A research program for Evolutionary Morphology

Background

For most biologists, morphology is restricted to the description of organs, organ systems and organisms. The term morphology is even often used synonymously to the object which the discipline investigates (as in the term morphological data). Although morphology does indeed proceed from description, it is not restricted to it. Riedl (2000) defined morphology as the ‘science of the gestalt, or more precisely of its decryption. It thus has an epistemological aspect and includes the alternation of analytical and synthetic processes' (our translation). To put it differently, morphology is the science of form, the study of how and why organisms look the way they do, of why certain parts of an organism possess certain predicates, but also of why not all the potentially possible combinations of elements exist and why it might not even be possible for them to exist. Morphology is the discipline which investigates the causes of the disparity of life, which therefore turns it into a science of explanation (contrary to the excessively restrictive definition put forward by Ghiselin 1980, 2006). Post-Darwinian morphology is always an evolutionary morphology even if in certain aspects of functional morphology and biomechanics evolutionary questions only play a subordinate role.

Although this essay is not intended to provide an overview of the history of morphology, it is enlightening to see that morphology has been an explanatory science since it began with Goethe and Owen (see Rieppel 2011, 2012, 2013; and Richter and Wirkner 2013 for certain aspects of the history of morphology).

A view of morphology as general as the one put forward herein has received repeated but isolated support over the past decades. Advocates include Weber (1958) and Seilacher (1970) in their conceptual publications on ‘constructional morphology’ (Konstruktionsmorphologie), Dullemeijer (1974) in ‘Concepts and Approaches in Animal Morphology’ and Riedl in his fundamental works ‘Order in Living Systems’ (Riedl 1975, 1978) and ‘Structures of Complexity’ (Riedl 2000). Wake (1982) outlined in a short essay the tasks of ‘functional and evolutionary morphology’, and Bock (1994) described ‘concepts and methods in ecomorphology’. After a long period of absence from mainstream evolutionary biology, morphology experienced a renaissance within the field of Evolutionary Developmental Biology, currently one of the most active areas of evolutionary biology. Olsson et al. (2006) described evolutionary morphology (as a historic phenomenon) as the direct precursor of Evolutionary Developmental Biology (evo-devo). Love (2006) highlighted the continuing importance of morphology alongside evo-devo, in particular by identifying innovations and novelties. Scholtz (2010, 2013) emphasizes the importance of morphology independently of evo-devo. Other proponents of evo-devo, however, fail to attribute any particular role to morphology in an extended evolutionary synthesis (e.g. Pigliucci and Müller 2010).

A reappraisal of morphological data (though, it is better to speak of phenotypic data; Rieppel and Kearney 2002) also occurred in the field of phylogenetic systematics, though here in the light of the growing dominance of molecular phylogenetic analyses. In this context, the importance of a proper critical evaluation of phenotypic data was emphasized (Jenner 2004). This has been followed by a new interest in morphology as a discipline (e.g. Sudhaus 2007; Müller and Rosenberg 2009). Some recent studies focus on the problem of the lack of uniform terminology and data standards in morphological analyses (the so-called linguistic problem; Vogt 2008; Vogt et al. 2010, 2011). The development of glossaries (e.g. Richter et al. 2010), and particularly of ontologies (Vogt et al. 2010, 2011 for theoretical background) is intended to resolve this problem (see also Mabee et al. 2007, Deans et al. 2012). We too have attempted elsewhere to summarize certain aspects of what we elaborate on here, including in a short essay in a volume published by the German Zoological Society on occasion of the 100th yearly meeting of the society (Richter 2007) and in a short overview of the descriptive and comparative aspects of Evolutionary Morphology (Wirkner and Richter 2010). Some parts of this essay have also been published previously in German (Richter and Wirkner 2013).

We attempt herein to (re-)conceptualize the discipline of Evolutionary Morphology. We use the term Evolutionary Morphology (rather than just morphology) simply to make it easier for the reader to separate our concept from the historical approaches, although ours does incorporate many elements from earlier accounts of morphology. Well aware of the shortcomings and imprecision of our proposal, our main aim is to open the discussion of Evolutionary Morphology as an explanatory science in the hope that our concept can eventually be refined and specified. What is equally important, however, is that the applicability of the concept be tested on a wide range of case studies (many of which already exist).

Evolutionary Morphology is an independent scientific discipline in its own right (comparable to mathematics which is the foundation of but not subsumable to chemistry and physics). Evolutionary Morphology includes various ‘areas of research’ (the term subdisciplines appears too formal) such as descriptive morphology, in which organisms and their parts are described; functional morphology, which includes functional analyses and biomechanics; comparative morphology, where homologies are identified and where the relative chronological order of transformations of evolutionary units are determined (by using phylogenetic analyses); and finally, causal morphology. Causal morphology, in turn, can be broken down into two areas, one focusing on the phenomenon of adaptation, the other on potential limits to the evolvability of certain structures that only exist in combination with others (the area we have termed ‘coherence morphology’, Kohärenzmorphologie, Richter 2007). However, none of the ‘areas of research’ should be considered separate approaches, for it is only by applying them together that we can begin to explain the disparity of life. Phylogenetic Systematics and Evolutionary Developmental Biology are mature disciplines in their own right but also have an explicit role to play in Evolutionary Morphology.

The tasks of Evolutionary Morphology

Evolutionary Morphology comprises the description, comparison and explanation of predicates of phenotypic objects (parts of an organism). These objects are conceptualized differently at different levels: as morphemes (see Richter and Wirkner 2013) in the area of descriptive morphology, as functional units in the area of functional morphology and as evolutionary units in the areas of comparative and causal morphology. Although all these terms refer to real, sometimes identical objects, they identify them from different perspectives. Descriptive, functional, comparative and causal morphologies are areas with specific aims and roles in Evolutionary Morphology, each asking different questions and each complementing the other areas. This will be outlined in the following.

Descriptive morphology

Descriptions in morphology refer primarily to single organisms or even semaphoronts (exemplar approach sensu Prendini 2001); the study of more than one organism is undertaken to test the potential for generalization of these descriptions (primarily on species level) and to exclude artefacts. This does not imply a failure to identify variation, however. On the contrary, it is by comparing the descriptions of different organisms that variation is revealed. In general, all organisms are integrated entities and as such, all (or at least most) attempts to divide them into parts are going to result in subdivisions that are in some way artificial (Dullemeijer 1974; Young 1993; Vogt et al. 2012). While certain parts of organisms do constitute recognizably separate entities in many cases (e.g. the bones in the vertebrate skeleton, particularly if separated as in fossils), the distinction between elements in other cases appears to be imposed by us. The vertebrate heart, for example, is part of an entirely closed vascular system and directly connected to veins and arteries (though a phenotypic separation might be possible using the heart valves as a marker). The heart is also a good example of how a functional definition – pumping organ (sensu Toepfer 2005) – might come into play. Equally possible are definitions based on development (the common anlage before differentiation) and on evolution (e.g. the contrast to the absence of a heart in Branchiostoma, though this would imply that only the part anterior to the sinus venosus belongs to the heart). It becomes obvious, yet, that any definition based on function, development or evolution already goes beyond the remit of mere description. Vogt et al. (2012) recently discussed this problem from the angle of bona fide versus fiat boundaries of objects. In their view, the demarcation of bona fide objects (i.e. mind-independent objects) needs to take into account functional aspects. We hesitate to accept this because of the implication that bona fide objects would not be identifiable as such were their function not known.

