Methods and Practices in Paleo-Evo-Devo
- 234 Downloads
Paleo-evo-devo is the discipline studying the developmental biology of fossil organisms and its evolutionary implications. In adopting a paleo-evo-devo approach, fossils have to be understood as once-living organisms, and the developmental patterns of extant organisms have to be comparatively investigated. For some types of fossils, it is comparably easy to investigate ontogeny, as they preserve earlier portions of the process throughout their entire life, for example as growth lines, or as they have been fossilized while bearing offspring inside their bodies. Yet, in most cases the ontogeny of fossil organisms (and also of some extant ones) has to be reconstructed based on plausibility. Major aspects for this approach are increasing differentiation or number of structures as well as continuity in development. Despite the difficulties in reconstructing the ontogenies of fossil organisms, studying fossilized development can provide important insights into the evolution of developmental patterns not available only from the study of extant organisms. Also the workflow in the practical work in paleo-evo-devo is shortly outlined.
KeywordsFossilized development Deep time Evolutionary reconstruction Heterochrony Character polarization
Paleo-evo-devo or evolutionary developmental paleobiology is a biological discipline combining approaches from developmental biology and paleobiology into an evolutionary framework. The discipline is still developing; the term summarizing the principal approach – paleo-evo-devo – has to our knowledge first appeared in the foreword to Minelli and Fusco in 2008. While some of the principal approaches are definitely significantly older, the comparably recent appearance of a distinct term for such approaches underlines that the discipline is still maturing. In the following, a short outline of paleo-evo-devo is provided (see also Sánchez-Villagra 2012; Urdy et al. 2013).
Paleo-evo-devo is a research field in which data and knowledge from developmental biology and paleontology are combined to draw conclusions about the evolution of a group of organisms. The inclusion of these data can help to explain how the stepwise transformation of morphological characters evolved, especially in lineages where the extant adults possess very distinctly differing morphologies. With such an approach, a more complete view on the evolutionary history of a group of organisms can be achieved (see also Haug and Haug 2016a).
This chapter is focused on the methodological aspects of paleo-evo-devo to provide a guideline for practical research. The methods which are applied in paleo-evo-devo are significantly different from those used in most other fields of evo-devo, especially due to special challenges of the data acquisition, which makes the explanation of the practical aspects necessary. The theoretical aspects of paleo-evo-devo can be retrieved from other publications (e.g., Hall 2002; Wilson 2011; Urdy et al. 2013).
Fossils as Biological Entities
Fossils are remains of once-living organisms. As such, they offer a view into the past. Reconstructing the extinct organism based on its fossil remains is a truly biological task. Although paleontology, the scientific discipline focusing on fossils, is in many modern university curricula deeply nested in the geological sciences, understanding the biology of an extinct organism obviously demands for biological approaches. This is also true for developmental aspects of the extinct organism. Although it might sound trivial, one needs to be aware that also extinct organisms, like living ones, developed from a single cell, then through embryogenesis, were born or hatched, grew up, and ideally reached adulthood. This statement is in fact far from trivial given its consequences: one needs to recognize that a fossil specimen might not necessarily represent an adult. This also means in further consequence that two morphologically differing specimens do not necessarily represent different species, but could potentially represent different developmental stages of the same species. Making this distinction in a specific case is far from simple and has given rise to significant debates in recent years concerning various groups of organisms (e.g., dinosaurs: Horner and Goodwin 2009; arthropods: Haug et al. 2012; cycloneuralian worms: Haug and Haug 2015a; for a detailed discussion see Haug and Haug 2016a). Hence, careful considerations are required within a biological framework, developmental biology included, when trying to identify and interpret a fossil. Such considerations are therefore some of the basic tasks for paleo-evo-devo.
Reconstructing the Ontogeny of a Fossil Organism
In general, one could assume that paleo-evo-devo faces three major methodological problems: (1) the problem of reconstructing ontogenies; (2) the problem of recognizing conspecifity; and (3) how problems 1 and 2 are to be confronted in the case of fossils. Problems 1 and 2 are highly interlinked and are often part of an iterative process where first an ontogenetic series is reconstructed and only after certain inconsistencies in this series may lead the researcher to conclude that there are in fact two species in the sequence. The specimens included into the ontogenetic sequence are sorted again, and two ontogenetic sequences are reconstructed, which are again checked for consistency (see the example in Haug et al. 2010a, where Cambrian specimens originally described as one crustacean species turned out to be two species). Hence, in the following no distinctions will be made between these problems.
