See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/365002477

The Hierarchical Basis of Serial Homology and Evolutionary Novelty

Article in Journal of Morphology · November 2022

DOI: 10.1002/jmor.21531

CITATIONS READS
7 404

3 authors:

James DiFrisco Alan C. Love

The Francis Crick Institute University of Minnesota Twin Cities
27 PUBLICATIONS 504 CITATIONS 147 PUBLICATIONS 2,554 CITATIONS

SEE PROFILE SEE PROFILE

Gunter Wagner
Yale University
527 PUBLICATIONS 28,043 CITATIONS

SEE PROFILE

All content following this page was uploaded by James DiFrisco on 04 November 2022.

The user has requested enhancement of the downloaded file.

The Hierarchical Basis of Serial Homology and Evolutionary Novelty

James DiFrisco* 1, Alan C. Love 2,3, and Günter P. Wagner 4,5,6,7

1) Institute of Philosophy, KU Leuven, Leuven, Belgium

2) Department of Philosophy, University of Minnesota, Minneapolis, MN, USA
3) Minnesota Center for Philosophy of Sciences, University of Minnesota, Minneapolis, MN, USA
4) Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT, USA
5) Yale Systems Biology Institute, Yale University, New Haven, CT, USA
6) Department of Obstetrics, Gynecology and Reproductive Sciences, Yale Medical School, New
Haven, CT, USA
7) Department of Obstetrics and Gynecology, Wayne State University, Detroit, MI, USA

*Corresponding author: [email protected]

Conflict of Interest Statement: the authors declare no conflicts of interest

Funding: JD is supported by the Research Foundation – Flanders (FWO), Grant Number 88559.
ACL and GPW are supported by the John Templeton Foundation (JTF), Grant Number 61329.
The opinions expressed in this paper are those of the authors and not those of the JTF.
Manuscript Word Count (including legends and footnotes): 13,122
References Word Count: 2,427
4 Figures
1 Table

1
Abstract

Given the pervasiveness of gene sharing in evolution and the extent of homology across the tree
of life, why is everything not homologous with everything else? The continuity and overlapping
genetic contributions to diverse traits across lineages seem to imply that no discrete determination
of homology is possible. Although some argue that the widespread overlap in parts and processes
should be acknowledged as “partial” homology, this threatens a broad base of presumed
comparative morphological knowledge accepted by most biologists. Following a long scientific
tradition, we advocate a strategy of “theoretical articulation” that introduces further distinctions to
existing concepts to produce increased contrastive resolution among the labels used to represent
biological phenomena. We pursue this strategy by drawing on successful patterns of reasoning
from serial homology at the level of gene sequences to generate an enriched characterization of
serial homology as a hierarchical, phylogenetic concept. Specifically, we propose that the concept
of serial homology should be applied primarily to repeated but developmentally individualized
body parts, such as cell types, differentiated body segments, or epidermal appendages. For these
characters, a phylogenetic history can be reconstructed, similar to families of paralogous genes,
endowing the notion of serial homology with a hierarchical, phylogenetic interpretation. On this
basis, we propose a five-fold theoretical classification that permits a more fine-grained mapping
of diverse trait-types. This facilitates answering the question of why everything is not homologous
with everything else, as well as how novelty is possible given that any new character possesses
evolutionary precursors. We illustrate the fecundity of our account by reference to debates over
insect wing serial homologues and vertebrate paired appendages.

Word count: 267

Keywords: serial homologues, homology, levels of organization, insect wing evolution, paired
appendages

Research Highlights (characters w/spaces = 251)

● Serial homologues form hierarchical, phylogenetic relationships based on a history of

evolutionary origination like gene paralogues
● Our novel classificatory scheme for serial homology provides an improved theoretical
framework for comparative biology

2
1. The Need for Theoretical Articulation

In a recent paper (Ralevski et al., 2022), a striking discovery was reported. A mitochondrial

gene in Arabidopsis that has a highly conserved homologue in mammals appears to play similar

developmental and physiological roles across this yawning phylogenetic gap, including effects on

movement (for mice, locomotion; for plants, hyponasty). Setting aside the question of what is

meant by “homologous functions” (Love, 2007), this discovery harbors two conceptual issues at

the foundations of evolutionary biology: the pervasiveness of gene sharing in evolution

(Piatigorsky, 2007) and the nature and extent of homology across the tree of life (Wagner, 2014).

These jointly form a problem that remains unresolved. Given the fact of common descent, why is

everything not homologous with everything else? The composition and developmental formation

of “traits” at different levels of organization across taxa are affected by extensive gene sharing that

make it difficult to discriminate, at the genetic level, one trait from another. As a result, drawing a

sharp contrast between traits that are homologous and others that are not seems to rely on

selectively ignoring some portion of this ubiquitous common inheritance. The continuity and

overlapping genetic and developmental contributions to diverse traits across lineages seem to

imply that no discrete determination of homology is possible.

This problem is especially acute for hierarchical notions of homology that focus on the

developmental-mechanistic basis of characters at different hierarchical levels. It has been known

for some time that homologous characters develop from different and often diverse sets of causes

(de Beer, 1971; Spemann, 1915). Empirical studies in evolutionary developmental biology (evo-

devo) have demonstrated that homologues can be induced by different signaling pathways in

different species and can arise from different germ layers and cell populations, sometimes

exhibiting unexpected variability in the rearrangement of components and underlying processes

3
(Haag, 2014; Haag & True, 2021; Hall, 2003). However, it is also the case that key signaling

pathways and cell types are involved in developmental mechanisms that give rise to different traits;

the existence of shared components or genetic networks for characters that are seemingly non-

homologous (e.g., eyes or limbs) suggests “deep homology” (Shubin et al., 1997, 2009). Other

authors, building from this data, argue that the widespread overlap in parts and processes should

be acknowledged as “partial” homology (Abouheif, 1999; Moczek, 2014; Roth, 1984). Since most

characters share some genetic and developmental commonality (i.e., “partial” homology can

always be detected), the distinction between homology and non-homology seems to blur. This

threatens a broad base of comparative morphological knowledge assumed to be robust by most

biologists.

The problem is compounded when we turn to inherent difficulties with the concepts of

serial homology and evolutionary novelty. Both have faced criticism within and among

communities of biologists. For serial homology, it is unclear why repeated structures within the

same organism are an instance of homology if homology requires phylogenetic relationships

(Fusco, 2022; Schmitt, 2017). For evolutionary novelty, the very idea seems anchored in the

distinction between homology and non-homology (i.e., novelty = non-homology; Müller &

Wagner, 1991). Assuming this distinction, “partial” homology implies that nothing is a novelty.