We suggest that the subdivision and definition of the parts of organisms should be exclusively phenotypic, avoiding any reference to function or evolution (see also Dullemeijer 1974; Young 1993; Scholtz 2010). Young (1993) identified four such classes of objects, all of which can be used at various levels of description: (1) disjointed, intrinsically identified: separate elements that are, additionally, intrinsically identifiable, for example on the basis of their cellular composition, for example a protonephridium in a body cavity, (2) disjointed, extrinsically identified: identically composed yet clearly separate elements, for example the bones of the human hand, (3) adjoining, intrinsically identified: generally speaking homogenous objects which are intrinsically separated by their composition (e.g. cellular structure) – examples being the mesodermal midgut and ectodermal proctodaeum in insects and (4) adjoining, extrinsically identified: homogenous parts of an organism which are indistinguishable on the cellular or ultrastructural level and where identification is based mostly on position in comparison with other elements of the same system, for example most parts of the nervous system and the vascular system in arthropods (e.g. Fritsch and Richter 2010; Wirkner et al. 2013). In most cases (but not all), these distinctions depend on the level of magnification and differ, for example, according to whether the observer is using the naked eye, light microscopy or electron microscopy. In other words, they are scale-dependent, a phenomenon well known to morphologists (see also Vogt et al. 2012). A particularly interesting question – and one yet to be answered – is the extent to which the delimitation of these elements (this might be particularly important with regard to those only extrinsically identified) is influenced by the human ability to perceive and recognize forms (‘Gestaltwahrnehmung’, Lorenz 1943). This, however, is not the focus of the present essay.

One particular problem is the demarcation of parts of an organism in time, which is a problem of fuzzy sets. Ultimately, all parts of an organism possess precursors or anlagen, which in turn also have precursors. For practical reasons, units of description should be stable (identifiable) for a certain amount of time (McShea and Venit 2001). Where intrinsic predicates exist, they can be used for the purposes of demarcation. Where any kind of concentration of neurites is identifiable, for example, one can talk about ganglia or ganglia anlagen (Richter et al. 2010).

The next question is what term should be used for the parts of an organism that constitute the objects of description (terms such as ‘organ system’ and ‘organ’ have functional implications and, additionally, a significant historical burden). The term ‘character’ is often used, referring as it does to the fact that the objects in question are characterized by something. However, ‘character’ has multiple meanings (see, for example, contributions in Wagner 2001) and in the context of phylogenetics is mainly used to signify an evolutionary unit (the difference between character and character state will be discussed below). We suggest that this term should be avoided at the descriptive level of morphology. A neat term would be ‘structure’, but unfortunately ‘structure’ also has another meaning – composition, which means there is no structure without a structure. For this reason, we suggest that this term be avoided too (see also Dullemeijer 1974; Young 1993). The term ‘feature’ (Bock and von Wahlert 1965) is too often used in a general way to be of any great use for our purpose. Relatively neutral are the terms ‘morphological element’ (Dullemeijer 1974; Young 1993) and ‘morphological trait’ (Vogt 2008), but neither are defined specifically enough. Because we are interested in clear terminology and the clear demarcation of terms, we suggested a new term for the parts of an organism that are the objects of morphological description: ‘morpheme’, according to its use in linguistics (see Richter and Wirkner 2013). While morphemes in linguistics are restricted to the smallest units of meaning (e.g. Römer 2006), morphemes in morphology exist on different hierarchical levels. The heart (of a vertebrate) might be a morpheme, but ventricle and atrium are also morphemes, and even within the ventricle and atrium morphemes might be identifiable. Eventually, we would reach the cellular and subcellular level, and here too the nucleus of any heart cell might be identified as a morpheme. Morphemes are scale-dependent in the same way as description is scale-dependent. This does not imply that a single publication will only deal with one level of description; on the contrary, many studies start with an overview and then go on to provide details on various descriptive levels (often depending on the techniques used, for example MicroCT, histology, transmission electron microscopy). Objects might reasonably be identified as morphemes when they represent the smallest units at a particular level of description.

Form and function of morphemes

In practice, descriptions will be shaped by the question which the study sets out to investigate. A description (which always needs to include a written part and cannot consist only of drawings or photographs, see Vogt et al. 2010) with no question behind it is certainly possible but might not be particularly useful. Every describable property of a morpheme is referred to here as ‘form’ (Bock and von Wahlert 1965; Dullemeijer 1974; Scholtz 2010). ‘Form’ does not just mean ‘shape’ (e.g. round, triangular, globular) but also includes ‘composition’ (or structure in the strict sense), especially on the level of tissues or cells. Consequently, ‘material’ (e.g. calcite shells in molluscs or foraminifera, cartilage, bone) also belongs to the ‘form’ of a morpheme. In summary, all non-process-related properties belong to the form of a morpheme.

Initially, functional (i.e. process-related) properties such as strength, extensibility and elasticity seem to fit logically into the concept of ‘form’, but they first need to be identified via a functional analysis. Function, then, is directly derivable from form (Dullemeijer 1974), which implies that there is no morpheme without function (sensu Dullemeijer 1974). On the other hand, several morphemes together can form functional units (see below).

Terminology

In recent years, the need for a uniform and standardized terminology in morphology has become clear (Ramirez et al. 2007; Edgecombe 2008; Vogt 2008; Sereno 2009). Vogt et al. (2010) even consider the absence of standards in terminology (the linguistic problem) to be one of the main problems in morphology (It should be mentioned that their view of ‘morphology’ is less inclusive than our concept of Evolutionary Morphology). We argue that while morphological description proceeds from single organisms, terms clearly need to be transferable, that is referable not just to a morpheme of one specific organism but to all comparable ones. It must be possible to give morphemes which display a very high degree of correspondence across organisms the same name. On the other hand, it might be more useful still to define terms in glossaries and ontologies in such a way that one and the same term can be used to refer even to morphemes with a less high degree of correspondence. Richter et al. (2010), for example, apply the term ‘brain’ to all bilaterians, despite the very obvious differences that are found. Because we agree with Vogt (2008) and Vogt et al. (2010) that a uniform morphological terminology should be free of assumptions about homology, terminology should primarily refer to morphemes. What makes this so challenging is that whether or not a term is used purely descriptively or to imply a potential homology often depends on the background situation. Calling the most proximal podomere in spiders a ‘coxa’ in a taxonomic work implies nothing about its homology to the coxa in crustaceans and insects. On the other hand, when Walossek (1999) avoids the term for chelicerates and restricts it to crustaceans, a very specific hypothesis of the evolution of arthropod limbs is implied. A homology-free terminology should only use a formulation like ‘most proximal podomere’ (for the sake of the argument, the term ‘podomere’ has no further implications here) rather than coxa or basipodit if the description in question deals with various different arthropods.