Some organisms are very “cooperative” to the researcher interested in development, preserving aspects of their ontogeny in their morphology, or at least in certain parts of their morphology. As a result, a single specimen may be informative for its entire ontogeny or at least parts of it. Well-known examples are shells and other hard parts of various metazoans. These often show distinct growth lines, which indicate the shape of the shell at an earlier time slice of the ontogeny (Fig. 1). Molluscan shells, for example, often even preserve the embryonic shell (protoconch) and hence provide a comparably complete developmental history of their shell. Such information can be and has been successfully exploited also in fossil organisms (e.g., gastropods: Nützel 2014; ammonites: De Baets et al. 2012; echinoderms: Sumrall 2008).
Serially organized organisms
A related case to preserved ontogeny is provided by still developing, serially organized organisms. In animals that add body segments during their ontogenetic sequence, such as annelid worms or many arthropods, the anterior segments are usually further developed than the posterior ones (e.g., Zhang et al. 2007). With such an anterior-posterior gradient, a single specimen can be (at least roughly) informative of the ontogenetic changes of certain structures. In arthropods, it is for example possible to compare the morphology of trunk appendages along the series and reconstruct the ontogenetic changes based on these segments (see also below, 4d) Increase in differentiation).
If an organism can be bred in captivity, it is indeed possible to directly observe its development. Yet, even in this apparently ideal case it is often necessary to deviate from studying the entire ontogeny of a single individual continuously as certain ontogenetic stages might be missed and most methods for microscopical inspection require preparation of the specimen and thus to sacrifice it. Hence, the ontogenetic sequence of an organism is reconstructed based on a series of individuals of different ages, which is a well-established approach in neontology. Such a strategy can, in principle, be transferred also to fossil organisms, although identifying the conspecifity of all individuals used to reconstruct the sequence may be tricky; relatively complete ontogenetic sequences are needed here.
There are certain exceptions among fossils that allow the identification of conspecifity through aspects of their breeding biology. One example includes gravid animals, where the mother is still carrying the developing embryo. Very well known are ichthyosaur females with embryos in their womb (Fig. 1). Further examples of clear conspecifity in fossil organisms are those in which already hatched immatures are retained in a special cavity of the mother. Such brood pouches occur for example in ostracod crustaceans (Siveter et al. 2007).
PlausibilityIn cases in which neither breeding nor barcoding (i.e., molecular methods to identify which species a specimen belongs to) is successful (or practical), alternative criteria have to be used for identifying conspecifity. In modern forms (e.g., from plankton or meiofauna samples), one is indeed also sometimes forced to apply a chain of reasoning to construct plausible cases of conspecifity; a similar concept can be well applied to fossils (the exact criteria are listed below, a–e). Criteria applicable here are related to the fact that there are certain expectations for ontogenetically differing conspecific forms. Accordingly, it can be partly tested whether several individuals are likely to represent separate stages of a single ontogenetic sequence or are separate species:
Co-occurrence: Co-occurrence is an important datum, as two specimens from the same sample have a higher probability to be conspecific than specimens belonging to samples from different locations. This also applies to fossils: two fossil specimens from the same horizon (namely, a layer of rock with a specific composition) in the same locality are more likely to be conspecific than two fossils from different horizons and/or different localities. As a rule of the thumb, the longer the distance in space and time, the less likely the specimens are to be conspecific. But also here one can find exceptions. Like today, plausibly also in former times there were globally distributed species. For such species, it might easily be possible to find two conspecific specimens on different continents. The same is true for time. A properly formulated and applicable concept for species in time is still lacking, hence even a distance of 20 million years between two specimens may not be enough to discard conspecifity. Therefore, absence of co-occurrence in the fossil record is not sufficient to exclude conspecifity of two specimens, while direct co-occurrence may point to it.
Morphometric aspects: One can assume in general an increase in size throughout ontogeny, but this is not universal. It does, for example, not necessarily account for embryos developing within an egg. Additionally, there are cases of nonfeeding larvae that slightly decrease in size during ontogeny. Furthermore, one needs to accept a certain variance of size also in corresponding developmental stages, which might lead to cases in which supposedly further developed individuals are smaller than less developed ones. Still, with a larger sample size of specimens, it should be possible to test whether all presumed conspecific forms cluster around a single trend line in a scatter plot or not. If they do not, conspecifity cannot be excluded but is more difficult to explain. Even though clustering around the same trend line is not a fully conclusive argument for conspecifity, it makes the case more plausible.
