Ever since I started having suspicions of the role of olfaction in scavenging behaviours of Tyrannosaurus rex, I have been interested in the olfactory capabilities in predators and how they correlate with their behaviours. So much that I bought a nice book called Predator-prey dynamics: the role of olfaction. However, I shan't write about that book today, even though I quite like it. Instead I shall comment on a paper that became available in Proceedings of the Royal Society B.

I refer to the new study by Darla Zelenitsky and colleagues on the olfactory bulbs in theropod dinosaurs and alligator. Zelenitsky et al. (2009) provide the "first quantitative evaluation of the olfactory acuity in extinct theropod dinosaurs". They calculate relative olfactory bulb sizes (olfactory ratio: the size of the olfactory bulb relative to the size of the cerebral hemisphere) in 21 species of theropods (including Archaeopteryx) and 1 species of crocodilian (Alligator mississippiensis) using the greatest linear dimensions measured from endocasts, CT-scan slices, or from the impressions left on the ventral surfaces of the frontals and parietals. This last bit of information, I thought is very interesting. Although the authors themselves state that there are uncertainties associated with olfactory ratios calculated from the fronto-parietal measurements and thus exclude them from independent contrasts analysis, it demonstrates that we don't necessarily need complete endocasts to extract information regarding that dinosaur's brain, however partial the information may be; it still provides some information regardless. In calculating the olfactory ratios, Zelenitsky et al. (2009) use the greatest dimensions, meaning that their ratios are not restricted to homologous measurements, i.e. depth/depth or height/height, but attempt to capture a purely geometric ratio between maximum measurements; height/depth or depth/height ratios, whichever ones happened to represent the greatest diameter - so that's kind of neat.

Their total data of 29 points are replotted here (see above) as log olfactory ratios against log body mass. A least squares regression (LSR) line is fitted onto 19 of these data points. LSR is conducted on all theropod taxa, so excluding Alligator but also Gorgosaurus and Albertosaurus because the data for these two taxa are derived from the less reliable fronto-parietal measurements. Each species is represented by a single datum, i.e. redundant points are excluded. The LSR line with 95% confidence intervals on this reduced data set is shown superimposed onto a bivariate plot of the total 29 data points.

As you can see, I followed Zelenitsky et al.'s (2009) procedure in data reduction but conducted LSR directly on log transformed data while they performed a LSR on independent contrasts. In other words, their regression analysis is based on data with any phylogenetic signals inherent in the dataset removed. However, my plot and LSR best-fit line above looks identical to their Figure 2, which is labeled "independent contrast least-squares regression", but their actual independent contrasts LSR presented in their supplementary materials show a line forced through the origin so is clearly not the same line as presented in Fig. 2. But the stats presented (slope and r value) in the same fig are for the independent contrast LSR. A minor point, I know...

So regardless of what LSR best-fit line is presented, it is quite clear that the relationship between olfatory ratios and body mass for theropod dinosaurs is probably different from that of Alligator; although as crocodilians are only represented by three members of a single species A. mississippiensis there is no knowing what the crocodilian relationship truly is. However, it is obvious that Alligator has a much higher olfactory ratio compared to those in theropods of similar sizes. Perhaps, the role of olfaction is different in alligators from those in theropods.

Another thing to note is that all theropods including Archaeopteryx seem to fall along the tragectory of the best-fit line. The notable exceptions are tyrannosaurids (excluding Dilong), dromaeosaurids, ornithomimosaurs and Citipati, the former two groups above and the latter two below the line. Zelenitsky et al. (2009) determine if these groups are higher or lower than expected based on the 95% confidence intervals, but I'm not entirely sure if that is a good criterion as these confidence intervals are surely for the slope and intercept and nothing to do with the quantile of data distribution...? Analyses on residuals and other regression diagnostics are used to identify outliers so why not use that? - well, I don't know, who cares...

Anyway, the point of this paper was that, it was the first attempt at quantifying olfaction in extinct theropods, and it shows that aside from Tyrannosaurus rex, which always gets the spotlight for having high olfactory ratios, other large tyrannosaurids but also dromaeosaurs also have relatively high olfactory ratios, and that this is not unique to T. rex. On the other hand, ornithomimosaurs and an oviraptorosaur seem to possess relatively low olfactory ratios. Now, the physiological or ecological interpretations from these results are still unknown, but we could infer that there was something different in the physiology/ecology of these groups of theropods from an "average" or primitive theropod condition.