Technical advances in the area of descriptive morphology

In the last few decades, major progress has been made in the area of descriptive morphology. Three-dimensional reconstruction in particular has completely changed our view. Although it was possible in the past to three-dimensionally reconstruct serial sections and document internal anatomy using wax models such as those produced by the Freiburg manufacturer Ziegler (Hopwood 2002), computer-based 3D reconstruction is now a standard technique in descriptive morphology (Wirkner and Richter 2010). Immunohistochemistry, which permits great specificity in the documentation of morphemes, confocal laser scanning microscopy (cLSM; e.g. Hessling and Westheide 2002; Rothe and Schmidt-Rhaesa 2010), micro-computer tomography (micro-CT; e.g. Wirkner and Richter 2004) and magnet-resonance tomography (MRT; Ziegler et al. 2012) are equally important tools in the study of internal anatomy today. In all of these techniques, it is the three-dimensionality of the documentation and the opportunity to actually manipulate these 3D reconstructions that have revolutionized our understanding of form (Murienne et al. 2008). Another front on which great progress has been made is that of video microscopy (Weiss et al. 1989), thanks to which we now view the cell as a ‘living’, not static, entity. Continuing from this, 4D microscopy now permits the study of developmental processes such as cleavage (Schnabel et al. 1997; Hejnol and Schnabel 2006) and enables us to differentiate morphemes in time.

Comparative morphology as homology research

We now leave the field of descriptive morphology and move on to the field of comparative morphology. Although comparisons might include other aspects as well, we restrict the area of comparative morphology to questions concerning homology. This is where the evolutionary aspect comes into play, because of the ‘hidden bond of common descent’ (Darwin 1859) which connects all organisms and species. Homology, which is (to put it simply) ultimately ‘sameness’ caused by common descent, exists between parts of organisms and cannot be applied to entire organisms.

Although homologues and morphemes might refer to the same objects (though this is not necessarily the case), the terms represent different ways of looking at these objects. Morphemes are units of description without reference to evolution, while homologues have, under the majority of homology concepts, evolutionary implications (although we are not arguing here against the concept of serial homology).

There can be very few terms (if any) that remain as controversial today as homology. The core of the controversy has persisted unchanged since Owen and Darwin. It centres on how to incorporate into the concept of homology an additional, mechanistic, biological explanation for ‘robust pattern transmission’ (sensu Szucsich and Wirkner 2007) that can exist alongside the historical explanation of descent with modification (Laubichler 2000). Whatever one's view, it is very important to distinguish between definitions of homology on the one hand, and ‘criteria’ that allow us to hypothesize homologies between objects (e.g. Bock 1989; Dohle 1989; Young 1993) on the other. We support the view that homology can only exist if correspondences between homologues are indeed present (see e.g. Richter 2005 and Szucsich and Wirkner 2007).

Distinguishing evolutionary units from morphemes

As mentioned above, morphemes and evolutionary units are not the same. Evolutionary units are often termed ‘characters’, but this is unfortunate because of the various meanings of the term character. Moreover, we think that the concept of ‘evolutionary unit’ can be understood intuitively. We follow Lewontin (1978) in viewing it as a ‘quasi-independent unit of evolution’ and agree with Wagner (2001) that evolutionary units (his characters), if properly defined and delineated, are the real subjects of adaptations.

An evolutionary unit might correspond to a morpheme (the same object can be described as a morpheme and conceptualized as an evolutionary unit), but an evolutionary unit might also correspond to a group of morphemes (e.g. those forming a functional or biological role unit, see below) or only to part of a morpheme (if the morpheme belongs to a higher level of description). Evolutionary units have to be identified or conceptualized (Wirkner and Richter 2010). The conceptualization of evolutionary units, however, needs to be a multistep process, as a priori conceptualization would require knowledge of the evolutionary course. Delineation of an evolutionary unit, therefore, already represents a hypothesis.

There are some cases in which it appears obvious that morphemes cannot constitute evolutionary units, one of them, for example, being phenotypic plasticity (the capacity of a genotype or individual to express different phenotypes across a range of environments, see Moczek et al. 2014). In the circulatory system of the marbled crayfish (Procambarus fallax f. virginalis; an exclusively parthenogenetically reproducing crayfish where all organisms are genetically identical; Martin et al. 2007), differences in the branching pattern of arteries exist which are obviously not genetically determined and are therefore not inherited (G. Vogt et al. 2009). Nevertheless, it is easily possible to identify and describe these arteries and their branching patterns as morphemes. In the case of the marbled crayfish, then, the conceptualization of evolutionary units must clearly be restricted to those patterns which are constant (S. Scholz, S. Richter, C.S. Wirkner, unpublished data). Another, slightly different example is provided by the bowfin Amia calva. Here, it is the shape of the nasal and frontal bones of the cranium which differs. All four (sometimes fewer because of fusion) bones together cover the same area, but the area covered by each bone in isolation (and therefore the shape of each bone) varies between individuals (Grande and Bemis 1998; see also Schwenk 2001). This implies that the bones in isolation cannot be considered single evolutionary units, although they can certainly be described as single morphemes, each with a particular shape. The evolutionary unit would probably need to take in all four bones together. Bilateral symmetry is another field in which morphemes and evolutionary units are not the same. If no asymmetry is obvious (but see below), single evolutionary units are to be conceptualized which take in both the left and the right side together, for example forelimbs or hindlimbs in tetrapods (and these, notably, are not even identical morphemes because they are mirror symmetrical). Interestingly enough, many descriptions actually omit one side of the organism altogether (in particular in documentation, e.g. in drawings).

The difference between morphemes and evolutionary units can be illustrated in another, more complex example (see Richter and Scholtz 2001 and Poore 2005 who considered this problem in different ways in real data matrices). Amphipod crustaceans possess 14 pereiopods (all the thoracic limbs except the first ones, the maxillipeds), each of which represents a morpheme at the relevant level of description. One important aspect of amphipod pereiopods is that they are uniramous, that is the exopod is absent (when compared with the biramous pereiopods found in most other peracaridan crustaceans; Gruner 1993). The question we ask is whether or not all 14 of these limbs (focusing for the sake of argument on the fact that they are uniramous, although many other aspects of their morphology could also be considered) constitute separate evolutionary units. The first objection might stem from the fact that the pereiopods (at least in terms of all being uniramous) are bilaterally symmetrical. This probably implies that we are only looking at seven evolutionary units (as the left and right pereiopods evolved together). The pairs of pereiopods, however, can in turn be divided into three functional units: the first two pairs are bent forward, carried under the head and not used for locomotion, the third and fourth pairs are used for locomotion, and pereiopods six to eight are bent backwards and used to grip the substrate (Gruner 1993). Do the pereiopods only constitute three evolutionary units, then, corresponding to the functional units? This assumption certainly has some merit because the biological role which would lead to adaptation is fulfilled by the functional units and not by single pairs of pereiopods (see section on the role of functional morphology). To complicate matters, Wolff and Scholtz (2008) were able to show that only one developmental mechanism is responsible for the non-development of exopods and thus for uniramous limbs. The single developmental mechanism could imply that all 14 pereiopods represent a single evolutionary unit (with reference to the absence of exopods). What this hypothesis fails to take account of, yet, is that the mechanism may have evolved in a single pair of limbs and then been transmitted to the others in the course of evolution. This shows clearly that evolutionary units must be conceptualized, and that morphemes are not equal to evolutionary units.