For any pair of characters, they are only more or less diverged. If this distinction is rejected (i.e.,

novelty does not require non-homology), the ubiquity of “partial” homology implies that

everything is partially novel. And this doesn’t even address the tendency of biologists to label traits

as novel according to diverse and conflicting criteria (Brigandt & Love, 2012).

In the history of biology, and in morphology specifically, a strategy appears repeatedly

when biologists are faced with difficulties in conceptualizing complex phenomena, usually

4
because new data arrives rapidly and thereby contributes to the disruption of theoretical categories.

This strategy involves introducing distinctions to extant concepts, sometimes revising or

narrowing their original meaning, to produce increased contrastive resolution among the labels

used to represent biological phenomena. Richard Owen famously exemplified this strategy when

he distinguished “general homology,” “special homology,” and “serial homology” (Owen, 1848).

More recent examples include “positional homology” (Minelli, 1998), “organizational homology”

(Müller, 2003), and “deep homology” (Shubin et al., 1997, 2009). We label this strategy

“theoretical articulation” because it involves articulating an elaborated set of conceptual

relationships to yield productive theory for doing empirical work. Importantly, this strategy can

fail, evidenced by terms and distinctions that do not catch on (e.g., Lankester’s (1870) “homogeny”

or Sattler’s (1984) “dynamic homology”) or underlying difficulties with the distinctions (e.g., the

connection between “general homology” and a presumed ideal archetype, or the fact that two body

parts can stand in a relation of deep homology while being homoplasies, serial homologues, or not

even candidate homologues [see (DiFrisco et al., 2022)]). And sometimes there is value in

exploring connections across different approaches to theoretical articulation (Novick 2018). Yet a

successful theoretical articulation can enrich biological investigation (e.g., the separation of

orthology and paralogy for gene homology), and eventually undergird much of the basic

vocabulary utilized daily by biologists.

We think the time is ripe to undertake new theoretical articulation in relation to serial

homology, and in a way that complements and augments recent work of this kind for evolutionary

novelty and co-option (DiFrisco et al., 2022). Here we adopt a specific tactic to pursue this strategy

that involves examining the successful patterns of reasoning that have coalesced in talking about

homology at the level of gene sequences (orthology, paralogy, gene families). These reasoning

5
patterns are now stable in genomics, population genetics, and protein biology; together they offer

a basis for additional theoretical articulation to address the problem described above. In

combination with important desiderata for the concept of homology, such as the requirement that

homologues form equivalence classes and therefore involve “sameness” (i.e., in contrast to

similarity or partial correspondence), we elaborate an enriched characterization of serial homology

as a hierarchical, phylogenetic concept with a five-fold theoretical classification. This proposal is

based on an account of what it means for a part to have identity as a biological character (DiFrisco

et al., 2020) and permits a more fine-grained mapping of diverse trait-types discussed by biologists,

while also answering the question of why everything is not homologous with everything else. We

illustrate the fecundity of our account by reference to standard empirical problems involving serial

homology, such as the evolution of insect wing serial homologues, the vertebral column, and

paired appendages of gnathostomes, which have been active sites of discussion within evo-devo

over the past decade. An advantage of the resulting hierarchical picture of serial homology is that

evolutionary novelty can be established without a complete knowledge of precursors and

transformation series, thereby blunting concerns that “new” characters are always in the eye of the

beholder and subject to deflationary interpretations based on future empirical studies. In closing,

we consider the broader significance of how concepts are constructed and used in scientific

inquiry, highlighting domains where the status of theoretical articulation remains an open question

(e.g., “character swarms”).

2. Is serial homology really “homology”?

Homology, in our view, is a hypothesis about the evolutionary traceability of biological

characters, which is underwritten by their developmental individualization. Sameness implies a

6
kind of continuity that is discovered in an act of tracing. In contemporary biology, the tracing of

genes across species uses sophisticated algorithms to identify the most likely or parsimonious

historical scenario to support the notion that a particular gene a in species A is in fact the same

gene as gene b in species B (i.e., it can be traced to an ancestral gene). This includes an initial

discovery of similarities (sequence alignment) and the reconstruction of the evolutionary history

of the genes across species (phylogenetic inference). Similarly, morphological homologies are

discovered via a process of initial similarity discovery (“primary homology”) (de Pinna, 1991),

which leads to the construction of a historical scenario connecting two characters to a hypothetical

ancestral character, thus establishing a continuity of evolutionary history. Although this is

relatively uncontroversial for special homology (i.e., the same character in different species), it is

unclear whether this way of thinking also applies to serial homology (i.e., the relationship between

repeated parts of the same organism, such as ribs in a vertebrate or the legs of a centipede).

The conceptual intricacy of serial homology derives primarily from the fact that serial

homologues are not just intra-organismic structures—segments, leaves, teeth, hepatocytes, or

cartilage—but also are shared across taxonomic groups. Cervical and thoracic vertebrae are serial

homologues within an individual organism but having a complement of vertebrae is also a

synapomorphy of the tetrapod vertebrates as a clade. This dual aspect of serial homology generates

three possible ways of understanding serial homology: (A) phylogenetically, (B) intra-

organismally, or (C) some combination of both.

(A) Tracing special homologues through phylogeny comes with a well-established framework of

tools and tests from systematics and cladistics, but it is not obvious how to apply these to characters

within the same organism. Some authors have conflated the seeming lack of phylogenetic tests

7
with the lack of any tests for serial homology, concluding that it is a defective concept (e.g.,

Patterson, 1982). The idea is that different parts of the body are not related to each other through

a process of phylogenetic change and lineage splitting; thus, they cannot be accommodated within

an evolutionary understanding of homology. From this perspective, serial homology is a misnomer

and a confusion. Alternatively, serially repeated structures in an organism can be grouped together

as comprising a single character—for example, vertebrae comprising the vertebral column—which

can then be investigated as to its phylogenetic relationships, just like non-serial characters such as

the brain or skin (Nixon & Carpenter, 2012). This perspective tends to assimilate serial homology

to special homology for characters with iterated subparts.

(B) Serial homology might be understood as applying to structures within an organism that are

homologous according to morphological criteria, such as sameness or symmetry of position

relative to major body planes, or that possess the same developmental mechanisms controlling

their identity. The reasoning to establish such homologies might rely on comparative methods to

infer the existence of these mechanisms, but this is nonetheless consistent with serial homology as

an intra-organismal relationship that is independent of phylogeny (Goodwin, 1994).