Reference:

Zelenitsky, D. K., Therrien, F., and Kobayashi, Y. 2009. Olfactory acuity in theropods: palaeobiological and evolutionary implications. Proc. R. Soc. B. 276: 667-673.

I refer to the new study by Darla Zelenitsky and colleagues on the olfactory bulbs in theropod dinosaurs and alligator. Zelenitsky et al. (2009) provide the "first quantitative evaluation of the olfactory acuity in extinct theropod dinosaurs". They calculate relative olfactory bulb sizes (olfactory ratio: the size of the olfactory bulb relative to the size of the cerebral hemisphere) in 21 species of theropods (including Archaeopteryx) and 1 species of crocodilian (Alligator mississippiensis) using the greatest linear dimensions measured from endocasts, CT-scan slices, or from the impressions left on the ventral surfaces of the frontals and parietals. This last bit of information, I thought is very interesting. Although the authors themselves state that there are uncertainties associated with olfactory ratios calculated from the fronto-parietal measurements and thus exclude them from independent contrasts analysis, it demonstrates that we don't necessarily need complete endocasts to extract information regarding that dinosaur's brain, however partial the information may be; it still provides some information regardless. In calculating the olfactory ratios, Zelenitsky et al. (2009) use the greatest dimensions, meaning that their ratios are not restricted to homologous measurements, i.e. depth/depth or height/height, but attempt to capture a purely geometric ratio between maximum measurements; height/depth or depth/height ratios, whichever ones happened to represent the greatest diameter - so that's kind of neat.

Their total data of 29 points are replotted here (see above) as log olfactory ratios against log body mass. A least squares regression (LSR) line is fitted onto 19 of these data points. LSR is conducted on all theropod taxa, so excluding Alligator but also Gorgosaurus and Albertosaurus because the data for these two taxa are derived from the less reliable fronto-parietal measurements. Each species is represented by a single datum, i.e. redundant points are excluded. The LSR line with 95% confidence intervals on this reduced data set is shown superimposed onto a bivariate plot of the total 29 data points.

As you can see, I followed Zelenitsky et al.'s (2009) procedure in data reduction but conducted LSR directly on log transformed data while they performed a LSR on independent contrasts. In other words, their regression analysis is based on data with any phylogenetic signals inherent in the dataset removed. However, my plot and LSR best-fit line above looks identical to their Figure 2, which is labeled "independent contrast least-squares regression", but their actual independent contrasts LSR presented in their supplementary materials show a line forced through the origin so is clearly not the same line as presented in Fig. 2. But the stats presented (slope and r value) in the same fig are for the independent contrast LSR. A minor point, I know...

So regardless of what LSR best-fit line is presented, it is quite clear that the relationship between olfatory ratios and body mass for theropod dinosaurs is probably different from that of Alligator; although as crocodilians are only represented by three members of a single species A. mississippiensis there is no knowing what the crocodilian relationship truly is. However, it is obvious that Alligator has a much higher olfactory ratio compared to those in theropods of similar sizes. Perhaps, the role of olfaction is different in alligators from those in theropods.

Another thing to note is that all theropods including Archaeopteryx seem to fall along the tragectory of the best-fit line. The notable exceptions are tyrannosaurids (excluding Dilong), dromaeosaurids, ornithomimosaurs and Citipati, the former two groups above and the latter two below the line. Zelenitsky et al. (2009) determine if these groups are higher or lower than expected based on the 95% confidence intervals, but I'm not entirely sure if that is a good criterion as these confidence intervals are surely for the slope and intercept and nothing to do with the quantile of data distribution...? Analyses on residuals and other regression diagnostics are used to identify outliers so why not use that? - well, I don't know, who cares...

Anyway, the point of this paper was that, it was the first attempt at quantifying olfaction in extinct theropods, and it shows that aside from Tyrannosaurus rex, which always gets the spotlight for having high olfactory ratios, other large tyrannosaurids but also dromaeosaurs also have relatively high olfactory ratios, and that this is not unique to T. rex. On the other hand, ornithomimosaurs and an oviraptorosaur seem to possess relatively low olfactory ratios. Now, the physiological or ecological interpretations from these results are still unknown, but we could infer that there was something different in the physiology/ecology of these groups of theropods from an "average" or primitive theropod condition.

Reference:

Zelenitsky, D. K., Therrien, F., and Kobayashi, Y. 2009. Olfactory acuity in theropods: palaeobiological and evolutionary implications. Proc. R. Soc. B. 276: 667-673.