Increase in number of structures: At least for certain organisms, one can expect an increment of certain structures throughout ontogeny. This includes, for example, an increase in number of plates in echinoderms, ribs and whorls in ammonoid molluscs, body segments in annelid worms and many arthropods, or for the latter also number of appendages (e.g., Sumrall and Wray 2007; De Baets et al. 2012; Haug and Haug 2016a; Fig. 2). Such an increase should be (at least roughly) coupled with a gain in size. If there is no (or only a weak) correlation between difference in number of structures and growth, it is unlikely that the individuals represent the ontogenetic sequence of a single species. Rather, it is more likely that they represent different species or different morphs of a species. Yet, also here certain variation in size is possible, leading to slightly smaller individuals, but with a moderately higher number of structures. If there is a positive correlation between increase in size and increase in number of structures, conspecifity is more likely.
Increase in differentiation: One can also assume an increasing differentiation, i.e., the individual structures become better developed throughout ontogeny. This can be recognized partly by a relative size (e.g., when limb buds become functional limbs) or by an increment in number of substructures (e.g., joints on an arthropod appendage; Fig. 2). Such an increase in differentiation should also show a distinct correspondence to growth. Also here a certain degree of variation is to be expected. There are indeed cases where it is also possible that certain structures degenerate during ontogeny (e.g., the horns in pachycephalosaurid dinosaurs, Horner and Goodwin 2009). Yet, also such a decrease should then show a more or less strict correlation to size increase, though a negative one.
Continuity: Considering all correlations discussed above, a certain continuity in development should also be expected, i.e., no repeated change from increase of a structure to decrease of this structure and back to an increase of it. It is thus clear that it is easier to recognize gradual developmental patterns as these provide such continuity. In these cases, the different stages of the studied species will also show more morphological similarities with each other. In forms that develop in a less gradual way, possibly even involving metamorphosis, it will be difficult to link premetamorphic and postmetamorphic forms.
With such considerations in mind, plausible cases can be constructed that indicate possible conspecifity of morphologically differing specimens, identifying those as different life stages. Such approaches can be applied to fossils as well, but remain plausible assumptions. Therefore, a close comparison to modern forms for which ontogenetic sequences are already (at least partly) available is necessary (but see below, Practical Work, (1) Primary data acquisition).
One also has to accept certain limitations. In a hypothetical case of a progenetic (small-sized larva-like) species and its “normal” sister species, it would most likely not be easy to distinguish reliably between a juvenile of the “normal” species and the adult of the progenetic one. Most likely both would be recognized as a single species. Notwithstanding, such a case mainly demonstrates the possible resolution (or its limitations) of paleo-evo-devo approaches instead of disproving their usefulness.
It is also important to note that in many cases it will be easier to reconstruct certain phases of the ontogeny than others. First of all, the embryonic phase is rarely accessible, but also here notable exceptions have been found, such as ichthyosaur embryos inside the mother or isolated embryos of cycloneuralian worms (Fig. 1; e.g., Donoghue et al. 2006; Haug and Haug 2015a). Larval phases of different animals can be reconstructed comparably reliably, but it is less easy to link them to possible postmetamorphic/non-larval juvenile or adult forms. Non-larval juvenile development can again be comparably easily reconstructed although also here pronounced morphological changes may occur complicating the process. Also in many groups of extant organisms, juvenile development is less well documented than the embryonic or larval phase (and the adult morphology), and therefore often less comparative data are available.
The latter discussed cases for reconstructing the ontogeny of extinct organisms all demand for a large sample size. Yet, in many cases only very few specimens might be available. As pointed out above, in some groups of organisms a single specimen can be informative for its developmental history. In cases in which only a single immature fossil is available, the reliability of any interpretation must be seen as even less strong, but also in such cases some inferences might be possible. When restricting attention to a distinct well-delineated systematic group also information of parts of the ontogeny of clearly nonconspecific specimens might be important. In the right framework even the finding of a single specialized immature form may be telling.
Fossils in Evolutionary Reconstructions
Fossils provide minimum ages
The advantage that fossils can deliver information which cannot be inferred from extant forms has even been recognized in a modern molecular-dominated biology: fossils act as anchor points in time. This is also true for developmental aspects. Fossils may give minimum ages for the occurrence of specific developmental patterns or specific larval forms (Fig. 3). In many instances, such occurrences seem to be inferred based on the presence of specific adult morphologies. Yet, larval morphology or more general developmental patterns are not necessarily strictly coupled with a specific adult morphology (e.g., Scholtz 2004, 2005). In other words, there might be modern looking adult forms in a certain horizon, not necessarily meaning that they did already have the same developmental pattern as their modern relatives. And vice versa, specific developmental patterns or larvae might have evolved before the modern-type adult morphologies, for example, in brachyuran crabs (Haug et al. 2015a). Only direct fossil evidence will therefore allow a reliable reconstruction of the time when a specific novelty in a developmental pattern first evolved. For such cases also single (isolated) larval forms may be highly informative as recent descriptions of holometabolous insect larvae have shown (e.g., Nel et al. 2013; Haug et al. 2015b).