(C) The crucial phenomenon missed by both alternatives is that characters—and not just species—

form phylogenetic trees of ancestral and derived identities, and it is often the case that multiple

branches deriving from the same ancestral character coexist within the same organism. The latter

characters are serial homologues. We can express this idea that combines both phylogenetic and

intra-organismal aspects into a compact definition:

8
 Serial homology: Two characters in the same organism are serial homologs if they are
derived by differentiation from a single character or a set of multiple identical characters
in an ancestral species.

The clearest picture can be obtained by starting from gene paralogy. Genes can be orthologous if

the same gene is found in the genomes of two different species and if there is a one-one relationship

between them (i.e., if there is no indication of a gene duplication event in evolutionary history).

However, there are also genes in the same genome that are phylogenetically related to each other,

so-called paralogous genes. These genes originated by duplication and have acquired different

functions because of undergoing largely independent evolutionary transformations. Classic

examples include the vertebrate Hox gene clusters, which arose by whole genome duplication at

the base of the vertebrate tree (Gehring, 1998). These are genes in the same genome that are

phylogenetically related to each other and thus can be traced to an unduplicated gene in a

hypothetical ancestor, as confirmed by unduplicated genes in outgroup taxa (e.g., amphioxus).

They are clearly homologous but, since they are in the same organism (genome), must be serially

homologous. If the number of paralogous genes is larger than two, then it is useful to think of the

set of paralogous genes as a gene family that has a history of duplication and divergence, which

can be represented by a gene tree. This phylogenetic tree reflects the evolutionary history of the

members of the gene family. Hence, in the case of paralogous genes, we have two or more

characters in the same organism that are phylogenetically related—serial homologues. Can this

way of thinking be applied to morphological characters? To approach this question, let us move

up one level of organization and first consider cell types.

In the last two decades, there has been a growing realization that cell types can be

understood as units of evolutionary change just like genes at the molecular level (Arendt, 2008;

Arendt et al., 2016; Wagner, 2014). Akin to genes or morphological characters, cell types from

9
different species can stand in the relationship of homology. In addition, there are cell types that

are clearly differentiated from one other but more closely related to each other than to other cell

types. For instance, macrophages are cells from the innate immune system of vertebrates and are

perhaps homologous to the phagocytes of invertebrates (Buchmann, 2014; Hartenstein &

Martinez, 2019). However, there are also specialized sub-groups, such as the resident macrophages

of most organs: microglia of the brain, Kupffer cells of the liver, alveolar macrophages of the lung

and uterine macrophages of the endometrium (Okabe & Medzhitov, 2016). Most likely, these are

different cell types specialized for different functions, phylogenetically related to each other and

derived from an ancestral phagocyte of pre-chordate origin. The same is likely true of neurons,

which probably arose once or twice in animal history (Kristan, 2016). There are many neuron types

in the brain and the diversity of cerebral neuron types likely arose from a history of evolutionary

cell type differentiation (Arendt, 2008; Arendt, et al., 2019; Arendt et al., 2016). Cell types from

the same organism appear to form cell type families, where they display different levels of

relatedness based on an evolutionary history of sister cell type differentiation (Arendt et al., 2016).

Thus, cell types in the same body that arose in evolution via differentiation from an ancestral cell

type are examples of serial homology. Again, we see a strong relationship between serial

homology and phylogenetic (special) homology.

Next, we can ask whether this pattern of reasoning scales up to morphological characters.

The so-called epidermal appendages (e.g., hair) are an example of characters that occur in the same

organism and are phylogenetically related like cell types (see Figure 1). A mammalian hair is not

just a keratinous shaft growing out of the skin, but a complex organ that consists of a sebaceous

gland, muscles, and the hair shaft proper with a follicle. In comparative anatomy, it is referred to

as a pilo-sebaceous unit (PSU). There is an interesting relationship between the PSU and mammary

10
gland, the latter of which is evolutionarily derived from (and thus related to) the PSU (Blackburn,

1991; Oftedal, 2002). The connection between PSUs and mammary glands is supported by the fact

that the developing mammary glands in marsupials and some eutherian mammals produce hair

that is lost later during development. Hence, the PSU and mammary glands are serial homologues

and phylogenetically derived from an undifferentiated PSU.

Realizing that serially homologous cell types and structures are related to each other as

represented in a phylogenetic tree makes it clear that they can have different degrees of relatedness,

just as some species are more closely related to each other than they are to other species (e.g., two

species of mice are more closely related than they are to cows). The same is true for paralogous

genes; HoxA1 and HoxB1 are more closely related to each other than HoxA1 is to HoxA2, which

duplicated earlier than HoxA1 and HoxB1 (Prince & Pickett, 2002). Degrees of relatedness among

cell types reflect the phylogenetic history of differentiation rather than degrees of similarity or

number of shared parts. All members of a cell type family are serially homologous, with their

relatedness determined by the number of branching points separating them on the character tree.

Figure 1. A simplified hypothetical character tree of epidermal appendages. On the right, feathers and scales are
characters that exist in “reptiles” (including birds); feathers are evolutionarily derived from ancestral body scales and

11
thus feathers and scales are more closely related to each other than each is to hair. On the left, three examples of
mammalian skin appendages are shown. As drawn, the tree suggests that different types of body hair have arisen more
recently than mammary glands, but hair (i.e., the “pilo-sebaceous unit” [PSU]) is the ancestral state from which
mammary glands evolved (see main text for references). Note: the hypothesis depicted here requires us to make
assumptions about the body cover of pre-mammary gland mammalian animals.

Importantly, not all iterated structures that are commonly considered serial homologues

form traceable character trees. Some repeated cellular elements are simple developmental copies

of each other or copies of a prototype, such as different cells of the same cell type (e.g., two

neutrophil blood cells), which are different instantiations of the same differentiation program.

They are iterations of the same cell type and have a developmental traceability to a progenitor cell

type (e.g., stem cell) in the life of an organism. However, developmental traceability does not

connect them to the same individual stem cell because different instances of the stem cell can give

rise to the same neutrophil cell type. Thus, cells of the same type in a body do not need to be

developmentally traceable to the same cell in the embryo.

There also are repeated elements that are highly similar but more individuated than cells of

the same cell type. A morphological example is the forewing and the hindwing of a bee. Both are

similar blade-like structures composed of two apposed epithelial cell layers and the associated

cuticle. Yet they occur in different locations of the body (mesothorax and metathorax) and differ

in size and shape across insect taxa. Shape differences can become extreme, like that between the

forewing and hindwing of a beetle or that of a dipteran. In beetles, the forewing is transformed into

a protective cover (“elytra”), while the hindwing often (but not always) functions as a wing. In

dipterans, the forewing is (in most cases) a functional wing and the hindwing a haltere (i.e., a club-

shaped sense organ). This example illustrates that these repeated elements can be similar, like in

the case of a honeybee or a mayfly, but in different instantiations show remarkable differentiation,

like in beetles and flies. Similarity of the forewing and hindwing suggest relatedness, but they also

12
are developmentally and evolutionarily individuated; the forewing, as a homologue across

pterygote insect species, has a different evolutionary history of transformation than the hindwing.

