Fossil functional traits decoupled from anatomical evolution
Comparing morphology and functional traits of the earliest echinoderms demonstrates that anatomical novelty arose rapidly and evolved in a more volatile fashion, serving as a prerequisite for later ecological innovation. We advocate here for greater attention to compiling fossil life-habits.
Were the earliest diversifications of animal phyla associated more with anatomical exploration or ecological innovation? The longstanding assumption (since Charles Darwin and formalized by George Gaylord Simpson) is that adaptive radiations require ecological opportunity before novel anatomical divergences can evolve. But testing this claim is not easy when both are changing quickly. Teasing apart which one is the primary driver therefore remains an open question. The underlying warrant of the test is a requirement to explicitly document ecological changes in order to identify ecological opportunities (Losos 2010). Our recently published paper in Nature Ecology & Evolution suggests one avenue to conduct such tests.
Like many collaborations, this one started as a “what if?” over beer at a conference. We were attending the Geological Society of America annual meeting in Denver, Colorado. Brad Deline was sharing an update on his goal to build an anatomical data set for all Cambrian and Ordovician echinoderms, his specialty.
To those unfamiliar, echinoderms are weird. Libbie Hyman (1955) famously quipped they were a “noble group especially designed to puzzle the zoologist” for their sheer morphological bizarreness. She was speaking mostly about the living forms. The extinct classes are even more alien (Fig. 1). Some appear to have either a whip-like tail (or perhaps it’s an anterior appendage?) or might be confused with a bobbin of wool. The genus Panidiscus tamiformis was so-named (Sumrall and Zamora 2018) because of its resemblance to a loaf of bread shaped like a Scottish tam o’shanter hat. And there’s also Bizaaroglobus.
Despite their weirdness, they share the virtue of having an excellent fossil record on account of their multiplated, body-encompassing architecture built using geologically stable biominerals. Their fossil record is also very well studied, so much that paleontologists trust that their geological record is generally faithful to their evolutionary history. And many important Cambrian fossils have been found in the past two decades, allowing much greater resolution of their evolutionary relationships.
Compiling a comprehensive data set for 366 genera was quite a task, indeed! Working over many years with several undergraduate students at the University of West Georgia (mostly co-authors), plus getting coding advice from many collaborators, Brad was nearly done building out this enormous database. The end result: 413 anatomical characters, which included not only well-established phylogenetically informative traits but also ancestral (plesiomorphic) traits and species-specific (autapomorphic) traits unique to individual species. The database was built to capture both the broad features that are shared across the phylum, but also the features that made every group unique. Given the goal was to describe the morphological disparity (the variability in anatomical form) of the entire phylum through time, it was critical he include more than the traits phylogeneticists depend on.
Such approaches at documenting morphological disparity through time have become de rigeur, especially among paleontologists. (Although most efforts reuse cladistic matrices rather than intentionally building disparity data sets from the ground up.) But what is so ambitious about Brad’s efforts is his having done so for an entire phylum, especially one as morphologically diverse as echinoderms. Most studies focus instead on a small group of animals over a small portion of time. Brad's data set encompasses essentially all 366 early echinoderms, and over their first two major radiations, the Cambrian and Ordovician radiations, spanning 100 million years.
For years, I have been attempting something similar (Novack-Gottshall 2007). However, instead of focusing on the number and shape of anatomical traits, I have been compiling information on how these same animals moved, interacted with their environment, foraged, reproduced, and other ecological dimensions of their life habits (Fig 2). When Brad told me he now had a comprehensive anatomical data set on the earliest echinoderms, it was natural for us both to wonder whether these two facets of echinoderm biology—their form and their function—evolved in the same manner, or whether one was a more important driver of their tremendous diversifications during the Cambrian and Ordovician radiations.
On top of this, Brad and our colleague Colin Sumrall (University of Tennessee, Knoxville) and his students had compiled a phylogenetic informal supertree that shows how all these animals are related, which allowed for much more robust analyses. The missing piece was a similarly fashioned data set of echinoderm life habits, as my existing database only included life habits for a small fraction of the genera in Brad's data set.
