules of cladistics, that of inclusiveness or inheritance. Simply, the rule states that all descendants of the species that represents the common ancestor of a clade remain a member of that clade, regardless of how differently the ultimate descendant species may have evolved from its ancestor. Under these rules, snakes are tetrapods, humans are mammals and birds are dinosaurs.

Some people argue the necessity of formal paraphyletic groupings, or groupings of animals that do not include one or more descendant species of a common ancestor. Their most frequently used paraphyletic group is a Dinosauria that does not include Aves, or birds. They argue that formal paraphyletic groups are necessary to demonstrate what they consider vast evolutionary differences between ancestor and descendant.

What descendant species should be removed to form the paraphyletic group is not clear nor well understood, a reason why many scientists do not approve of the practice. Supporters of paraphyly usually cite morphological distance, or the extent of difference between certain anatomical features of ancestor and descendant, as the reason for their formation of certain paraphyletic groups.

When arguing whether birds are truly morphologically different enough from their dinosaurian ancestors to warrant making the Dinosauria a paraphyletic group, paraphylists will often summarize their position by such statements as in the opening quote. The message behind those statements is that the morphology, or anatomical characteristics, of birds and dinosaurs is so vastly different that even a child, without the keen powers of observation of an adult, when presented with a bird and dinosaur side by side, would state that the bird is not a dinosaur.

I do not hide the fact that I do not agree with the very concept of formal paraphyletic groupings. However, even if I step into the paraphylists' shoes and look at the world from their viewpoint, I cannot justify a paraphyletic Dinosauria. I do not agree that "any child can see that birds are not dinosaurs." I decided to test whether Norm is correct. To test his hypothesis I formulated a poll and conducted it on my website, Dinosauria On-Line, from January 13, 1997, through February 1, 1997.

THE POLL

Imagine, for a moment, that you are an alien being from Mars. The Martian bureaucracy being what it is, you were sent to Earth despite having no knowledge whatsoever of Earth and its creatures. Upon arriving, you were ushered into a room and sat at a table with these three pictures on it:

Gallus, a bird	Compsognathus	Chasmosaurus
view more detailed pictures

You were told that two of these creatures are dinosaurs, and one is not. You must choose which of these animals is NOT a dinosaur. Remember, you are NOT to make your decision based on what you, as an earth-being, know. You are to choose based on being a Martian that knows nothing about earth-creatures, and must choose which one does not belong based only on the similarities and differences shown in the images.

CHOOSE:

CHOICE	RESULT	PER CENT
A	1680	58.92%
B	165	5.79%
C	1006	35.29%

(Note: neither the animal names nor the poll results appeared in the actual poll as given.)

STATISTICS

To start, we need to determine the statistical parameters that govern this poll.

First, is the sample size sufficient?2 We assume that the population proportion (percentage of people responding a certain way) should be within plus or minus 2 percentage points of the sample proportion, just for accuracy's sake. Since the Internet is a global network, we use the Earth's population of 4 billion. Calculating the sample size, at right, we see the sample size is sufficient for an interval of plus or minus 2 percentage points.

Second, what are the results of the hypothesis test?3 Properly formulating the null hypothesis in this case is a difficult proposition. What, exactly, is the mathematical representation of the qualitative statement "obvious?" Personally, I think that if a specific response is obvious then everybody, or 100% of the respondents, should choose that response. To be fair, one might choose 3/4, or 75%, as the limit; however, I'll choose 2/3, or 66%. Therefore, the null hypothesis is that the population proportion is equal to or greater than 66% when the answer is "obvious." The alternate hypothesis will be that the population proportion is less than 66%. The Type I error coefficient is set to 5%, meaning we want to risk only a 5% chance of rejecting a response as "obvious" when it really is. Running the hypothesis test, at right, we see that any response that was given by at least 64.55% of the respondents in the sampling can be considered "obvious."

Lastly, what is the probability of making a Type II error, accepting a response as "obvious" when it is not?4 The graph, at right, shows the probability of making a Type II error when the population proportion is as shown. For example, there is approximately a 70% chance of accepting a response as "obvious" when only 65%, or 1% less than our limit for "obvious," of the population would answer that way.