The examples of paralogous gene copies, sister cell type families, and iterated

morphological characters demonstrate that our proposed phylogenetic notion of character trees of

serial homologues is applicable primarily to developmentally and evolutionarily individuated

units. Traceability of historical relationships does not apply to identical copies of the same element;

at best, we can trace their developmental history in the lifetime of the individual in which they

occur. Cells of the same cell type form ahistorical classes rather than historical individuals and

thus should be distinguished from serial homology. Here the need for further theoretical

articulation is palpable since the terminology of serial homology alone does not provide that

distinction. Before we turn to this task, let us briefly consider instances of duplicated genes that

do not form gene families, as truly paralogous genes do, but behave more like copies of the same

cell type in a phenomenon labeled “concerted evolution” (Ganley & Kobayashi, 2007; Zimmer et

al., 1980).

Ribosomal RNA genes exist in the genome as multiple copies. This is needed to produce

enough ribosomal RNA to form the ribosomes necessary for a high level of protein production

rate. Surprisingly, different copies of the ribosomal RNA genes have a very shallow phylogenetic

signature; there are no one-one orthologues of individual rRNA genes from one species to the next.

These genes get duplicated and lost at a high rate by unequal crossover or become homogenized

with respect to their sequence through gene conversion. The result is that members of the rRNA

cluster do not have individualized gene copies, unlike other gene families (e.g., hemoglobin or

Hox genes). Identical copies or replicas of the same biological unit are a different phenomenon

13
than serial homologues of genes and cell types, which do have a historical phylogenetic

relationship to each other and thus exhibit an aspect of phylogenetic homology.

The distinction between copies or replicas and serial homologues is at odds with an earlier

proposal by one of us (Wagner, 2014, Figure 2.13A). In that proposal, replicas were subsumed

under the general notion of serial homology as “homomorphs” and distinguished from

“paramorphs” (i.e., individuated serial homologues). At that time, the notion of evolutionary

traceability as the conceptual core of identity had not been appreciated; here it is considered central

to any form of homology, serial or not. Again, the need for theoretical articulation is pressing;

further distinctions are required, especially to foster an empirically discriminating and fruitful

research program. A hypothesis of serial homology among parts of the same body implies that one

must investigate whether a plausible scenario for the evolutionary history of the putative serial

homologues can be empirically supported. Such a question would be moot for copies traced

developmentally (e.g., different cells of the same terminal branch cell type). Theoretical

articulation of this kind for serial homology takes on added significance when we turn to a variety

of frequently discussed morphological examples.

3. Understanding Serial Homology

3.1. Sameness, similarity, and partial homology

A key principle for making sense of serial homology is that homology is a form of sameness

rather than similarity. Homology between two characters means that they are the same kind or type

of character—a leaf, forelimb, or neuron—and that this kind forms a historically traceable

individual like a species (Wagner, 2014). It does not mean that homologous characters are similar

in every respect; they can be highly dissimilar, such as in the well-known transformation of the

14
jaw joint of reptiles into inner ear ossicles of mammals or the transformation of the ancestral

parietal eye in vertebrates to the mammalian pineal gland. This principle applies to special

homology and serial homology alike. One implication of this understanding of sameness is that

judgments of homology are binary or discrete, at least in principle: two characters are either

homologous or not, as opposed to having some degree of homology. Since homologues are the

product of evolution, we would expect there to be cases of incomplete individuation. However,

this is not the same as either mere similarity or partial homology.

Parting with the principle of sameness, some influential architects of the Modern Synthesis

(Mayr, 1969; Simpson, 1961) redefined homology as similarity due to descent from a common

ancestor. It is important to appreciate why a graded version of homology stands in conflict with

the central roles of homology in comparative reasoning.

Homology, understood as sameness, forms equivalence classes of characters. Equivalence

classes are formed by a relation that is reflexive, symmetric, and transitive. Thus, character A is

homologous with itself (reflexive); if A is homologous with B, then B is homologous with A

(symmetric); and if A and B are homologous, and B and C are homologous, then A and C are

homologous (transitive). A theorem of elementary set theory proves that all elements fulfilling

these conditions jointly form a class (i.e., a set that neither overlaps with other such sets nor is a

proper subset). These logical attributes combine with the hierarchical nesting of phylogenetic

descent to give comparative evolutionary reasoning its characteristic form. Within a given clade,

homology of a character applies to the clade as well as every nested sub-clade within it. This is

important for guiding comparative research because only a tiny fraction of species—extant or

extinct—can be studied. When a character is studied in one species in a clade, the results can be

projected across the clade to homologous characters, yielding a rich network of empirical

15
hypotheses with an underlying evolutionary rationale for their likelihood (Harvey & Pagel, 1991).

Work on specific cases should lead to likely inferences applicable to yet unobserved instances

within the same homology class.

Consider the logical implications of understanding homology as similarity for comparative

reasoning (which may not always manifest in scientific practice). Unlike sameness, similarity is

non-transitive: if A is similar to B, and B is similar to C, then A is not necessarily similar to C.

This is because things can be similar and dissimilar in different degrees and respects. For example,

if homology is measured by transcriptomic similarity, there can be overlapping transcriptional

profiles between characters A and B, and B and C, but not necessarily between A and C.

Combining this understanding of homology as similarity with phylogenetic hierarchy, the

homology of a character in a taxonomic group does not necessarily apply across the whole group.

Instead, each potential homologue would have to be compared in a pairwise fashion as to their

similarity or dissimilarity. Carrying out such research would be impractically demanding for

groups with even modest species diversity. It would make comparative research hyper-contextual

and local. It also would stunt empirical hypotheses and block projections that are otherwise enabled

by homology reasoning. Instead of relying on homology to project hypotheses across members of

a clade, and thereby reveal underlying biological commonalities between the characters under

comparison, we would have to know in advance all relevant similarities and dissimilarities in order

to establish homology. In this way, homology-as-similarity appears empirically inert—a stopping

point of investigation that makes homology a blunt if not useless tool in comparative reasoning. It

is perhaps not surprising that there was a relative absence of homology thinking (Ereshefsky, 2012)

in the evolutionary research associated with the Modern Synthesis (see Nuño de la Rosa, this

issue).

