July 22 , 2016

by Don Brinkman

I started at the Tyrell Museum in 1982. Back then, the dinosaur renaissance was in full swing. The public were clearly interested in understanding dinosaurs as living animals: they wanted to know how dinosaurs lived and interacted with one another, what they ate, and what their lives were like. And the Tyrrell’s approach to research was to be organized similarly: we were to undertake field work and research with a paleoecological focus. But there was a problem: while we had good phylogenetic methods which allowed us to understand the ancestry of dinosaurs, paleoecology was still viewed as little more than speculative story-telling. Our challenge, then, was to work out how to understand dinosaurs as living organisms in long-gone environments without foregoing scientific care or rigor.

Badlands of Dinosaur Provincial Park, taken from the air. This area is now a natural preserve, and only accessible by guided tour or permit.

One approach was already underway when I joined the team. This was the study of bonebeds in Dinosaur Provincial Park. Virtually no field work had been done in the park between the mid 60’s and the late 70’s, the time during which the Dinosaur Renaissance had taken place. This is partly because paleontological work in the province during this time was focused primarily on non-dinosaur vertebrates, particularly mammals and fish, and partly because the park administrators felt that their job of preserving the fossils present in the park was best done by preventing any excavation. However, dinosaurs from the park were in many of the museums and research institutes in North America and around the world, and much of the contemporary dinosaur knowledge was based on these specimens. So any paleoecologically-based studies that involved dinosaurs from Dinosaur Provincial Park relied on these museum specimens.

One kind of study that was undertaken focused on taphonomy, which is what happens between the time an animal dies and is preserved as a fossil. A survey had been undertaken of specimens in museum collections, and it was found that while in most of the fossil bearing formations of the western United States, like the Morrison Formation, dinosaurs were represented by isolated elements, most of the specimens from Dinosaur Provincial Park were articulated skeletons (Dodson et al. 1980). This gave the impression that there was something unusual about the formations exposed in the park leading to the preservation of complete skeletons. So this is what Phil Currie was expecting to see when he started doing field work in the park in 1979.

However, it didn’t take him long to realize that it wasn’t that there was a preferential preservation of articulated skeletons. Rather everything was more abundant. The dominance of articulated skeletons in museum collections was because of a collecting bias toward complete skeletons, whereas isolated elements were left behind. This was perfectly understandable when the goal of collecting was to provide material for taxonomic studies and to build museum exhibits. Currie also realized that isolated elements were not randomly scattered through the beds. They tended to occur in distinct layers, sometimes very concentrated. These concentrations were referred to as bonebeds. He felt that by studying bonebeds using the approaches to taphonomy that had been developed, information about the communities and environments at the time the dinosaurs were living (ie., paleoecology) could be obtained.

One way that he thought that data from bonebeds could contribute to paleoecological studies was by providing insight into the relative abundance of animals in the original community. The bonebeds were mixed and transported assemblages, but it was assumed that the abundance of an animal in the bonebed should be a reflection of the abundance of the animal in the original community. The taphonomic processes leading to bonebeds are different from the taphonomic processes leading to preservation of articulated skeletons, so this could be tested by comparing the abundance of an animal in a bonebed to the abundance of that animal as an articulated skeleton. This test could be done because of the high number of articulated skeletons that had been collected (over 300 at the time) and a system of marking quarries that had been initiated in the ‘30s. Phil Currie started by mapping bonebeds and recording what was seen on the surface, particularly the kinds and relative abundance of taxa.

A pattern soon emerged. Hadrosaurs were, by any measure, the most abundant kind of dinosaur present, usually around 60%. Ceratopsians and carnivorous dinosaurs constituted another 10% each. The remaining 20% consisted of a variety of herbivorous dinosaurs. There were exceptions to the pattern – pachycephalosaurs were more abundant as isolated elements than they were as articulated specimens, but this was easily explained because of the high preservation potential for the skull caps. More intriguingly, there were examples of bonebeds that were unusual in being dominated by the remains of one kind of dinosaur that was otherwise rare in the community.

