Slime versus dinosaur

In 2005, researchers made headlines when they reported that they had found intact blood vessels from a 68-million-year-old Tyrannosaurus rex. The discovery raised hopes that paleontologists could get their hands on the flesh and blood of vanished animals. This week, however, other scientists challenged the results, arguing that the dinosaur flesh was in fact just coatings of young bacteria. But the original researchers stand by their results, calling the new argument weak. “There really isn’t a lot new here,” says Mary Schweitzer of North Carolina State University in Raleigh.

That’s the opening for my story in today’s issue of Science on the slime vs. dinosaur controversy triggered by a paper that appeared Monday in PLOS One. I’ve posted the whole thing here.

This sort of story is a bit maddening to write. When I contacted outside experts to comment, I got lots of thoughtful, detailed responses pro and con–which I then promptly squeezed down to a handful of words. The quotations may leave people with the impression that the scientists were being dismissive or too eager to accept the paper or otherwise less than careful in their arguments. And then there were some people who kindly gave me their thoughts, which I didn’t even mention in the article. Ugh.

PLOS One encourages full, open discussion of their papers, so I’m going to post lengthy excerpts from the clashing emails I got from some of the sources below. Maybe this will be of interest only to die-hard biofilm and dinosaur nuts (you know who you are), but it may give you a feeling for how complicated these debate can get beyond short news articles.

Plus, I figured I might as well take use some of Discover’s hard drive space. [Bold-facing is mine.]

First up for the doubters, Frances Westall, an expert on fossil biofilms.

There are a number of misinterpretations and lack of basic data in this paper.  In the first place, it appears that the bone material was extracted from depth (although little information is provided about the actual context of the bones).  How could “swimming bacteria” infest the bones if they were not in a watery medium but rather “buried at depth?  What is the composition of the biofilms?  We are shown FTIR spectra but there is no indication of peak identification.  Since EDX was available, surely the authors made elemental analyses of the biofilms.  What is their composition??

Pyrite framboids are common occurrences in degraded organic matter, their presence relating to the activities of anaerobic microorganisms such as sulphur reducing bacteria.  If the biofilm is modern, this implies that there was previously reducing microbial activity before the formation of this film.  Is it not more reasonable to hypothesise the formation of a microbial biofilm at the same time as diagenetic degradation of the organic matter within the dead dinosaur in anaerobic conditions?  What is the history of the dinosaur bones after death of the animal??  This is important information for interpreting the post-death alterations that could have taken place.

The superb preservation of soft body parts by fossilised microbial biofilms is a well known phenomenon.  For instance, the wonderful fossils found in anoxic volcanic lake sediments at the Messel and Enspel sites in Germany are phosphatised microbial biofilms that grew on the soft body parts. (I don’t have the references with me as I am on travel but here is one and there are the original references within this article (it is in English): Liebig, K., Westall, F., And Schmitz, M., 1996., A study of fossil microstructures from the Eocene Messel Formation using transmission electron microscopy. Neues. Jh. Geol. Paläont. Mh., 4, 218:231) .  If we were provided with the compositional information for the biofilms, it would be easier to interpret the data.  Moreover, the filaments and spheres in the “osteocytes and lacunae” may be mineralogical in origin.  The resolution of the images is insufficient to determine structure and no EDX elemental data are presented.

Although the material was apparently dated as being modern, if the filmy structures are really microbial biofilms, they are more likely to be as ancient as the dinosaurs.  The “modern” date could be due to recent contamination (obviously the dinosaur bones have been handled quite extensively since their retrieval; also lying around in museum drawers or shelves is sufficient for contamination).

In conclusion, the hypothesis that the soft body parts in dinosaur bones and the like could be microbial biofilms rather than the original structures is a tenable hypothesis; such phenomena are common in specific diagenetic environments.  However, the authors fail to demonstrate that the biofilms are modern contamination from “swimming bacteria” and their arguments are not sufficient to refute the studies of Schweizer et al. on an exceptional event.

Now for a couple backers of the paper.

David Martill of the University of Portsmouth generally supports the paper, although he believes the original researchers probably found some genuine dinosaur proteins, if not entire chunks of tissue.

This piece of work demonstrates just how careful we have to be when attempting to analyse fossil bones for traces of original molecules or biomarker molecules. It seems that bacteria have the ability to get into the tightest of spaces and play havoc with the original biomaterials. Not only that, but they can do it at any stage in the fossilization process from the moment of death (of a dinosaur say) to the moment of discovery. I recall finding mould on a fossil insect from Brazil where contaminant fungi produced fruiting bodies over the fossil as it was still stuffed full of useable nutrients. Elsewhere on the slab there were only filaments of the fungus.. but oh boy did they hit it big time on the fossil.