16
 Some more nuanced proposals, though not embracing the identification of homology with

mere similarity, likewise counsel abandoning the idea of homology as a binary or discrete relation

in favor of graded notions like “partial homology” (Minelli & Peruffo, 1991; Abouheif, 1999;

Minelli & Fusco, 2013; Moczek, 2014; Roth, 1984; Sattler, 1994; Shubin & Wake, 1996). Partial

homology can be more complex than mere similarity if its assessment is based on multiple

evidential factors—for example, not just similarity, but also phylogenetic relationships,

topological position, and shared developmental mechanisms. The output of such assessments is

nonetheless understood as a graded degree of homology rather than a binary judgment of

homology. It implies that comparative research should be highly contextual and local: “We must

abandon the concept of homology as an all-or-nothing relation. We must rather look, in the

individual cases, for those particular aspects of homology which are recognizable in the context

and at the level of the particular modules we are dealing with” (Minelli, 1998, 344). Although the

rationales for partial homology vary, there are two broad motivations. The first is that evolution is

too contingent and variable for tracing discrete character identities across the vicissitudes of

phylogeny, and therefore binary judgments of homology are unrealistic. The second is that

homologies at different levels of organization can be decoupled from one another. For example,

homologous organs are not necessarily produced only by homologous developmental processes

and might not have homologous components.

It is true that differences between homologous characters can always be found, as should

be expected from the long-term evolution of independent lineages. However, the notion of

homology and definition of shared characters abstracts from biological variation and diversity,

picking out what is invariant or conserved across a given taxonomic group. The identification of a

morphological character does not need to include all of its features and components—if it did, it

17
is not clear how comparative generalizations would be possible. The question is whether

abstracting from these details is problematic, or significantly diminishes the realism of our

conceptual frameworks. We think the question cannot be decided a priori, but must be adjudicated

on the basis of the empirical success of research programs that rely on this conceptual strategy

compared to those that do not.

Shifting from categorical to graded notions may seem appealing to manage biological

variation and complexity, but “degrees” still require specification of the properties that living

systems can have degrees of. In practice, criteria of homology can be in tension (e.g., position

versus special quality) and must be weighted appropriately, yet the existence of different degrees

of evidence for homology does not entail degrees of homology itself. The invocation of graded

notions does not present clear advantages in this situation, whereas it introduces the need to make

sense of how an intransitive version of homology can avoid hyper-contextualism in evolutionary

comparisons. The situation parallels the difficulty that assessing “similarity” requires specification

of which similarities are relevant, a problem that confounded the mathematization of taxonomy

(Sterner & Lidgard, 2014). Principled decisions about which aspects are relevant (and which are

not) would hardly be less demanding than understanding homology as sameness of kind. These

challenges are not insurmountable, but it is not clear that they have been met with a principled

framework of partial homology that is sufficiently articulated to guide the practices of comparative

biology.

As noted, partial homology can be understood as a way of capturing the fact that

homologous characters do not necessarily have all of the same parts or mechanisms. In this sense,

homology is conceived as not “all-or-nothing.” This line of thought presupposes that homologues

would have to be conserved at all levels of organization if homology were to be categorical or

18
discrete, and thus deviations from total conservation imply that homology itself must be graded.

This assumption is unwarranted, however. Homologies are defined for characters identified at a

specific level—a level of organization and level of taxonomic inclusion (see below, Section 5)—

without requiring that all the parts of the character are also homologous. If homology and character

identity are level-specific, then deviations from total conservation across levels need not degrade

the concept of homology itself or make it a matter of degrees. At the same time, violations of the

assumption of total conservation do not entail the opposite extreme of a total independence of

levels of homology. For example, in our model, level-specific character identity is underwritten

by developmental mechanisms that are responsible for the developmental individuation of the

structure. Hence, the independence of character identity only applies to compositional and

developmental mechanisms that are not part of the character identity mechanism itself (DiFrisco

et al., 2020 and see below section 3.2).

Crucially, in our estimation, the idea of partial homology creates difficulties for assessing

how homology at one level of organization influences homology at another level (DiFrisco et al.,

2022). Many of the same genes are reused in different tissues in diverse metazoans. Does this reuse

make all those characters partially homologous? When we consider that key signaling pathways

like Shh and Wnt are involved in many unrelated developmental processes, this would threaten to

make everything partially homologous with everything else. We have returned to our animating

problem: given the pervasiveness of gene sharing in evolution, why is everything not homologous

with everything else? Since Shh signaling is involved in the development of feathers, head midline

structures, and paired—as well as unpaired—appendages of gnathostomes, the potential range of

partial homologies becomes dizzying, if not empirically intractable. Under this conception, the

19
notion of homology loses contrastive resolution, which reduces its usefulness as a tool for

comparative biological research.

3.2. The ontology of a morphological character

To situate our perspective on character identity and homology in contrast to accounts based

on similarity or partial homology, it is useful to briefly review the “ontology” of a morphological

character that gives substance to notions of character identity and character states. In our approach,

a morphological character is a circumscribed part of an organism that consists of at least one but

usually multiple cells and their products. This organismal part develops under the influence of a

character identity mechanism or ChIM (DiFrisco et al., 2020), which endows that part of the body

(as a unit) with the capacity to manifest distinct developmental states from other parts of the same

body. A ChIM is the mechanistic basis for the developmental and evolutionary individuality of a

character. In this model, developmental individuality goes hand-in-hand with evolutionary

individuality because developmental individuality facilitates some degree of variational

independence of an organismal part relative to other such parts. Although there is much more to

the ChIM framework (DiFrisco et al., 2020; DiFrisco et al., 2022; Wagner, 2007, 2014), it is

important to note that a ChIM is necessary for developmental individuality but does not dictate the

particular phenotype or state that a character displays. Character identity and character states (or

the realization of a character) are thus, to some degree, decoupled. This allows character identity

to be a separate matter from similarity between states or phenotypic attributes.

Character state differences can be described in terms of two complementary aspects: (a)

differences in the configuration of sub-parts of the character, and (b) differences of attributes of

the character. A configuration is a set of parts contained within the character that includes their

20
spatial relationships—e.g., zeugopod, stylopod, and autopod arranged along the proximo-distal

axis of the forelimb. Importantly, these parts or configurations of components can be

individualized characters themselves. Character states can also differ with respect to attributes or

properties of the character or its parts. These attributes include size and shape, as well as physical

configuration, such as the presence or absence of parts, the number of parts, or their spatial and

functional relations. A potential source of ambiguity is that the parts of a character are entities (and

potentially characters themselves), but having such-and-such parts is an attribute of the character.