In those cases where ambiguity makes a priori conceptualization unreliable, we suggest taking as evolutionary units the smallest individualizable units that might possess their own evolutionary history (this could imply that evolutionary units are, like morphemes, scale-dependent, which partly contradicts the restrictive definition put forward by Grant and Kluge 2004).

Once we start identifying evolutionary units in different taxa, we also have to name them. In this, we cannot avoid using assumptions about homology (see also Edgecombe 2008). Vogt et al. (2010) suggest adding a taxon attribute, for example ‘bilaterian protonephridium’, and we agree that in many cases this might help to clarify what is meant. Richter et al. (2010) came up with a proposal for terms for morphemes in neuroanatomy which are intended to be strictly descriptive and do not suggest homology between taxa. Examples are ‘brain’ and ‘tetraneurion’. Although there can be little doubt that the ‘gastrotrichan brain’ and the ‘entoproctan tetraneurion’ might represent evolutionary units (or a combination of evolutionary units), this approach throws up more difficult (and more important) questions such as whether or not an ‘entoproctan-molluscan tetraneuron’ can also be conceptualized as an evolutionary unit (see Wanninger 2009 for the biological background). The use of a terminology which is separate from that of morphemes might therefore be another solution. This point was touched on above on the example of the term ‘coxa’, which in our view should be restricted to a specific evolutionary hypothesis (see above and Walossek 1999 for background). In general, we believe that this aspect of the ‘linguistic problem’ (sensu Vogt et al. 2010) is far from being resolved.

Character conceptualization

The term ‘character conceptualization’ previously used by us (Wirkner and Richter 2010) refers to both the conceptualization of character states and the conceptualization of characters, that is the assembling of a character matrix. In cladistic terminology, ‘character’ often refers to a kind of ‘superordinate structure’ (Rahmenhomologie sensu Riedl 1975) often based on a topographical identity (Rieppel 1988; Rieppel and Kearney 2002), and ‘character states’ are subordinate to this (e.g. Freudenstein 2005; Brower and Schawaroch 1996; Sereno 2007). This, however, is different from the original concept introduced by Hennig (1966) and later called an ‘ideographic character concept’ (Grant and Kluge 2004). Hennig's ‘character states’ (or characters because Hennig himself was not very exact in his use of these terms) are part of ‘transformation series’. Transformation series do not exist in a single organism but constitute a combination of historical events (i.e. transformations). This is particularly clear in the case of ‘neomorphic characters’, whose two states ‘absent’ and ‘present’ cannot exist simultaneously in the same organism. (Sereno 2007 draws a distinction between this kind of character and the ‘transformational character’).

Grant and Kluge (2004) define ‘character states’ as ‘the least inclusive historical individuals that result from heritable transformation events’ – in other words, as elements of transformations series, that is the evolutionary units in our terminology. It is in the operational process of preparing a character matrix that these evolutionary units become ‘character states’ and are organized into ‘characters’ (=transformation series; Richter 2005). Theoretically speaking, we would actually prefer the terms ‘character’ (for character state) and ‘transformation series’ (for character), believing them to be more intuitive, but we are aware that ‘character states’ and ‘characters’ are well established terms and therefore use them here in the generally accepted manner.

Once evolutionary units have been identified in one organism, identical (or rather comparable) evolutionary units have to be identified in others (and both scored as the same character state). Ultimately, the criteria involved are those introduced by Remane (1952), although they have been rendered more precise and reformulated various times over the years (e.g. Riedl 1975; Dohle 1989; Rieppel and Kearney 2002; Richter 2005; Scholtz 2005; Szucsich and Wirkner 2007). Clearly, however, the simple absence of an evolutionary unit as a character state in two exemplars (sensu Prendini 2001) cannot be tested using these criteria (due to the lack of either correspondences or non-correspondences).

A rather more complicated matter is assigning character states to a particular character, that is transformation series. In this connection, the term ‘Rahmenhomologie’ (frame homology; Riedl 1975, 1978) may be useful. Rahmenhomologie implies that some kind of homology relationship must also exist between the states within a transformation series (the relationship which Patterson 1982 called ‘transformational homology’; see also Brower and Schawaroch 1996). Maximum correspondence cannot be the goal in establishing a transformation series, and the defining element, that is an evolutionary connection between the evolutionary units in a transformation series (see Grant and Kluge 2004), can, for obvious reasons, not be used as a criterion. In many cases, however, there will be enough correspondences between the evolutionary units to assign them to a single transformation series (e.g. between the forelimbs of tetrapods), and this is what the term ‘frame homology’ refers to. A general suggestion is that some kind of ‘topographical identity’ should have priority in the establishment of transformation series (e.g. Brower and Schawaroch 1996), but even if we accept this, there are bound to be cases where other aspects might be equally important (see Richter 2005). Where transformations are observable in ontogeny, the start and end points can be included in a transformation series (or the corresponding evolutionary units represented in the adults). Classic examples are the quadratum and articulare as parts of the primary jaw joint and the malleus and incus in the middle ear, or the endostyl in acrania and the thyroid gland in gnathostomes, where few correspondences between the adult evolutionary units exist. Whatever the case, this ‘transformational’ relationship is of a very different kind than that between corresponding evolutionary units (character states) in different exemplars (Szucsich et al. 2013).

Wagner (2007 and earlier publications) proposes, in a continuation of Riedl's (1975) ‘frame homology’ concept, a ‘biological character concept’ in which gene regulatory networks are the defining criterion in the establishment of ‘general’ characters. He suggests, for example, that the forewing of pterygote insects is one such character, which implies that derivations of all kinds, for example the haltera in Strepsiptera or the elytra in coleoptrata, be assigned to ‘character states’. In addition to the practical problem of there being various kinds of elytra to be subsumed under this single character, we fear that dressing what are certainly important developmental findings in cladistic terminology might increase confusion rather than resolving it. The pterygote forewing exists either as haltera or elytra (or as one of many other character states), but not as a generalized construct. On the other hand, there can be little doubt that all kinds of pterygote forewings could be assigned to a single transformation series. If gene regulatory networks in the framework of cladistics were nothing other than a tool for the identification of transformation series, we would agree with Wagner (2007).