ANALYSIS

Clearly, based on the hypothesis test, one must reject the hypothesis that it is "obvious" that the bird is not a dinosaur. No one choice received enough responses to equal or exceed 64.55% of the total responses, the amount necessary to be "obvious."

It should be noted that the distribution of the responses is probably not an artifact of someone who understood the "game" skewing the poll. Since I could not prevent people from responding more than once, a single individual theoretically could have hit the "submit" button multiple times and skewed the results. However, I monitored the results, checking the distribution several times a day. The distribution never deviated by more than one or two percentage points from the final distribution shown above.

In formulating the poll, I was very deliberate in which dinosaurs I chose. I picked Gallus and Compsognathus because of their close relative sizes. I originally picked Triceratops for my other dinosaur because I do not think Triceratops visually looks as close to Compsognathus as Gallus, thus giving a solid visual demonstration of the different morphologies within the Dinosauria. A huge visual difference between two animals unquestionably in the same group helps in determining whether a third animal, if picked by a sufficient number of people, is the "obvious" exclusion. A good, skeletal Triceratops image was not available, so I used the closely related Chasmosaurus instead.

I controlled as much as I could the factor of bogus visual cues. I did not want differences in the size or color of the images, for example, to create false characteristics for comparison that would bias the results. The cues controlled for include:

• Image height and width. Differing height and width among the images could fool the viewer into thinking the relative proportions of the animals within the images were different than what was actually shown. Therefore, I made sure all three images were exactly the same height and width.
• Resolution. The smaller the picture, the more difficult it is to pick up detail. This could bias the results by making the viewer miss some detail, or think they see something that is not really there. Thus I gave the viewer three, high-resolution images to view in addition to the thumbnail images.
• Scale. Since the poll was predicated on looking only at anatomical characteristics, I needed to make sure choices did not cause a bias due to their relative sizes. Further, objections had been raised in Dinosaur Mailing List debates that size alone was one of the characteristics that made dinosaurian paraphyly "obvious." I thus felt that objections would be raised to my methods if the bird and other theropod were shown as the same size. Therefore, I made sure the bird and other theropod were close to scale, their relative sizes appearing in the images as they would in real life, and reduced the ceratopian to Compsognathus size to eliminate choice of the animal simply because it was, in real life, so much bigger than the other two.

While conducting the poll I received an e-mail stating that the directions were not worded properly. The author stated that if I truly wanted to discover what people thought were dinosaurs that I should have them pick the dinosaurian pair, rather than the odd creature out. However, this was not what I was testing for. The hypothesis as stated by Norm is that it is "obvious" that birds are not dinosaurs, rather than it being "obvious" that certain animals are dinosaurs. In polling, the same question worded in an affirmative manner or negative manner can produce very different results even though the answer is conceptually the same for both.⁵ Therefore, the wording of the instruction, "choose which of these animals is NOT a dinosaur," was chosen very carefully to reflect the information that was truly being sought.

One question about the poll that needs addressed is whether people really "played the game" properly. Did they truly step back and "forget" what they already knew about dinosaurs? Did they truly pick from an unbiased standpoint? Since the results were skewed toward the bird not being a dinosaur, a position I do not agree with, I obviously choose to conclude that those who chose Gallus were not following directions. In reality, the results could be interpreted two ways: either that 58.92% of all respondents honestly thought Gallus is not a dinosaur, or that they were unable to forget that which they already knew and this biased the results toward Gallus. Perhaps it was a combination of the two. Regardless, assuming the results were genuine and honest, the fact that 41.08% of the respondents picked what Norm would consider the "wrong" answer debunks the hypothesis that it is "obvious" that birds are not dinosaurs.

One really interesting result of the poll is that 5.79% of all respondents picked Compsognathus, a theropod dinosaur, as not being a dinosaur. When guessing what the results might be before beginning the poll, I never in my wildest dreams thought that anybody would choose Compsognathus. Assuming that the respondents were sincere in their choice, rather than doing it as a joke or to be a jerk, it suggests that the differences between bird and non-avian dinosaur are less "obvious" to the general public than anybody imagined.