The same ambiguity exists for whole characters: feathers are biological characters, but the

presence or absence of feathers is a descriptive state of the organism’s phenotype. Overall,

differences of character state can include differences in any attribute (e.g., size, shape) or

differences in the kinds and numbers of sub-parts as well as their relations (i.e., their

configuration).

As an example, consider the paired appendages of gnathostome vertebrates (extant jawless

vertebrates do not have paired appendages). All paired appendages of gnathostome vertebrates are

homologous and their development is guided by a conserved set of signaling centers, such as the

apical ectodermal ridge and the zone of polarizing activity, as well as dorsally and ventrally

differentiated ectoderm. In teleost fishes, the paired appendages have a different complement of

skeletal parts compared to limbs. Teleost pectoral fins have four radials, a row of small nodular

distal radials, and a series of dermal fin rays. In contrast, a forelimb has, at least ancestrally and

often early in development, a single proximal stylopodial element (humerus), two zeugopodial

elements (radius and ulna), small wrist elements (carpals), palm elements (metacarpals), and digits

(phalanges). The difference between a pectoral fin and a forelimb is thus due to differences in the

complement of their parts (e.g., limbs have no fin rays) and their attributes (e.g., outward shapes).

21
Hence, fins and limbs are two different character states of the pectoral paired appendage character.

To be more precise, fins and limbs are two classes of character states for this morphological

character because both encompass a variety of states collectively described as fins or limbs. None

of these differences affect the identity of the pectoral paired appendages, which are controlled by

the same ChIMs (see below, Section 6). They only differ in their configurations and attributes, and

there is clear evidence for the continued existence of paired appendages in the lineages leading to

extant tetrapods, notwithstanding that limbs have been lost in some lineages (e.g., snakes,

gymnophiones). Importantly, fins and limbs are not very similar but nonetheless homologous;

there is a traceable history of the presence of paired pectoral appendages in gnathostome lineages.

3.3. Serial homology: character identity mechanisms and positional indexing

The above ontology of morphological characters allows us to distinguish more precise

categories within the broad and inclusive phenomenon of serially repeated structures (Table 1). A

key distinction is between morphological attributes and structures. Attributes of a body part—

including features like size, color, curvature, overlap of parts, and number of parts—are essentially

dependent on the body part they are an attribute of, and have no separable existence (i.e., there is

no colored phenotype without a body part to have that color). By contrast, structures are separable

parts, and they can either be developmentally individualized or not.

22
 Condition of serial feature Term Examples Traceability

Non-separable attribute of Character state Size, color, curvature,

No
body part, no ChIM attribute overlap, number

Zygomatic arch,
Separable structure of a body Character state transverse and spinous Sometimes
part, no ChIM structure process, peripheral (contextual)
vascular elements

Body parts with the same Different particular

ChIM but no positional Replica (copy) cells of the same cell No
indexing type, hairs, leaves

Body parts with the same Mammalian pentadactyl

ChIM, with positional Homomorph digits 2-5, right and left Yes
indexing pectoral appendages

Body parts with

Hox-individuated
phylogenetically related but
Paramorph arthropod segments, Yes
distinct ChIMs and with
vertebrae
positional indexing

Table 1. Categories differentiating the various phenomena associated with serial homology. “Serial homology” is
often loosely applied to each category from character state structures to paramorphs. In our framework, the key
distinction is between paramorphs and non-paramorphs because only the former include distinct individuated
characters.

Primary examples of non-individualized structures include tissue outgrowths in vertebrates, such

as the zygomatic arch (lateral cheekbones), the transverse and spinous processes on vertebrae, and

peripheral vascular elements (Figure 2). These are entities or parts rather than attributes, but they

are structures unlikely to be developmentally individualized by a dedicated ChIM. Nonetheless,

when these structures are stably represented across generations, they are traceable as serial and

special homologues. Often, this is based on their position relative to individualized structures (e.g.,

the spinous process of the C4 vertebra, or the right transverse processes of the thoracic vertebrae).

This is a form of contextually dependent traceability that does not require developmental

individualization. By contrast, character state attributes do not form traceable units or entities at

all. Although one can compare attributes, such as size or color of body parts, such comparisons

necessarily depend on tracing the body part rather than the attribute itself.

23
Figure 2. Examples of morphological structures that are likely not developmentally individualized: (A) The zygomatic
arch is formed by projections from the temporal and zygomatic bone, with variable relative contributions between
species; (B) vertebral processes on thoracic and lumbar vertebrae; and (C) peripheral veins whose formation is due
more to epigenetic factors like shear stress and pressure differences than regulatory specification.

Within the category of developmentally individualized structures or characters, there are

further distinctions relevant to the issue of serial homology. The primary differentiating factors are

whether the serially repeated characters (cell type, tissue, or organ) have the same ChIM and

whether there are stable positional differences that permit a descriptive indexing of the positions

of their different instances. Characters that have the same ChIM but no positional indexing are

simple copies, which we label replicas. These include cells of the same cell type, mammalian body

hairs, and leaves. Mammalian hair of some specific type (e.g., contour hair) is developmentally

individualized vis-à-vis other skin appendages, such as mammary glands, but single hairs are not

individualized vis-à-vis other instances of the same type of hair and lack positional identifiers.

Without differences in identity mechanisms and without stable positional attributes, it is not

meaningful (and may often be practically impossible) to trace such structures across individuals

and species. Cells are only replicas if they belong to a terminal branch in a cell type tree (e.g.,

erythrocytes rather than blood cells) because instances of a non-terminal type can be further

24
distinguished with respect to other terminal types (i.e., blood cells differentiated as lymphocytes

versus myeloid leukocytes). The same principle applies to tissues and organs.

Some developmentally individualized characters do not differ in their underlying ChIM

but possess stable positional differences that allow descriptive indexing: the right or left pectoral

appendage, or digits 2-5 in mammalian pentadactyl limbs. We call these serial structures

homomorphs (Wagner, 2014). The categories of homomorph and replica are morphologically

distinct but mechanistically the same (see Stewart et al. 2019); their instances will vary together

and their capacity for independent variation is limited by the influence of positional factors on

development (e.g., local growth rates or hormonal influences on 2:4 digit ratio).