Caution is often needed to ensure that characters (i.e. transformation series) are logically independent (see also Wilkinson 1995), as illustrated by the example of the number of maxillipeds and pereiopods in malacostracan crustaceans. In all cases, eight thoracopods are present, and in most cases, these are differentiated into one maxilliped and seven pereiopods. In decapods, however, three pairs of maxillipeds and five pairs of pereiopods are present (Richter and Scholtz 2001). The ‘number of maxillipeds’ and the ‘number of pereiopods’ are not independent from each other and cannot constitute two different transformation series. For us, the logical independence of transformation series is therefore the most important criterion (see also Patterson 1982; Grant and Kluge 2004).

It is not necessary here to describe in detail exactly how morphological phylogenetic analyses are performed (see e.g. Schuh 2000; Wägele 2005), but the central idea is that character states hypothesized to be homologous (through the allocation of the same codes – 0, 1, 2, etc.) are now tested against each other. The most parsimonious distribution of character states is reflected in the most parsimonious topology (Wiley 1981; Richter 2005; Vogt 2007). De Pinna (1991) calls the homology hypotheses prior to the test against other homology hypotheses during the cladistic analysis ‘primary homologies’ and those which are supported by this test ‘secondary homologies’. The identification of characters (=transformation series) and the assignment of the same character states (or their codes, to be precise) to different taxa are tantamount to the establishment of primary homology hypotheses. The cladistic analysis which follows not only reveals the most parsimonious (and therefore preferred) topology; it also constitutes a test of the homology hypotheses of character states (e.g. Richter 2005; Vogt 2007). This should make it clear why cladistic analyses are an indispensable part of Evolutionary Morphology. The result of such analyses is that some (primary) homology hypotheses are corroborated and others are not. The latter are not falsified in the strict sense, though, because according to Popper (1994: p 212), falsification needs to be absolute (see Richter 2005). In this case, a re-evaluation of corroboration and non-corroboration would follow in additional research cycles (Kluge 1997), taking into account new transformation series and new character states. However, the initial analysis should make it possible to estimate (on the basis of branch support, for example), the reliability of certain (secondary) homology hypotheses. In those cases where the homology of a particular character state is not corroborated by phylogenetic analyses, a new ‘character analysis’ (test of primary homology) should be conducted, or the entire character conceptualization process (including the conceptualization of transformation series) revised.

The result of a cladistic analysis is an unrooted network (topology), representing the most parsimonious distribution of character states (i.e. evolutionary units; for each of the transformation series) but not their relative chronological order. It is the rooting with an a priori defined outgroup that transforms a topology into a cladogram and thus into a real phylogenetic hypothesis (Richter 2005). Secondary homologies (sensu de Pinna 1991) become synapomorphies or symplesiomorphies. Only now do the evolutionary units of a transformation series (we are now leaving the technical field of cladistic analyses) fall into an order that reflects the chronology of their appearance. This can be analysed (using various optimization approaches such as ACCTRAN or DELTRAN) and documented using software such as Mesquite (Maddison and Maddison 2002-2011). It is also possible to use maximum likelihood approaches to calculate the probability of a particular chronological order (see Cunningham et al. 1998), but this will not be explored here (see Wirkner 2009). ‘Character mapping’ (based on parsimony or ML approaches) can also be conducted when the underlying phylogenetic analysis is based on molecular data, but we believe that even here, phenotypic data still need to be analysed. This can take place either in comparison to or in combination with molecular data (total evidence; Kluge 1989), but is important as a way of subjecting the analysis as a whole to a more severe test (sensu Popper 1994).

Once we come to the chronological order of evolutionary units in transformation series, we have reached the earliest point at which it becomes possible to venture a causal explanation for evolutionary transformation (see also Lauder 1981). Just as it is important for an understanding of the history of opera to know that Händel and Mozart predated Verdi and Wagner, it is important for a causal understanding of the evolution of the tetrapod limb to know that it started as a pentadactyl appendage (at least in the crown group; Carroll 1988). Both of these examples, however, should also make it clear that knowing the chronological order does not suffice to develop a causal understanding and both should show that the incompleteness of transformation series (or the absence of historical sources) has a major influence on our causal understanding. This highlights the importance of including fossils (if available), both as representatives of potentially unknown character states which might feature in certain transformation series (see Sudhaus 2007), and to help with the absolute dating of certain transformations which can then be correlated to particular environments for a better (causal) understanding of these transformations. A useful notion when it comes to arriving at causal explanations is that of the ‘Phylogenetische Bürde’ (phylogenetic burden; Riedl 1975, 1978). Certain transformations can be ‘explained’ simply by their phylogenetic origin. The presence of seven cervical vertebrae in giraffes and whales, when in the one case more and in the other fewer might be expected, is explained by the fact that both are mammals (Riedl 2000; see also Müller et al. 2010 for a recent approach). Although it can be argued that this does not constitute a full causal explanation, this kind of explanation may satisfy some.

The causal aspect of Evolutionary Morphology

The term ‘causal morphology’ was coined during a relatively short period of Entwicklungsmechanik (Developmental Mechanics) which ended with a switch of focus towards the genetic mechanisms responsible for changes in form (Mocek 1998). Because this period can be viewed as one of the early precursors of Evolutionary Developmental Biology (evo-devo), it should be relatively straightforward to extend the term to explanations generated in evo-devo, too. We use ‘causal morphology’ slightly differently to take in all the areas of research in Evolutionary Morphology which deal in some way with explanations of biological disparity. Schwenk and Wagner (2004) subsume all the approaches which attempt to explain phenotypic evolution under two main headers, externalist and internalist, and go some way towards providing a synthesis.

We start with the relative chronological order of evolutionary transformations. As one evolutionary transformation follows another in a specific order, the result is a limitation of the transformations still possible (in other words of evolvability), and this is referred to as the ‘phylogenetic burden’ (Riedl 1975, 1978). Clearly though, this does not explain the transformations themselves. We will show that it is not always possible to differentiate between proximate and ultimate causes (Mayr 1961). Having said this, we agree with Love (2006) that what counts as an evolutionary cause is much broader than natural selection and those processes which primarily concern populations.

The role of functional morphology in Evolutionary Morphology

A central concept of Darwinian evolutionary theory and of the modern synthesis is that specific traits (here defined as evolutionary units) are the result of adaptation, and that adaptation is forced by natural selection (Darwin 1859; Lewontin 1978; Losos 2011). This implies that evolutionary phenotypic transformations can be explained on one level in terms of the selective advantage of one evolutionary unit over another of the same transformation series given a certain environment (in the broad definition put forward by Losos 2011). This level of explanation might be called ultimate (Mayr 1961). While the selective advantages we are talking about here might be directly testable at population level (e.g. Herrel et al. 2005), on a higher, macro-evolutionary level they are more elusive (Losos 2011).