And the dinosaur was always a ceratopsian.

Some of these ceratopsian-dominated bonebeds were incredibly rich. In one case the exposed bonebed was covered by gravel formed of fragments from the eroding bones. There was no question that the behaviour of the animals played a part in the formation of the bonebed. For some reason, a large number of individuals must have been together at the time they died. A study of this bonebed was planned. An archaeological approach was undertaken.

The bonebed was marked off in meter square grids and each meter was excavated systematically. The identity, size, and orientation of the bones were recorded. After the first season of excavation, a clear pattern emerged. Both juvenile and adult individuals were present, a distinct alignment of bones was seen, and a bias in the kinds of elements was present, small round bones like phalanges being underrepresented. The key element observation for biological interpretations were that it was indeed dominated by a single taxon of ceratopsian, Centrosaurus apertus, and that large numbers of juveniles and adults must have been together at the time of death.

A section of the ceratopsian bonebed currently on exhibit at the Royal Tyrrell Museum of Palaeontology. For exhibit a three by six meter section of the bonebed was removed in 1 meter by 2 meter blocks. Each block was prepared with bones left in place and the pieces were reassembled in the exhibit.

Looking at modern analogues, the best fit was that ceratopsians were herding animals and that a large number of individuals from a herd had died at one time in one place. Crossing a river in flood was envisioned as the kind of killing mechanism that would cause this. It was proposed that herds formed during a north-south migration of thousands of kilometers in response to seasonal changes in climate. Work continued for another 9 years, and eventually a master’s thesis was written analysing the data that was collected.

When I started to develop a paleoecologically directed field-based research program in the early 80’s, I had an interest in a different kind of bonebed. These were vertebrate microfossil localities (also referred to as microsites) -- bonebeds dominated by elements of small size, usually in the one to five millimeter size range.

It had long been recognized that such localities preserve a wide diversity of taxa:

Much of this diversity were rarely found outside these settings, and that large sample sizes could be obtained by a bulk sampling technique in which matrix (the rock and substrate in which the fossils are preserved) from the locality was soaked in water to break it down and screened -- a process referred to as screen washing.

The fossils would remain in the insoluble residue, and this could be examined under a microscope to pick out the fossils. I already had experience working in vertebrate microfossil localities from my undergraduate days working for Dr. R. C. Fox at the University of Alberta, who was processing vertebrate microfossil localities in a search for mammals. One of my jobs was to sort the microvertebrate concentrate, and from this I was familiar with the diversity of taxa and elements present. This led to wondering if these could be used for paleoecological studies in some way.

Inspiration on how to make studies of vertebrate microfossil localities the basis for paleoecological studies came from a research project undertaken by Peter Dodson (Dodson 1983). As part of the survey of bonebeds, Peter surface collected from a large number of localities and statistically compared the relative abundance of taxa present in the collections that were made. This was innovative, in that samples were taken from a large number of localities and statistical approaches were used to identify patterns of distribution. This was in contrast to studies that focused on one or two particularly rich localities in an attempt to get a full understanding of the diversity present. I felt that I could adopt this comparative approach, and combine it with previously established techniques for bulk sampling such localities using underwater screen-washing techniques that I had learned from working for Dr. Fox as an undergraduate. Drawing on my exposure to sedimentology, I felt that looking at patterns of abundance relative to depositional environment might be a way of distinguishing between different communities. At the prodding of the sedimentologist on staff at the museum, David Eberth, I also looked at abundance relative to stratigraphic position. The goal of this study wasn’t to reconstruct trophic structure within a community, but instead was to identify how communities differed in terms of kinds and relative abundance of taxa present.