Fossils provide important character polarizations
When reconstructing evolutionary changes along a phylogenetic tree, the direction of character change is crucial. If two different character states are known in two sister groups, a comparison to the closest related group of the two, a so-called outgroup comparison, helps to evaluate which of the two states was ancestral (= part of the ground pattern) and which state is derived. Such an evaluation is called character polarization. Often evo-devo scenarios are heavily influenced by model organisms. Yet, model organisms are not necessarily chosen because of their plesiomorphic appearing developmental patterns, but mostly according to their availability. Hence, especially for polarizing detailed character reconstructions along phylogenetic trees, fossils (but also some modern representatives of the “forgotten branches”) may act as important polarization points, providing a clear direction for ordering step-wise character evolution, for example, in early arthropods (e.g., Maas et al. 2006; see also review in Edgecombe 2010).
Fossils may display more plesiomorphic traits
Many fossils indeed show ontogenetic patterns that are also known from modern forms. Such patterns are comparably easy to recognize. Yet, fossils have been heralded for cases in which they possess character combinations that are no longer represented in the modern fauna (e.g., Donoghue et al. 1989). This holds also true for developmental patterns. Fossil species may have developed in a different manner than their modern counterparts (Fig. 3). They might have differed in pattern, i.e., characters appearing in a specific order might have appeared in a different order in the past, or might have developed through quite different larval stages. Such cases are much more difficult to infer, yet they are especially interesting. They again demonstrate that also ontogeny evolves. Identifying such cases is one of the main strengths of the paleo-evo-devo approach.
In some cases, it could be shown that all modern representatives of a group exhibit derived ontogenetic patterns, but that this pattern is not ancestral for the modern group, but that several lineages have evolved it convergently from a now extinct pattern. Such cases are known, for example, from spiny and slipper lobsters (Haug and Haug 2015b, 2016b) and various lineages of insects (e.g., Shear and Kukalová-Peck 1990; Haug et al. 2016a). Such findings are therefore only possible with the inclusion of fossils.
Fossils can break a single “evolutionary jump” into substeps
Often fossils have been termed “missing links” or “connecting links.” Both terms show that there is a general misunderstanding of evolutionary theory and should thus not be used. The fossils often referred to with these inappropriate expressions usually have already evolved some, but not yet all, characteristics of a specific modern group (often these are therefore referred to with the likewise infelicitous term “stem-representative”). As a result, such fossils break down an apparent evolutionary jump, the acquisition of several characters in one step, into at least two substeps. In this way, they provide a clear order in which certain characters evolved (A before B). Such cases can also be found for developmental patterns. Most simple, larval forms that possess only part of the characteristics of their modern counterparts provide an important marker point for determining in which order such larval specializations evolved (Fig. 3).
Heterochrony and the emergence of adult “novelties”While developmental data are mostly informative about the evolution of developmental patterns on first sight, paleo-evo-devo may also be informative for the evolution of modern morphologies or “novelties” in adults. Heterochrony, the evolutionary shift of developmental timing, plays an important role in this respect (Fig. 3). Numerous examples of fossil animal species whose ontogeny has been reconstructed show a rather gradual developmental pattern compared to their modern counterparts; especially well documented examples come from early crustaceans and insects (e.g., Shear and Kukalová-Peck 1990; Walossek 1993; Haug et al. 2016a). This thus often demonstrates a step-wise formation of certain structures that appear in modern forms in a single step. Alternatively, some adult characters in derived forms might be identified as former larval characters that become shifted into the adult phase. For instance, species of the Cambrian arthropod group Naraoia lack dorsal joints of the trunk segments, a feature originally present in larval stages of the closely related trilobites (Fortey and Theron 1994; Haug et al. 2010b). With such examples, also quite aberrant appearing morphologies can be explained by small steps, for example, the evolution of the shovel-shaped antennae in slipper lobsters from elongate feeler-type antennae (Haug et al. 2016b). This again emphasizes the strength and necessity for a paleo-evo-devo approach.