Developmentally individualized, serially repeated characters with different ChIMs but

derived from the same ancestral ChIM are called paramorphs, parallel to gene paralogues

(Wagner, 2014). 1 Paradigmatic examples are forelimbs and hindlimbs of mammals, which in spite

of their obvious homology are developmentally individualized through the action of the paralogous

transcription factors Tbx4 and Tbx5 (see discussion in Section 6). For tissues and organs, the

presence of distinguishable identity mechanisms automatically implies that there is stable

positional indexing of paramorphs (McKenna et al., 2021). Often, in the absence of mechanistic

knowledge, positional indexing is used as a proxy to name different serial structures and

differentiate them descriptively. Comparisons of variation over a wide taxonomic scope also can

suggest hypotheses about developmental individualization across serial structures based on

correlated variation. However, it is not possible to confirm the status of some serial structures as

homomorphs or paramorphs without the mechanistic study of their development. For example,

1 Others have introduced the term “paramorph” for similar phenomena (Minelli, 2000, 2002). However, this
introduction of the term “paramorph” does not distinguish phenomena we would label as “homomorphs” and
therefore the reader should be aware of the subtle differences in vocabulary.

25
although mammalian digits 2-5 are homomorphs, digit 1 (thumb) is a paramorph of digits 2-5 and

developmentally individualized from them (Stewart et al., 2019), while also being positionally

distinct. In contrast to the replica-homomorph distinction, the categories of homomorph and

paramorph are morphologically the same but mechanistically distinct. Paramorphs are distinct

characters with capacities for independent evolutionary trajectories due to being developmentally

individualized from one another.

The status of some serial structures as homomorphs versus paramorphs can change over

the course of development when multiple ChIMs operate sequentially in the same regions. For

example, in Drosophila, trunk segments along the antero-posterior axis develop from a

segmentation cascade controlled by the gap gene network, pair-rule network, segment polarity

network, and Hox genes that differentiate segmental identities. Prior to the action of Hox genes,

the trunk segment domains are homomorphs, having the same ChIM—the segment polarity

network (Chipman, 2015; DiFrisco & Wagner, 2022; Lev & Chipman, 2021; Lev et al., 2022;

Wagner, 2014; see Figure 3). 2 They are positionally indexed and traceable only in a weak sense

(e.g., “anteriormost segment”). In comparing myriapods, which have more segments and more

segment number diversity than insects, it might not be meaningful to ask (at this stage) which

segments have been added and which are the same. However, after Hox genes activate cascades

of quasi-independent regionally-specific developmental processes, the segments become

paramorphs (Hughes & Kaufman, 2002). In arthropods generally, segments are serially

homologous as segments, but also have additional identities, such as thoracic segments or a T2

2 Lev and colleagues (2022) have recently used the ChIN/ChIM concept to assess serial homology between
arthropod segments, finding that pre-gnathal segments of the head are not serially homologous to trunk segments
due to their not having the same ChIM (the segment polarity network). This insight is then used to support a novel
hypothesis about the evolutionary origin of head segments.

26
segment. In this case, paramorphs also are positionally indexed and strongly traceable; they can

be meaningfully compared across species that have (or lose) them.

A feature unique to paramorphs is that they form hierarchies or trees of character identities.

Because of this, it is necessary to specify the “levels” at which the characters are homologous or

non-homologous when comparing paramorphs. For example, a segment must be distinguished

from a thoracic segment and a vertebra must be distinguished from a thoracic vertebra. This feature

also has an important implication: evolutionary novelty should also be understood in a hierarchical

manner (Love 2006).

4. Levels of Organization, Paramorph Hierarchies, and Evolutionary Novelty

Homology is often acknowledged to be a “hierarchical” relationship (Hall, 1994), but there

are different senses in which this is true. One sense refers to levels of organization: robust part-

whole relationships that capture wide-ranging biological regularities (Brooks et al., 2021). For

morphological characters, the primary levels of anatomical organization are cell types, tissues,

organs, and whole body plans. These different characters have level-specific kinds of character

identity mechanisms (ChIMs). For cell types, these are primarily core regulatory complexes of

transcription factors; tissue ChIMs include extracellular matrix and cell–cell signaling networks

that maintain appropriate distributions of cell types; and organ ChIMs are interdependent signaling

centers that control the distribution of tissues in a region (DiFrisco et al., 2020). Body plan identity

mechanisms, or ChIMs for integrated body plans, are similarly composed of interdependent

signaling centers, but the signals cross embryonic structures that are not developmentally

individualized (DiFrisco & Wagner, 2022). The ChIM model revises and qualifies a common

contention about homology and levels—namely, that homology between characters at one level is

27
often independent of homology at another level. While acknowledging this fact, the central

hypothesis of the ChIM model is that character identity is linked to the identity of specific

underlying (i.e., lower-level) mechanisms, which are subsets of the total developmental

mechanism for building a body part in a particular species (DiFrisco, 2021).

A different sense of hierarchy pertains to phylogenetic trees of species or characters, where

the branching points define different levels of taxonomic inclusion or specification. With special

homology, the branching pattern of characters coincides with the speciation events of the lineages

having those characters. When serial homologues are treated as single phylogenetic characters

(e.g., the vertebral column; see Section 3.2), the situation is the same. However, with paramorphs

the pattern is more intricate because multiple terminal branches of a character tree coexist within

the same organism. Thus, in making a statement about serial homology between paramorphs, it is

necessary to specify the level of inclusion at which that relationship holds. This can be captured in

the language of “levels”: the vertebrae C4 and T3 are serially homologous (paramorphic) at the

level of vertebrae but not cervical vertebrae, whereas C4 and C5 are serially homologous

(homomorphic) at the level of vertebrae and cervical vertebrae. The hierarchical or tree-like

relationship among paramorphs entails that they can be more or less closely related, though this

does not imply that there are continuous degrees of homology (e.g., insect segment T2 is not “X%

more homologous” to T3 than to abdominal segments). Rather, the relatedness of paramorphs is

determined by the number of (discrete) branches or levels separating them on the relevant character

tree.

Paramorph hierarchies can be found at multiple biological levels of organization, from

paralogous genes to cell type families, tissue paramorphs, and organ paramorphs. At still higher

levels of organization, sexual homologies between male and female genitalia are paramorphic

28
(Pavlicev et al., 2022), as are caste types in insect colonies (Abouheif, 2021). Given that paramorph

hierarchies comprise characters that are developmentally individualized from one another, their

existence implies a corresponding hierarchy for the associated ChIMs. The ChIM for

homomorphic arthropod segments is the segment polarity network (SPN), but the ChIM for the

prothorax includes the SPN plus Sex combs reduced (Scr), which individualizes the T1 segment

(Figure 3). In this ChIM hierarchy, the SPN is at a more basal node, whereas the ChIM that

includes the SPN and Scr is a terminal branch. In the same way, “arthropod segment” is a more

basal character and “prothorax” is one among several terminal branches—a developmentally

individualized paramorph of arthropod segments.