Although this revelation (only really proven because these bacteria were recent and hence could be carbon dated), may seem like a major set back, we shouldn’t loose heart and molecular palaeontologists should persist with their search for the elusive molecules of these long dead animals. But we will have to be much more careful with our analyses, and before we jump to conclusions, we must double and triple check that we are not dealing with contaminant molecules.

Also backing the paper is David Gillan of the Free University in Belgium:

I read the report published by Kaye and colleagues in PLoS One in July 2008 and I fully agree with their conclusions. I am not an expert in dinosaur bones, nor in mineralogy, but to my humble opinion of microbiologist, the presence of bacteria in the bones may explain the structures observed. After the death of the dinosaur most of the flesh would have been completely digested by aerobic heterotrophic microorganisms. However, the inside of the bones is a particular micro-environment : it is filled with organic matter with a lot of red blood cells containing iron. After the death of the dinosaur I can imagine that the interior of the bones would have become anoxic. In such conditions, sulphate reducing bacteria (SRB), which are very abundant in the environment, were able to produce hydrogen sulphide (H2S) while degrading the organic matter (most of the H2S in the sedimentary environment is produced by SRB). This degradation liberates the iron that is complexed to the organic matter. This was shown in one of my papers, in which a living biofilm precipitated iron oxides using Fe(III) organic complexes (Gillan et al. 2000, Geomicrobiol J. 17:141-150). The result of this iron liberation, together with the H2S produced inside the bone by bacteria, is the precipitation of iron sulphides (and/or pyrite) and the formation of framboids. Later, during diagenesis, iron sulphides of the framboids may have been oxidized and replaced by iron oxyhydroxides. Schweitzer and Horny (1999, Ann. Paleontol., 85: 179-192) have discovered iron and oxygen-rich microstructures in the bone tissues of Tyrannosaurus rex but rejected the structures as framboids due to their composition of iron oxides. However, Sawlowicz and Kaye have documented various stages of framboidal pyrite replacement by oxides in dinosaur bones (Mineralogia Polonica Spec. Pap. 29, 184-187). So the ‘‘dinosaurian soft tissues” observed in the bones are probably best explained as bacterial biofilms that are very common throughout nature.

And, finally, let’s hear from Mary Schweitzer, whose original work sparked the new research:

There really isn’t a lot new here, although I really welcome that SOMEone is attempting to look at and repeat the studies we conducted. There are really several errors in wording (and spelling and grammar) in the paper by Kaye et al. that seem to underlie a fundamental misunderstanding of our work, our data and our interpretations.

Something that is not fully appreciated by the outsider is that science is a process. One makes an observation, forms a testable hypothesis about the observation, gathers data, and the data either support or refute the hypothesis. It is then refined and retested. If the hypothesis is tested multiple times, it is strengthened, and eventually moves to become a theory, one of the strongest statements in science.

If one chooses to challenge a hypothesis and the data put forth by another researcher to support it, one is under the obligation to 1. form a hypothesis that provides an alternative to the first; 2. reinterpret the original data presented in such a way that it __better supports__ the new hypothesis than the original, and 3. produce new data that, in addition to the original, more strongly supports the alternative hypothesis than the original. That is the progression of science. Hypotheses are continually being reformulated in this way, because science IS a process, and undergoes revision as new data become available.

We first reported apparent soft tissues in dinosaur material in 2005. We then asked 2 questions, based upon our initial observations (in which, by the way, we were careful not to overstate the data we had at that time). We asked ‘how widespread is this phenomenon?’ and, ‘what is the composition of the material we observed, and is it consistent with original, but altered, organismal components? To address the first issue, we hypothesized that if these were indeed original soft tissues, they would not be limited to a single organism or type, but be represented in bone of other animals of different ages. We then conducted a time-transgressive study examining fossil material from over 30 specimens spanning ages from present day to Triassic specimens. These represented multiple depositional settings, crossed several continents and represented many taxa. Indeed, about the only thing that our samples had in common was that they were bone. The (morphological) presence of at least 3 of the 4 components we first reported were present in about half of the specimens tested, and this did not seem to be independent of environment—sandstones preserved this material best. This allows us to form models of degradation and preservation for further (and ongoing) testing.

To address the second question, we conducted a careful chemical and molecular analyses of just one of the 4 components we reported, that of the collagenous matrix (fibrous, NOT filamentous as stated by Kaye, this is a big difference!). Our multiple analyses conducted in different labs by different investigators supported the hypothesis that material consistent with collagen was preserved in these dinosaur tissues. We did not rely solely on morphological studies, but employed many assays. We did not limit studies to those conducted by the primary author, but our results were validated by others, independently.