## Comments

Archaeopteryxhad the hearing capacity of an emu (which can't hear very well).Modern crocs, being so often submerged, have doused their sense of smell in favor of better eyesight and hearing abilities. I'm actually surprised how well they did, considering their environment.

Archaeopteryxcomes out as having olfactory ratios typical for a theropod of that body size. Troodontids also are typical while dromaeosaurids are higher. So the ancestral condition for these three taxa should be typical or close to typical as well. Primitive theropods like ceratosaurs and allosauroids also possess "typical" olfactory ratios. So, olfaction didn't get worse towards the birds but a primitive condition was retained throughout theropod evolution with certain outlying groups as the exceptions. High olfactory ratios in the Tyrannosauridae can be considered a derived condition, because the basal tyrannosauroidDilonghas a typical olfactory ratio, so the ancestral condition of that clade is probably "typical".We don't know what the primitive olfactory ratio is for archosaurs because that was not within the scope of this paper. For all you know, crocs may show some derived condition of heightened olfactory ratios.

So you're totally right--crocs might be doing their own totally derived thing! :-)

Archaeopteryxas having the worst sense of smell.On the other hand, what strikes me now that I look back at it, is that this supposedly dimensionless number still has some significant correlation with body size. In general, the bigger the theropod, the higher the olfactory ratio. That's rather interesting.

beingbigger?That is quite interesting - that would mean theropod brain shape show

allometric scaling, i.e. different parts increase in size with different

proportions. And olfactory bulb size seems to be proportionally larger in larger theropods. So, maybe as you say, it is nothing more than some ontogenetic/phylogenetic development in relation to size.

>Raptor Lewis

You'd need to test for correlations with stratigraphy to say anything

conclusive, but dromaeosaurs, ornithomimosaurs and oviraptorosaurs are

also from the Cretaceous and they don't have extremely high olfactory ratios.

Well, there is a correlation between sense of smell (represented by olfactory ratio) and body size. You can see this by simple visual inspection of the scatter plot; all points roughly follow a certain tragectory (olfactory ratio is almost consistently higher in larger theropods). This trend may not necessarily be a linear one, but a best-fit linear regression line explains almost 90% of the total variance in the data - so the goodness of fit of this particular linear model is pretty high. So we don't know if this particular model really explains the relationship between size and sense of smell, but we can say that larger theropods generally have better sense of smell, i.e. higher olfactory ratios.

cluster 1: larger theropods which show some relationship between body mass and olfactory ratio

cluster 2: alligators with to few data points to tell whether there is a relationship

cluster 3: small-medium size theropods which display the same olfactory ratio independent from size

belonging to no cluster:

ArchaeopteryxEven if I assume that I can derive a single model for all data points, there is the bias which lies in the the involvement of

Archaeopteryx. The problem is that theArchaeopteryxdata point is far away - at least one order of magnitude on both axes - from all other data points.So it has a huge impact on the whole model (and no wonder it is fitting well). If you would do the fit leaving out a single data point alternately, like in a cross-validation test, you will find that consideration or non-consideration of

Archaeopteryxwould turn out to have the largest effect.I suppose that a linear relationship (of the log-log data) would be no longer a good hypothesis if you would in include

AlbertosaurusandGorgosaurusand leave outArchaeopteryx. For the model you depicted the residuals look already quite suspicuous.Archaeopteryxis an influential outlier with regards to this dataset. But who knows though. Data from other small theropods such asCompsognathus,Microraptor,Mei, etc. are lacking, the inclusion of which in a future study (if at all possible from squished fossils) may or may not alter that...Alligator on the other hand is not included in the LSR analysis and is just superimposed on the theropod analysis for reference.

Alternative explanation: What if the olfactory ratio ("the ratio of the greatest diameter of the olfactory bulb to the greatest diameter of the cerebral hemisphere") is rather constant in all theropods up to a certain body size and becomes more variant in the largest species.

This could be because the size of the olfactory bulb is more dependent on body size than the brain size (which is not increasing proportionally in the largest species).

I think a better plot to demonstrate what they wanted to demonstrate would have been:

Plot the ratio of olfactor bulb diameter and (logarithmized) body mass

against the ratio of cerebral diameter and (logarithmized) body mass.

You would see a better split-up of good and bad sniffers on the diagonal (the other diagonal would represent body size).