My first summer’s field work in 1985 involved sampling five vertebrate microfossil localities in three different depositional environments. Two of the sites were deposited in small, temporary ponds that formed as flood waters retreated. Two were deposited in in river deposits, and one in an ox-bow cut-off. Based on this small sample, there seemed to be a strong correlation between abundance and environment of deposition – frogs, salamanders and teleost fish were distinctly more abundant in the small pond deposits, while crocodiles, the fresh-water ray Myledaphus, and gar were more abundant in the river deposits. However numbers were low, not of specimens per locality, but of localities. Each locality was well sampled, but each depositional environment was represented by only a few localities. For the statistical analysis, each site was a single sample, so the sample size was only 5, much too small to be shown to be significant.

Thus the goal for the following summer was to collect from more sites to test whether or not these patterns were really significant. By the end of that summer, a total of 22 localities had been sampled, and the pattern was quite different (Brinkman 1990). As the specimens were identified and counted and the localities were compared, a different pattern emerged. I had anticipated that a correlation could be made with depositional environment, but that proved not to be the case. Rather, a significant correlation with stratigraphic position was present. Frogs, salamanders, and teleost fish were generally more abundant in localities low in section than they were higher in section, while crocodiles, gar, and Myledaphus were more abundant in localities higher in section than they were lower in section. With the large number of specimens present (22 localities, each with generally more than 500 specimens), statistical approaches could be used to demonstrate that the differences were significant. Since the sea that covered the central part of North America was extending westward from central Saskatchewan to western Alberta during this time, stratigraphically higher localities would have been geographically closer to the inland sea. Thus what I was seeing was variation in distribution of taxa across the coastal plain. Those that were abundant higher stratigraphically would have been more abundant in coastal areas than they were in inland areas. Most of the patterns made sense based on what we know of the biology of living relatives but there was one surprise: ceratopsians were more abundant in coastal areas than they were in inland areas. What did that mean biologically? A couple of different ideas of ceratopsian biology had already been presented. The general wisdom was that certopsians were upland animals, analogous to the rhinoceros. However, Tom Lehman (1987) had looked at distribution of dinosaurs at the very end of the Cretaceous and found that Triceratops was found in more coastal areas. Thus he proposed that it was more of a swamp dweller. Rather than the rhinoceros, the hippopotamus might be the best modern analogue. The distribution of ceratopsians in Dinosaur Park suggests that this was a common pattern for the group as a whole.

This presented another anomaly: if ceratopsians preferred swamps, then why were they herding animals, and why were the bonebeds mostly found low down in stratigraphic position (that is, in an inland setting)? To bring these conflicting observations together another scenario had to be developed. Ceratopsians are nesting animals, so there must have been a time when they stayed in one place. Their abundance in coastal areas may be because this was the area that they nested. In a wet, swampy environment, ceratopsians would have had a locomotor advantage relative to the albertosaurs Gorgosaurus and Daspletosaurus, the large meat eating dinosaurs that would have been their major predators. With four large feet, ceratopsians would have been able to move in muddy areas more easily than the albertosaurs, whose weight is all on two feet. In a dryer, more upland setting, they would lose this locomotor advantage, so herding may have been a way of avoiding predation there. This scenario implies that the migrations associated with the herds wouldn’t have been from the far north to Alberta as was initially assumed, but east-west migrations in response to seasonal storm events.

But why move inland in the first place? The answer to this came from sedimentological studies undertaken by David Eberth. He showed that there were regularly occurring massive floods of the coastal plain as a result of tropical storms (Eberth et al. 2010). Moving inland would be one way of avoiding getting swamped by these floods.

This scenario ties a number of different observations together, yet it is speculative. To be science it needs to make some predictions that can be tested. And one prediction is that hatchling sized individuals will be found in the more coastal areas rather than the more inland areas. The smallest ceratopsian elements found in the bonebeds are about 30% adult size. A hatchling sized element would be about 10 to 20% adult size.

Hatchling-sized hadrosaurs are common in Dinosaur Provincial Park, but where are the hatchling sized ceratopsians? If the scenario is correct, they should be in the coastal areas, either stratigraphically very high or geographically in more easterly locations. Right now, the number of ceratopsian hatchling sized elements that have been found is very small: only two. Indeed, finding one was so exceptional that a locality was named after it: the Baby Ceratops Locality. Although the sample size is not significant, both are found in very costal setting so it is at least consistent. In the meantime, we keep searching.