While Kaye et al. address the morphology of the structures we observed, and find their own explanations for these, they do not address the considerable chemical and molecular data we put forth to support our hypothesis of endogeneity. We did propose biofilm production as a possible explanation for the material that we see, but we determined that based upon the data we had, microbial biofilms were not a parsimonious explanation for the data (see Schweitzer et al., 2007, Proc. R. Soc. Lond. B). Subsequent studies (paper in preparation) are examining material that was collected from deeply buried specimens and examined by multiple assays, again conducted independently by many investigators, within a month of recovering this material. Looking at all data _together_, the idea that biofilms are completely and solely responsible for the origin or source of the structures we reported is not supported. For example, there is no evidence in the literature that biofilms form branching _hollow _tubes as we observe. Biofilms are universally reported and indeed defined, as flat films containing microbial bodies and their exopolymeric secretions. Kaye et al. state that no microbial bodies were observed, only an ‘undulating’ film. Furthermore, because of gravitational “down’ in the positioning of these bones, it is highly unlikely that a biofilm would be evenly distributed across the channels as we report. Rather a biofilm would almost certainly be thicker at gravitational ‘bottom’, something we did not observe. While Kaye et al, confirm that their 14C data indicate a recent biofilm they do not identify microbial bodies, a hallmark of biofilm. I have seen similar features to what they figure in specimens that have EDTA residue remaining–ie insufficiently rinsed.

We did not do 14 C dating on our material precisely because we used a buffer to extract these vessels and cells that is rich in modern carbon (EDTA) and would naturally leave a residue on vessels and cells liberated from mineralization this way. Because the Kaye et al data have a ‘greater than modern’ 14C date, this means that their biofilms would have to be laid down, and mineralized, more recently than 1950, the zero date for such studies. Kaye et al state that ‘biofilms would coat vascular canals’ but provides no evidence for this argument, and as such, they are engaging in purely a thought experiment at this point.

Kaye et al. did not address our immunological data, and controls. They did not address the phylogenetic analyses of sequence as reported by Organ et al., 2008 or offer any explanation for how ‘biofilm’ proteins from dinosaur could cluster with chicken, while ‘biofilm’ from mammoth and mastodon cluster with elephant. Nor did they explain the internal, or ‘intracellular’ structure we report for observed osteocytes. And finally, they did not state how the rounded structures we reported could persist /_free floating_/ in a hollow biofilm as we described for the ‘vascular’ inclusions in dinosaur vessels. Indeed, it seems that they only addressed aspects of our study that fit conveniently with their preconceived ideas, as they pick and chose what to focus on. As we stated often after our paper came out, morphology alone is insufficient to make any claims about the origin of such material, hence we provided a host of other data to support the hypothesis of endogeneity. Kaye et al. did less than this to support their claim that the material they observed is biofilm.

Kaye et al also overstate their FTIR data, and show a misunderstanding of what these data can be used to say. There are many different molecular vibrations and rotational vibrational modes in a heterogenous sample as represented by their dinosaur material. It is far from a pure sample. In this case, IR absorption peaks overlie each other, and resolution of the spectrum is not high enough to separate them. The spectrum is dependent upon both the composition of th emixture and the relative concentrations as well. Combining all these variables to conclude that one reading is more similar to one than another is really meaningless, especially when trying to interpret peaks in the fingerprint region of 1500-400 cm-1. The low resolution IR spectra figured in this paper is not adequate to draw any concludsions about a heterogenous sample.

Finally, while we have made no claims for the origin of these red, round structures within the vessel structures of our material, the structures we observed did not exhibit the microcryst structure know to characterize framboids, and no where in the literature does it discuss framboids that have a central core of denser material, and of different structure, then the translucent outer regions, as we demonstrate in multiple specimens. We do not, nor have we ever, stated that these are ‘blood cells’. The most we have allowed is that they are not consistent with pyrite framboids, and perhaps the iron we demonstrate in association with these microstructures may be organically derived, a hypothesis we are still testing.

We continue to test the hypothesis that original material is retained in fossil bone, even dinosaur bone, by employing multiple assays and working with experts in many disciplines. We are examining fossils preserved in many different ways, from many time spans, and from a variety of depositional settings, to try to determine factors contributing to preservation. While we welcome the skepticism of colleagues, we hope that the reviewers and readers hold them to the standards to which we are held. Science progresses by scrutiny and testing. We will do our part to be as cautious as possible, and to not overstate our data, while we examine all possible explanations for the observations we make. I guess the best I can say is—stay tuned!

[T. rex Image: David Monniaux, Wikipedia]