Such functional ecology analyses typically require large numbers of functional (life habit) traits in order to provide sufficient statistical power (Mouillot et al. 2021). Some of my biological peers doubt the fossil record can provide the same kind of data richness that functional ecologists routinely employ. But that is not the case. Like most data collection efforts, all that’s needed is to know where to look, and to then put in the work to carefully compile the data.
Most of the work involves tracking down the systematics literature that originally described the fossils. (For the ecological data set in our paper, we used 180 references. Many of these were also used in compiling anatomical information on the same species.) It should not be surprising that systematics articles provide a wealth of information relevant to the functional ecology of fossil organisms. The systematists describing ancient species are as interested in how these animals lived as they are in their anatomy and evolutionary relationships. It may not be in vogue to care about “natural history” in the current scientific environment, but such meticulous observations and careful descriptions are critical for these types of analyses.
In my case, tracking down functional traits has sometimes required a deep dive into the natural history literature and obtaining journal articles and monographs not published in English. (Shout out to the phenomenal Field Museum Library for providing access to critical journals and monographs.) More than 20% of the references in my library of life-habit literature dates before the invention of PCR, with many important references dating pre-1960 (and that’s surely an undercount if my library were historically comprehensive). And approximately 20% of the genera in our study were originally described in the nineteenth century.
Evidence that can be used to infer life habits takes many forms. The ideal are instances of in situ preservation, when an organism was preserved in life position at the moment of burial and death (Fig. 3). Although many tend to associate this style of preservation with rare Lagerstätten, it is quite common throughout the fossil record. Even when organisms are poorly preserved, it is often possible to find direct evidence of their attachment structures. For motile animals, trace fossils can provide an excellent line of evidence on how they moved, fed, and other behaviors (Hsieh and Plotnick 2020). For echinoderms with extant relatives, we can sometimes make reasonable extrapolations, but this only goes so far when dealing with Cambrian and Ordovician echinoderms because so many classes are now extinct. But it is still possible to draw powerful conclusions by extrapolating knowledge of the functional morphology of anatomically similar living forms.
A critical piece of information I use when inferring life habits is body size. This is most obvious when coding the distance an echinoderm lives above or within the sea floor. But the importance of body size is also critical for other functional traits. In the case of echinoderms, it is even more important. Because suspension-feeding echinoderms do not generate their own feeding currents (i.e., they are passive filter feeders), feeding rates largely depend on the distance the feeding organ is raised into the water column. This requires not only measuring the size of the animal’s body, stem (if raised), and feeding organ, but also making inferences about the manner the body was oriented while alive. Mobility, diet, and many other functional traits are also influenced by body size. The life habits of tiny animals are functionally constrained to be quite different from their larger relatives.
For the current study, we used 19 ecological (functional) characters with 40 states, chosen to represent how these 366 echinoderm genera feed, reproduce, move, and functionally inhabit their environments (Fig. 1). Like for Brad’s data set on anatomy, compiling my ecological data set took several years to achieve, and also involved the efforts of three undergraduates (all co-authors). The end result was two data sets, compiled independently but in a similar manner, and both spanning the same genera.
If we were focused on a single clade of echinoderms, we could have done much better. Selina (Lena) Cole (NMNH) has been compiling detailed information on 28 functional traits for Ordovician crinoids by carefully measuring ecomorphological measurements on individual fossils (Fig. 4) (Cole and Hopkins 2021, Cole et al. 2019). Because her effort is focused on a single (and morphologically and ecologically diverse) class, she is able to code functionally important characters that are relevant across the clade, but would be irrelevant for most non-crinoids. Her approach allows for an ever-more detailed overview of how crinoid ecology has evolved through time, and when her data set encompasses more crinoids, will provide an excellent fine-grained test of our broader-focused analyses. For example, Cole and Hopkins (2021) found strong correlations between morphological and functional (ecological) disparity throughout 160-million years of disparid crinoid evolution, in contrast to our decoupled patterns for echinoderms as a whole. In both cases, we concur that it is not sufficient to assume morphology and ecology are correlated. It is critical evolutionary biologists build comparable data sets to evaluate each facet of history separately.