Looking at this history of the initial development of bonebed research from a philosophical point of view, a couple of things stand out. One is the importance of scenario building. The idea that a herd of dinosaurs was wiped out while crossing a river in a flood can never be proven, and indeed has subsequently been shown to be incorrect, but it tied together a lot of independent observations and showed why some kinds of data are important and why that data should be collected. And the scenario of an east-west migration in response to seasonal floods makes predictions that can be tested.

Another is the importance of a comparative approach. The ceratopsian bonebed was recognized as unusual and of biological significance because data was present on a large number of bonebeds in the area. The initial assumption was that by studying the one locality intensely an understanding of its history could be developed. However, generalized patterns could not be recognized until similar studies of other monospecific bonebeds were undertaken. These additional studies provided a comparative data base showing that not all have the same cause. Some are best interpreted as nesting sites, others as a result of death from disease, and others death associated with drought. As it became clear that different causes could be recognized, the evidence that the Dinosaur Park ceratopsian bonebed was a result of flooding became stronger.

Along with sharing ideas, sharing primary data is also important. In the case of vertebrate microfossil localities, this means the counts of specimens. All the papers on the vertebrate microfossil assemblages that were published included raw counts of occurrences, not just the results of the analysis, and subsequent studies have used these data to look at questions in different ways.

Underlying all this is the importance of sharing ideas in science. Looking back in the literature at the time the ideas were being developed, it is difficult to find a formal paper presenting the herding ceratopsian hypothesis. Rather, the observations and interpretations were widely disseminated through talks and public presentations. Although informal, the results had a big impact. Herding in sauropods had been documented previously based on track sites, but this was the first time a taphonomic approach to the study of fossil bones had been used. Researchers are often instinctively protective of ideas and data, but making them available proved to be very valuable both for the study itself and for science in general. By presenting ideas informally prior to publication, research benefited from feedback, and science benefited by being able to incorporate the new approaches into other studies. The excavation of the ceratopsian bonebed in Dinosaur Park wasn’t the first time a ceratopsian bonebed had been systematically excavated. Wann Langston Jr. had done exactly this for a pachyrhinosaur bonebed at Scabby Butte. But previous work had focused on taxonomy. The new approach was innovative because it was very much directed at understanding behaviour and biology. It didn’t take long for other researchers to adopt this approach for other monospecific bonebeds, and now such studies are standard practice. Everyone has benefited.
References

Brinkman, D. B. 1990. Palaeoecology of the Judith River Formation (Campanian) of Dinosaur Provincial Park, Alberta, Canada: evidence from vertebrate microfossil localities. Palaeogeography, Palaeoclimatology, Palaeoecology 78: 35-74

Dodson, P. 1983. A faunal review of the Judith River (Oldman) Formation, Dinosaur Provincial Park, Alberta. Mosasaur 1: 89-118

Eberth, D.A., D.B and Barkas, V. 2010. A centrosaurine mega-bonebed from the Upper Cretaceous of southern Alberta: implications for behavior and death events. New perspectives on horned dinosaurs. Edited by MJ Ryan, BJ Chinnery-Allgeier, and DA Eberth. Indiana University Press, Bloomington, IN, pp. 495-508

Dodson, P., Behrensmeyer, A.k. and Bakker, R.T., 1980. Taphonomy of the Morrison Formation (Kimmeridgian-Portlandian) and Cloverly Formation (Aptian-Albian) of the western United States. Mem. Soc. geol. Fr., N.S., 1980 no 139: 87-93.

Lehman, T. M. 1987. Late Maastrichtian paleoenvironments and dinosaur biogeography in the western interior of North America. Palaeogeography, Palaeoclimatology, Palaeoecology 60: 189-217