At this stage, we need to come back to the relationship between form and function, which will now be discussed in more detail. In the main, we follow a concept put forward by Bock and von Wahlert (1965) almost 50 years ago. Three of their terms in particular – function, faculty and biological role (see also Verraes 1981) – are useful for understanding evolutionary transformations. The term function refers to all those process-related properties which arise directly from the form of morphemes, for example from material properties or shape and composition, without reference to the organism's environment (sensu Bock and von Wahlert 1965). The term faculty refers to the combination of form and function. Because every form (always the sum of all predicates) can have a variety of functions, a number of different faculties might also exist. The biological role refers to the actual use of the faculty in the natural environment.

To illustrate this with an example, the chela of a pedipalpus of a scorpion (consisting of the tibia, the immobile finger, and the tarsus, the movable finger) has different faculties (form-function complexes) depending on the species. It can be a ‘slender gripping chela allowing high speed’, for instance, or a ‘robust crunching chela allowing strong forces’ (van der Meijden et al. 2010). The biological role of the chelae is primarily to catch and manipulate prey. In our case, the two groups of scorpions concerned hunt for quite different prey, which implies that the biological roles of the chelae differ. Interestingly enough, the chelae also have a second biological role: male scorpions use them to direct the female to the spermatophore in what is called the ‘promenade à deux’ (Polis and Sissom 1990). In some cases, a sexual dimorphism exists, which helps the males to grip the females. The selective advantage comes from the interaction between biological role and environment (we use the term environment in a general way and ignore Bock and von Wahlert's distinction between umwelt and umgebung). In a specific environment, the fittest individuals (those with the highest reproduction rate) are those individuals of a population which perform a biological role better than others (for the sake of the argument, we focus on one biological role only, although there are clearly other factors in the fitness of individuals). Our example suggests that both natural selection and sexual selection might be of importance, although the manipulation of prey is probably the overriding factor in this particular case because it affects more aspects of the form of the chela (van der Meijden et al. 2010). It is the biological role in a particular environment (here different kinds of prey) which becomes optimized in the process of adaptation. This optimization takes place through changes in the faculty, which again shows that form and function cannot be separated but evolve together.

Our example shows that while functional units might consist of more than one morpheme (e.g. the chela of the two fingers), it is only in conjunction with each other that these morphemes are able to perform the biological role. One might follow Bock and von Wahlert (1965) in claiming that there is no morpheme without function (because there is no form without function) and attribute some kind of function to each of the two fingers, but the actual performance of the function, crunching or gripping, only comes about when the two fingers are combined. An important question is whether or not the ‘biological role unit’ corresponds to the evolutionary unit. Although the two parts of the chela certainly evolve together in some way (in coherence), they do not necessarily only constitute one evolutionary unit; coherence might also exist between different evolutionary units (see section on the role of ‘coherence’ in Evolutionary Morphology).

While function is explicitly unrelated to environment, environment has to be taken into consideration when determining the biological role. In the case of extinct taxa, this can be challenging. Biomechanical studies which demonstrate the remarkable biting force of Tyrannosaurus jaws allow us to draw pretty detailed conclusions about their function (Erickson et al. 1996; Meers 2002), but this alone does not reveal their biological role. It is still well possible that Tyrannosaurus was a scavenger which used its powerful jaws only to tear apart carrion (like some of today's hyenas), as Ruxton and Houston (2003) conclude from their reconstruction of the ecosystem in which Tyrannosaurus lived.

As outlined above, the study of form falls into the area of descriptive morphology. All the investigations and experiments which aim to deduce function from form belong to the area of functional morphology (Dullemeijer 1974). Because knowledge of form and function is necessary (though not sufficient) for an understanding of biological role, functional morphology plays an essential part in Evolutionary Morphology (though some proponents might prefer to define Functional Morphology as an independent field). The field related to functional morphology but which focuses on the biological role has been called ecomorphology (e.g. Bock 1994; Betz 2006).

One very successful field with a direct link to functional morphology is that of bionics (Nachtigall 2010). In bionics, knowledge of organismic form and function is used to imitate faculties. While it is obvious that machines cannot have a ‘biological’ role anyway, even the ‘roles’ they are built to perform differ from the biological role of their biological prototypes (despite comparable faculties). Examples are the ‘lotus effect’ (Barthlott and Neinhuis 1997) and the use of climbing robots to clean pipes (Mämpel et al. 2008; Schmidt et al. 2008).

The significance of sexual selection in explaining evolutionary units was touched upon above. Many additional examples could be cited. If morphemes possess particular functions such as ‘gripping’, ‘holding’ or ‘keeping distance’, their biological role in copulation (Sepsidae; Puniamoorthy et al. 2009; Tan et al. 2011) or in defending territory (deer head, Sudhaus and Rehfeld 1992) is the crucial aspect. In morphemes which are exclusively for display (e.g. the red gular sac in frigatebirds, Schreiber and Burger 2001), however, the situation may be different. This might be the only case in which no clear function can be identified without knowledge of the biological role.

Adaptation is mostly discussed with reference to external features which interact directly with the environment, but it is clear that it must refer to internal features, for example musculature and nervous system, as well. That function might arise directly from form is self-evident to the (electron-) microscopic anatomist, independently of the fact that it is always useful to test assumptions of this nature within the framework of physiological experiments. On the basis of the shape and electron density of the crystalline cone in arthropod compound eyes, it is possible to draw conclusions about refraction ability and, ultimately, optical design, though electrophysiological experiments to support these conclusions certainly would not go amiss (Land and Nilsson 2012). Muscles are characterized by their origin and insertion, and this information alone can often suffice to reveal their function as flexors, extensors, abductors or adductors. However, a full picture of muscle function is only provided by an analysis of the fascicles and fibres and their mechanical properties (e.g. Fischer and Lilje 2011). Understanding the function of nervous system structures can be even more complex. Knowing the course of a nerve might allow us to narrow down its possible functions (in particular if the target is known, e.g. sensilla or muscle), but on the basis of ultrastructure alone, it is at least difficult to conclude whether the nerve is motoric (efferent) or sensoric (afferent). To confuse things even further, in many cases (if not all), both muscles and nerves only have a biological role (or various roles) in combination with their target organs. Whether or not they can still be deemed to constitute evolutionary units in their own right remains to be shown (see discussion below).

In conclusion, we need to turn our attention to material constraints (a field which has its origin in the seminal work of D'Arcy Thompson 1917; see also Seilacher 1970, 1973). Material constraints are important not only for our understanding of form (in that they limit form), but also, following the logic discussed above, for our understanding of function and biological role. Material constraints are not necessarily insurmountable over the course of evolution, but they do limit the function of morphemes and, ultimately, the evolvability of biological roles. To give an example, the evolution of orb webs was made possible by the evolution of a particular silk protein in the ancestors of Araneomorphae. The silk of the other aranean taxa Mesothelae and Mygalomorphae is not strong enough for orb webs (Swanson et al. 2009). Thus, the evolvability of certain traits related to orb web building (particularly behavioural traits) was and is limited in the two taxa displaying the plesiomorphic condition of silk.