Armed with two complementary data sets, we were now able to carry out our analyses, which included time-scaling the phylogeny, calculating rates of character change in each data set, documenting trends in disparity (functional diversity), assessment of convergence, among others. (A huge shout out to Graeme Lloyd and David Bapst for advice and suggestions for carrying out R analyses.)
We discovered that morphological evolution in these earliest echinoderms was consistently much faster than their ecological evolution, and that these differences persisted throughout the Cambrian and Ordovician radiations. The initially greater rate of anatomical divergence allowed Cambrian echinoderms to explore a wider range of forms during their outset, while ecological (functional) disparity lagged behind. Similarly, when echinoderm lineages diverged phylogenetically, anatomical evolution involved a much greater number of character changes and a more volatile movement in morphospace. By the Ordovician, however, the slow and ultimately more constrained pace of ecological evolution accumulated in a rich ecological diversity involving novel ecological strategies.
Interesting patterns also emerged when examining convergence. Although instances of major convergence were relatively uncommon (as would make sense during the earliest radiations of any taxon), we found that ecological convergence was much more common than morphological convergence. Similarly, life-habit convergence often played out across rather unrelated taxa (most often belonging to different classes or subphyla), whereas anatomical convergence was most often restricted to echinoderms within the same class.
In short, our analyses provide new evidence that anatomical divergence can be dissociated from functional divergence, at least during the earliest radiations of this phylum. Future tests, such as those ongoing by Lena Cole, will demonstrate the generality of these patterns. However, it is critical that such tests be done using independently collected data sets that are of similar structure and dimensionality, and that can be analyzed in identical manners. We end by encouraging paleontologists to continue compiling these kinds of functionally important and variable-rich ecological data sets. And we want functional ecologists to recognize that the kinds of data they value can be gleaned confidently from the fossil record, even for long-extinct life.
Read the paper here: https://www.nature.com/articles/s41559-021-01656-0. And make sure to check out Lena's accompanying News & Views here.
Cole, S. R., and M. J. Hopkins. 2021. Selectivity and the effect of mass extinctions on disparity and functional ecology. Science Advances 7(19):eabf4072.
Cole, S. R., and D. F. Wright. 2022. Niche evolution and phylogenetic community paleoecology of Late Ordovician crinoids. Cambridge University Press, Cambridge.
Cole, S. R., D. F. Wright, and W. I. Ausich. 2019. Phylogenetic community paleoecology of one of the earliest complex crinoid faunas (Brechin Lagerstätte, Ordovician). Palaeogeography, Palaeoclimatology, Palaeoecology 521:82-98.
Deline, B., J. R. Thompson, N. S. Smith, S. Zamora, I. A. Rahman, S. L. Sheffield, W. I. Ausich, T. W. Kammer, and C. D. Sumrall. 2020. Evolution and development at the origin of a phylum. Current Biology 30(9):1672-1679.
Hsieh, S., and R. E. Plotnick. 2020. The representation of animal behaviour in the fossil record. Animal Behaviour 169:65-80.
Hyman, L. 1955. The Invertebrates. Vol. IV. Echinodermata. McGraw-Hill, New York.
Losos, J. B. 2010. Adaptive radiation, ecological opportunity, and evolutionary determinism. American Naturalist 175(6):623-639.
Mouillot, D., N. Loiseau, M. Grenié, A. C. Algar, M. Allegra, M. W. Cadotte, N. Casajus, P. Denelle, M. Guéguen, A. Maire, B. Maitner, B. J. McGill, M. McLean, N. Mouquet, F. Munoz, W. Thuiller, S. Villéger, C. Violle, and A. Auber. 2021. The dimensionality and structure of species trait spaces. Ecology Letters 24(9):1988-2009
Novack-Gottshall, P. M. 2007. Using a theoretical ecospace to quantify the ecological diversity of Paleozoic and modern marine biotas. Paleobiology 33(2):273-294.
Sumrall, C. D., and S. Zamora. 2018. New Upper Ordovician edrioasteroids from Morocco. Geological Society, London, Special Publications 485:SP485.6.
Zamora, S., B. Deline, J. J. Álvaro, and I. A. Rahman. 2017. The Cambrian Substrate Revolution and the early evolution of attachment in suspension-feeding echinoderms. Earth-Science Reviews 171:478-491.