Primary data acquisition
Actual work in paleo-evo-devo involves direct work on fossils. As indicated above in the section “Morphometric aspects”, larger sample sizes, larger sample sizes are of advantage. Yet, when exceptionally preserved fossils are studied, only some few or even a single specimen may be quite informative. For extracting the maximum morphological information from each fossil specimen different up-to-date methods can be applied, for example, computed tomography, virtual surface reconstruction, or contrast enhancing methods in macro- and microphotography (e.g., Haug et al. 2012; Sutton et al. 2014). Maximum information should include also details, which are not necessarily studied for pure taxonomic or stratigraphic reasons.
For comparison, the study of modern forms is necessary. Generally speaking, it is more or less impossible to exhaustively investigate the morphology of modern organisms. Therefore, studies are often restricted to specific aspects of their morphology, usually those that are considered taxonomically important. This is also true for the study of developmental sequences. Unfortunately, these characters are not necessarily the characters that are available in the fossil specimens. This makes in many cases necessary to study the modern species, specifically for the characters available in the fossils. Extensive data acquisition is therefore a crucial step for a paleo-evo-devo approach.
Reconstruction of developmental sequences
Based on the acquired primary data, ontogenetic sequences of the extinct species can be reconstructed. For each of the ontogenetic stages, the morphology of all structures needs to be reconstructed, which makes changes of these structures during ontogeny visible. In general, one expects an increase in size as well as in the number of substructures, such as joints, setae, or spines. However, not all structures usually grow at the same speed in size or number, which results in a mosaic-like development.
Based on the reconstructed morphologies, “local” or proximate interpretations are possible. Among them rough systematic identifications can be provided. Comparison to such forms may then alter some of the morphological or developmental interpretations. Hence, such steps should be followed in a reciprocal manner. Also functional and, based on these, ecological interpretations should be attempted to better understand the entire biology of the studied organism.
Within a phylogenetic framework, it is then possible to make ultimate (evolutionary) interpretations. These may include aspects of time of appearance and/or detailed reconstruction of character transformation. Also based on ecological interpretations, findings may contribute to further-reaching interpretations of faunal and ecology evolution.
We would like to thank curators and collection managers from different museums providing specimens (see figure captions). Furthermore, we are grateful to Roger Frattigiani, Laichingen, for providing the crab megalopa from Solnhofen limestones. JTH was supported by the German Research Foundation (DFG HA 6300/3-1); CH was supported by the LMU through a Bavarian Equal Opportunities Sponsorship (BGF). Both authors would like to thank J.M. Starck, Munich, for his support.
- Fortey RA, Theron JN (1994) A new Ordovician arthropod, Soomaspis, and the agnostid problem. Palaeontology 37:841–861Google Scholar
- Haug JT, Haug C (2015a) Worm Paleo-Evo-Devo – the ontogeny of Ottoia prolifica from the Burgess Shale. Res Rev J Zool Sci 3(1):3–14Google Scholar
- Haug JT, Haug C (2015b) “Crustacea”: comparative aspects of larval development. In: Wanninger A (ed) Evolutionary developmental biology of invertebrates 4: Ecdysozoa II: Crustacea. Springer, Wien, pp 1–37Google Scholar
- Haug JT, Maas A, Waloszek D (2010b) †Henningsmoenicaris scutula, †Sandtorpia vestrogothiensis gen. et sp. nov. and heterochronic events in early crustacean evolution. Earth Environ Sci Trans R Soc Edinb 100:311–350Google Scholar
- Haug JT, Martin JW, Haug C (2015a) A 150-million-year-old crab larva and its implications for the early rise of brachyuran crabs. Nature Comm 6:art.6417Google Scholar
- Haug JT, Labandeira CC, Santiago-Blay JA, Haug C, Brown S (2015b) Life habits, hox genes, and affinities of a 311 million-year-old holometabolan larva. BMC Evol Biol 15:art.208Google Scholar
- Scholtz G (2004) Baupläne versus ground patterns, phyla versus monophyla: aspects of patterns and processes in evolutionary developmental biology. In: Scholtz G (ed) Evolutionary developmental biology of Crustacea. AA Balkema, Lisse, pp 3–16Google Scholar
- Sumrall CD (2008) The origin of Lovén’s law in glyptocystitoid rhombiferans and its bearing on the plate homology and the heterochronic evolution of the hemicosmitid peristomal border. In: Ausich WI, Webster GD (eds) Echinoderm paleobiology. University of Indiana Press, Bloomington, pp 228–241Google Scholar
- Sutton M, Rahman I, Garwood R (2014) Techniques for virtual palaeontology. Wiley-Blackwell, ChichesterGoogle Scholar