The role of ‘coherence’ in Evolutionary Morphology

Coherence morphology, or Kohärenzmorphologie (Richter 2007), is the term we give to the identification of coherence between elements of a structural or functional network of an organism. Coherence describes the non-accidental mutual presence of specific morphemes or evolutionary units, regardless of whether the mechanism responsible is external or internal (see Schwenk and Wagner 2004). Coherence can either be genetically (or epigenetically) fixed or the result of a historical or architectural constraint. The questions and problems surrounding coherence are well known (e.g. Dullemeijer 1974; Lauder 1981, 1990; Schwenk 2001).

We start with the seminal publication by Gould and Lewontin (1979), who criticized the tendency of evolutionary biologists to focus on the adaptation of individual ‘characters’ (the adaptationist approach). Their famous example, taken from architecture, relates to the spandrels of St. Mark's cathedral in Venice. Although they might appear to have been planned by the architect for the sole purpose of accommodating the cherubim painted there, they are actually born of architectural necessity as the only means of joining two neighbouring arches. There is, then, coherence between the arches and the spandrel. In our reasoning, the spandrel as an architectural element cannot be conceptualized as an evolutionary unit in its own right (for the sake of the argument, we treat it here as a biological object) because it only exists in combination with the two arches. The evolutionary unit is the combination of the three elements. In this case, coherence does not exist between evolutionary units (which potentially have their own adaptive value) but between the separate morphemes which make up a single evolutionary unit. Interestingly enough, the cherubim painted on the spandrel could easily represent an ‘evolutionary unit’ in its own right, and one independent of the other ornaments on the arches. We suggest that the term ‘architectural constraint’ (Gould and Lewontin 1979) be restricted to coherence between morphemes.

An important indication of coherence between character states of different transformation series is their mutual presence at multiple nodes in the cladogram. In the cases mentioned above, coherence should have been identified a priori and used in the character conceptualization process which took place at the level of comparative morphology. A helpful approach is to treat morphological analyses as research cycles sensu Kluge (1997) in which the evolutionary units themselves constitute hypotheses just as much as the potential homology between evolutionary units of different species does.

That coherence exists between the right and left side of bilaterian animals has already been mentioned. The left and right appendages in tetrapods (forelimbs and hindlimbs) or arthropods, for example, together form an evolutionary unit each. Asymmetries are rare but of particular interest if present because coherence is undone. ‘True asymmetries’ (e.g. those found in hermit crabs) are genetically fixed, while in ‘antisymmetries’ (e.g. the cheliped of lobsters or fiddler crabs), there is no genetic fixation of the side on which a particular form will appear. Some level of explanation is provided by evo-devo approaches, and asymmetry is the topic of ongoing research (reviews by Palmer 2004; Levin and Palmer 2007). One important question that still needs to be asked is whether it is always a divergence in the biological roles of the left and right appendages, which leads to the uncoupling of the processes involved in their development and results in the independent evolution of their form-function complexes, or whether it is sometimes the other way round (Palmer 2012a, b). Whatever the case, adaptation and developmental constraints, here regarded as the two main aspects of causal morphology, clearly need to be considered together.

An interesting case of coherence is that involving the allometries between body proportions (see Gould and Lewontin 1979). The two variables that are related to each other here are quantitative and not qualitative predicates of morphemes. If a specific and stable relationship (one of isometry or positive or negative allometry) exists across multiple organisms, the two variables might not be independent from each other but in some way coupled (i.e. coherent). Allometry is found in connection with the growth of organs in individuals (growth/ontogenetic allometry), but also, and more relevantly here, between organisms of a population or species (static allometry) and between species or even species groups and other higher taxa (evolutionary allometry; Shingleton 2010). An interesting and illuminating example is formed by the relationship between elytra and mandible length in male stag beetles (Lucanus cervus). In smaller males, the allometric relationship between these two variables is positive, while in larger males, it is negative, indicating that the relationship itself is also variable (Sudhaus and Rehfeld 1992). The developmental mechanisms behind this are only partly understood (Stern and Emlen 1999; but see Moczek et al. 2014 for a recent account on horned beetles). Evidently, the allometric relationship is changeable by selection, which implies that the coupling is reversible (Wilkinson 1993). In some cases, however, coherent relationships exist across higher taxa (e.g. the relative size of the cranium in Equidae and Felidae; Sudhaus and Rehfeld 1992), which implies that some types of coherence can prevail for millions of years.

Allometry research extends into the field of morphometrics, an area of study which also has its origin in the work of D'Arcy Thompson (1917) and which is very active today (Zelditch et al. 2004; Catalano et al. 2010). The central focus of morphometrics is the study of shape, which by definition cannot be changed by translation, rotation or rescaling. Morphometrics uncovers relationships between landmarks. In many cases, the character might be constituted by the relationship between landmarks not by the landmarks themselves. Coherence between landmarks thus needs to be taken into consideration in the process of character conceptualization. Recently, procedures have been developed for the cladistic analysis of morphometric data (Catalano et al. 2010; Goloboff and Catalano 2011).

In many cases, coherence exists not between morphemes (as part of evolutionary units) but between evolutionary units that are generally considered to have evolved independently. A useful term for such a group of evolutionary units is ‘evolutionarily stable configuration’ (ESC; Wagner and Schwenk 2000). Knowing about this kind of coherence is important for our understanding of phenotypic transformations, for it too limits the potential evolvability of the evolutionary units in question. Lauder (1990) suggests that entire classes of evolutionary units (e.g. phenotypic and behavioural) could be affected. An example is found in Dullemeijer's (1974) hypothesis that the size of the eyes in birds may not just influence the form of the orbita, as it is known to do, but also affect brain size and even beak shape. Vertebrate morphology provides many comparable examples, and the identification of ‘evolutionarily stable configurations’ of this nature has been the aim of a number of functional morphological studies (Schwenk 2001 and citations herein). Evolutionarily stable configurations are complexes, often comprising a range of different morphemes, that function together in a concerted manner to produce a functional output (Wagner and Schwenk 2000) and, it should be added, to fulfil a particular biological role. When the evolutionary units of which they are comprised are closely integrated, ESCs behave like a single evolutionary unit. Integration is likely the result of a combination of developmental constraint (canalization), functional constraint (internal selection) and external selection. Under certain circumstances, however, these constraints are resolvable and the combined evolutionary unit may separate into independent evolutionary units again (Wagner and Schwenk 2000; Schwenk 2001).

An interesting but as yet only partially studied case of coherence might be the crab-like habitus of many decapod crustaceans, which has evolved several times in a process called carcinization (Borradaile 1916). Carcinization concerns not only external appearance (a carapace broader than it is long, a pleon folded under the cephalothorax) but may also influence features of the internal anatomy such as the circulatory system and the arrangement of endosternites (Keiler et al. 2013). Although we presume that there is some level of coherence between the different morphemes involved here (and perhaps between the evolutionary units; that is that their mutual presence is non-accidental), this hypothesis needs to be tested further. Another example concerns the extension of the circulatory and respiratory systems in arthropods, which appears to be mutually dependent (Richter and Wirkner 2004; Wirkner et al. 2013). Here too, however, the precise extent of the coherence(s) needs to be tested.