Given that result, one must wonder what would have happened had I chosen a theropod more closely related to birds than Compsognathus. If I had used Deinonychus or Velociraptor with their amazingly bird-like skeletons the results probably would have been even further in favor of Chasmosaurus not being a dinosaur.

I am sure somebody will object to the fact that I used skeletal reconstructions as the basis for comparison. A person might say that if you put skin on the animals there would be no question that Gallus would have been the "obvious" choice as the non-dinosaur, no doubt using the bird's feathers as the deciding factor. In fact, I think using fleshed out dinosaurs would have made Chasmosaurus even moreso the "obvious" choice.

Late last year the Chinese discovered a new dinosaur that they named Sinosauropteryx prima. This animal has been described by Dr. Phil Currie of the Tyrrell Museum as being a compsognathid dinosaur. If true, this makes Sinosauropteryx a close relative of Compsognathus. Two specimens of Sinosauropteryx have been found, and the fossils show what have been described as unmistakable down-feather impressions. If the published analysis (to appear soon in Nature) reveals that Sinosauropteryx was truly feathered, and was closely related to Compsognathus, then it is likely that Compsognathus, too, was feathered. Putting a feathered Gallus next to a feathered Compsognathus and a likely naked, or nearly naked, skinned Chasmosaurus would further result in more respondents choosing Chasmosaurus, not Gallus as our hypothetical objector might think.

CAVEATS

Although I regularly claim infallibility and omnipotence as character traits, I must, to be honest, point out possible problems with the poll.

That actual parameters of the poll do not wholly represent the hypothesis as worded by Norm. Norm states that his conclusion that ". . . birds are not dinosaurs . . . is obvious even to a child." In a poll conducted in the manner in which I gave it, it is impossible to restrict respondents only to children. Further, this was not a truly random sampling, as assumed in the statistical analysis above. Not only did I not poll only children, I did not poll just "any" person. For the results to fulfill statistical requirements of population representation, they must be from a truly random sample. The actual respondents of the poll were a specific subset of the total population, namely dinosaur enthusiasts that were readers of my webpage. Therefore, the actual hypothesis that was tested was "it is 'obvious' to dinosaur enthusiasts, of all ages, who are on the world wide web, who visit Dinosauria On-Line, that birds are not dinosaurs."

The fact that my poll was not a truly random sampling should not cause the reader to disregard its results. It should not be overlooked that respondents who were most likely teens and adults, who already knew about the characteristics of dinosaurs, still were not subtly biased enough to prove the "obviousness" of Norm's hypothesis.

What factor cultural bias⁶ may have played in the poll is anybody's guess.

The fact that people were looking at fixed two dimensional images rather than being able to examine three dimensional models may also have had an effect.

The biggest caveat of all is that there was an answer missing: none of the above. As far as I, and most dinosaur paleontologists, are concerned, birds, and therefore Gallus, are dinosaurs. All three of the responses were "wrong."

CONCLUSION

Some scientists and enthusiasts support the concept of a formal paraphyletic Dinosauria. They believe that morphological distance is so acute between avian and non-avian dinosaur that birds should be removed from the Dinosauria altogether. They often characterize these differences as "obvious" to the general public, specifically children. The poll conducted on Dinosauria On-Line to discover whether the differences truly are obvious shows that the differences are not, in fact, "obvious." When given a choice between Gallus, Compsognathus and Chasmosaurus, too few of the poll respondents chose Gallus, a bird, to support the hypothesis. Indeed, had more a more bird-like non-avian theropod been used, the support for the hypothesis may have been even weaker.

1. King, Norm. cladistic wrap-up, please! Public e-mail, Dinosaur Mailing List, 2 Dec 1996.
2. David R. Anderson, Dennis J. Sweeney and Thomas A. Williams, Statistics for Business and Economics (St. Paul: West Publishing Company, 1993), 781-783.
3. Ibid., 313-317.
4. Ibid., 320-322.
5. Dr. Glenn W. Milligan, MBA 870 Statistical Analysis for Decision Making class lecture, The Ohio State University Columbus campus, Autumn 1993.
6. Dr. Glenn W. Milligan, MBA 870 Statistical Analysis for Decision Making class lecture, The Ohio State University Columbus campus, Autumn 1993.