A very different case of coherence between evolutionary units is that of paedomorphosis, the phenomenon in which predicates of juveniles appear in sexually mature animals. Paedomorphosis is the result of two different processes, either progenesis or neoteny (Gould 1977). These processes simultaneously influence evolutionary units of quite different transformation series about whose independent evolvability there can be no doubt. In the case of the neotenic Urodela (e.g. the Axolotl), the external gills, the limbs, the eyes and the tail fin are all affected. In the admittedly questionable case of the progenetic cladocerans (Claus 1876), the number of limbs, the extension of the carapace and the function of the antenna all seem to be involved. Gould (1977) suggests that as a life history strategy, precocious maturation could constitute a selective advantage. What needs to be tested here, though, is clearly not the adaptive value of the single evolutionary unit but that of the entire organism. A recent study by Fritsch et al. (2013), however, shows that the developmental mechanisms, that is heterochronic processes, leading to a paedomorphic appearance are much more complicated than Claus (1876) believed.

Coherence between evolutionary units might also exist where selective pressure affects different evolutionary units in parallel, as in cases of adaptation to specific habitats such as the interstitial (here, it would concern the general shape of the animal, the double-gland system, internal fertilization). The fact that the same correspondences are found in different species (often of very different taxa) is strongly indicative of coherence (Rieger and Tyler 1985). This highlights the importance of ‘analogy research’ in morphology (Lorenz 1973). The term analogy as it is used here simply refers to all the correspondences between morphemes (or conceptualized evolutionary units) that are not the result of common origin (e.g. Riedl 2000); the term convergence is sometimes used in the same way (Losos 2011). Analogies might either be identified a priori (i.e. similar morphemes are conceptualized as different evolutionary units) or in the course of a cladistic analysis if character states are not corroborated as homologues (refuting the hypothesis of secondary homology; see Richter 2005). The occurrence of analogies is of particular interest in the area of causal morphology because analogues might plausibly have analogous explanations (e.g. constructive constraints, as in suckers; similar biological roles; a combination of both, as in the bivalved shells of mussels, ostracods and brachiopods; or identical developmental mechanisms), allowing us to draw ‘conclusions by analogy’.

The role of Evolutionary Developmental Biology (evo-devo) in Evolutionary Morphology

We conclude by examining that level of explanation offered by the study of the ontogeny which – as Evolutionary Developmental Biology – today plays a central role in the explanation of phenotypic disparity. Evo-devo was popularized by Kirschner and Gerhart (2005) and Carroll (2005) and the philosophical background behind it provided by authors such as Amundson (2005). Since its inception in the early 1980s (see Holland 2004), evo-devo has, inarguably, evolved into a mature discipline (Müller 2008). The direct connection between ontogeny and phenotypic disparity is obvious because all organs and parts of organs (all morphemes, we might say) develop, and development ultimately represents the realization of the genotype into a phenotype. All developmental stages (e.g. specific cleavage stages) are intermediate stages and part of the phenotype at the same time (Woodger 1945). The particular ontogeny of an individual is always the proximate cause of the form of any morpheme (Laubichler 2000). It is equally obvious, however, that evolutionary changes, that is transformations, are reflected in development.

While early evo-devo focused on heterochrony, that is differences in the timing of the appearance of differentiations with a direct influence on adult structures (Gould 1977), and developmental constraints (Maynard Smith et al. 1985), gene regulation (including gene regulatory networks) and spatial and temporal sequences of gene expression are the main topics of evo-devo research today (Carroll 2005; Müller 2008). Other mechanisms such as gene duplication and gene cooptation still need to be looked into (Oakley 2007), as does the addition of new gene regulatory elements (e.g. microRNAs; Peterson et al. 2009). Correspondences in gene regulatory networks might be responsible for the conservation of ‘biological characters’ (Wagner 2007) and ultimately provide an explanation for the phylogenetic burden (sensu Riedl 1975).

The search for correspondences and differences in gene expression patterns and gene regulation makes evo-devo a predominantly comparative science, but one which does also aim to provide causal-mechanistic insights into how evolution works. Evo-devo incorporates various traditions from evolutionary morphology (Olsson et al. 2006), and for Love (2006), the two disciplines meet in the identification and conceptualization of evolutionary innovations and novelties. Of particular interest is the correlation between transformations in phenotypic evolutionary units and transformations in developmental pathways. Changes in gene expression and regulation and the addition of new genes are seen as the final causes of all phenotypic transformations and ultimately also of the appearance of evolutionary novelties. Within the framework of Evolutionary Morphology, studying the gene expression and gene regulation patterns responsible for the realization of features in an individual allows us to compare the developmental basis of two evolutionary units in a transformation series. These ‘developmental units’ can initially be viewed as parts of a new transformation series which can be treated in the same way as other series. In the resulting phylogeny, transformation from one developmental pathway to another can be correlated with phenotypic transformations to provide a representation of the ‘whole’ transformation (see Abouheif 1997). Finally, the differences in developmental genetic pathways between two species are interpreted as being causally responsible for the phenotypic transformation(s) between these species. This kind of correlation might appear to be more direct than the kinds discussed above, as various experiments have shown that changes in gene expression directly influence the resulting organogenesis (Liubicich et al. 2009). Nevertheless, experimental and phenomenological approaches are still restricted to extant organisms. The correlation between changes in developmental pathways and differences between organisms has been demonstrated convincingly in many cases, but to conclude that changes in developmental pathways were also the key to phenotypic transformations and evolutionary novelties in evolutionary history are to conclude by analogy.

While we do not doubt the value of such conclusions, we would nevertheless like to stress that the study of the development (and this includes the developmental genetics) of single individuals gives no more direct insights into the evolutionary process than the study of selective forces in populations does into the mechanisms of adaptation on higher phylogenetic levels. And again, it is the phylogeny that indicates the direction of change. In our view, while evo-devo certainly adds to our understanding of the evolutionary causes of phenotypic transformations, it does not replace the other levels of causality. The other elements in the area of ‘causal morphology’, including selective forces, ecological environment at the time of transformations, potential coherences to other transformation series and the historical burden, are still necessary to our understanding of the evolution of form and disparity.

Conclusion

Evolutionary Morphology is described here as a program of research which takes in the description, comparison and explanation of organismic disparity. It is understood to be much broader than morphology in general. The various areas of Evolutionary Morphology – descriptive, comparative and causal morphology – are strongly interconnected by research cycles. Although it overlaps with other disciplines in evolutionary biology, Evolutionary Morphology is a discipline in its own right and is an indispensable part of an extended evolutionary synthesis.