This understanding of paramorph hierarchies has an important implication for the

comparative study of serial homology and evolutionary novelty. If we conceptualize novelties as

characters that are not serially or specially homologous to an existing character (Müller & Wagner,

1991), and paramorphy is a hierarchical relationship, then the status of some paramorph as a

novelty versus a serial homologue is also hierarchical. 3 That is, the difference between novelty

and serial homology is only well-defined relative to particular levels in a paramorph hierarchy.

Thus, when we ask whether the first digit in mammals (the thumb) is a novelty or a serial

homologue, the answer is both—relative to different levels of specification of the digit paramorph

hierarchy. As a digit, it is serially homologous to digits 2–5, but since it is developmentally

individualized from them, it is also a novel character. By contrast, a supernumerary 6th digit would

be serially homologous (homomorphic) to digits 2–5, but not a novelty if it is not individualized

relative to digits 2-5.

3 Our conceptualization of serial homology requires a slight modification of the earlier definition of evolutionary
novelty as non-homology (Müller & Wagner, 1991). An evolutionary novelty that arises from duplicated characters
introduces a new level in the hierarchy of serial homologues, but it can be serially homologous with respect to its
components at a deeper level of the hierarchy.

29
Figure 3. A simplified illustration of a character tree for arthropod segments, shown here in Drosophila. The Segment
Polarity Network (SPN) is proposed as the ChIM shared by all trunk segments, with individual segmental identities
differentiating from these under the influence of Hox identity genes (Sex combs reduced for prothorax, Antennapedia
for mesothorax, Ultrabithorax for metathorax, and so on). The segments and the corresponding ChIMs both form a
paramorph hierarchy.

This hierarchical understanding of novelty differs from a common but mistaken assumption

that genuine novelty is absolute and requires non-homology with respect to all existing characters

and their components: “Do new anatomical structures arise de novo, or do they evolve from pre-

existing structures?” (Shubin et al., 2009, 818). Given the pervasiveness of gene co-option and re-

use of existing structures, it is not clear how many multicellular characters would satisfy this strict

understanding of novelty; in other words, we would expect that almost all novelty evolves from

pre-existing structures and resources through individuation. The theoretical articulation

summarized in Table 1 takes into account the hierarchical dimensions of homology and (therefore)

novelty. The question is not whether a novelty evolved from pre-existing structures; new

characters arise from pre-existing structures just as new species evolve from pre-existing species

by descent with modification. The relevant question is how. This mechanistic question involves

not only discerning whether the putatively new character is developmentally individualized at a

particular hierarchical level, but also assessing the status of the pre-existing structures and whether

30
they were replicas, homomorphs, or paramorphs at a particular hierarchical level. And,

importantly, “hierarchical level” needs to be understood in both space and time. The relevant pre-

existing structures from which a novelty evolved might be at a different spatial level of

organization (e.g., gene regulatory networks) or at a different temporal juncture in the ontogenetic

origination of a character (e.g., iterated morphogenetic fields).

We have rediscovered the corollary of our original question: if everything is (partially)

homologous with everything else, then nothing is an evolutionary novelty. Our hierarchical

account of serial homology and novelty escapes these untoward implications, grounds the basic

vocabulary and knowledge base of comparative biology, and facilitates new inquiry by permitting

a more fine-grained mapping of diverse trait-types discussed by biologists. Thus far, our theoretical

articulation has been relatively abstract with a few scattered illustrations from well-known

examples. These ideas about levels of organization, paramorph hierarchies, and evolutionary

novelty can be clarified and elaborated by sustained application to more detailed empirical case

studies.

5. Insect Wing Serial Homologues

Wings represent a major innovation of the insect body plan that likely contributed to the

exceptional success and diversity of winged insects (Grimaldi & Engel, 2005). Wings are thought

to be monophyletic, but their evolutionary origin has proved a difficult problem in part because of

the lack of transitional forms in the fossil record between non-winged and winged insects

(Grimaldi & Engel, 2005). This has encouraged studies of potential serial homologues of wings

(e.g., on wingless segments) as a source of clues about wing origins (Clark-Hachtel & Tomoyasu,

2016). It is now known that many genes from the canonical wing gene regulatory network that

31
were revealed from earlier studies of Drosophila—including the “master wing genes” vestigial

(vg), apterous (ap) and wingless (wg)—are expressed in multiple structures along the insect body

wall, including in ancestrally wingless insects, such as the bristletail Pedetontus unimaculatus

(Niwa et al., 2010). The observation of shared key regulators suggests that either the wing

originated by co-option or it possesses some underlying serial homology with other body wall

structures. In the latter case, wings would be derived branches in a paramorph hierarchy that

includes other body wall outgrowths.

Historically there have been two competing serial homology hypotheses for wing origins

(Figure 4). The tergal origin hypothesis (also known as the paranotal hypothesis) holds that wings

are lateral extensions of the thoracic tergum, or dorsal body wall. The pleural origin hypothesis

holds that wings originated as outgrowths of the pleuron, or lateral body wall. More recently, some

authors have gathered compelling evidence for a dual origin hypothesis, which holds that wings

arose from fusion of both tergal and pleural source tissues (Clark-Hachtel & Tomoyasu, 2016;

Niwa et al., 2010). These hypotheses have different implications for serial homology between

wings and other body wall structures. If the tergal origin hypothesis is correct, then wings could

be serially homologous to specialized tergal outgrowths, such as beetle horns and the T1 carinated

margin. If the pleural origin hypothesis is correct, then wings would likely be serially homologous

with pleural outgrowths, such as lateral lobes or lateral sclerites. Under the dual origin hypothesis,

however, it is only the tergal source tissues of the wing that would be serially homologous to the

tergal outgrowths, while the pleural source tissues would be serially homologous to pleural

structures. By hypothesis, the tergal source tissues would be responsible for the flat wing blade,

whereas pleural plates with exite outgrowths would have provided a pre-existing articulation

mechanism and muscle attachment for the wing (Clark-Hachtel & Tomoyasu, 2016).

32
Figure 4. A basal wingless insect from the suborder Monura (order Archaeognatha), illustrating proposed source
tissues for wing origins. Modified and redrawn after (Kuklová-Peck, 1987).

This example highlights the need to clearly specify homology relations in terms of the

appropriate level of organization and paramorph hierarchy. If the dual origin hypothesis is right,

the wing is only “partially” serially homologous with tergal structures (and with pleural structures).

This does not mean that the homology relation itself is graded and partial, but rather that part of

one character (wing) is homologous to another character (tergal outgrowth). Likewise, the

conundrum of wing serial homology can be understood as a problem of determining what is the

basal character in the character phylogeny that includes wings. One recent theory combining

elements of pleural and tergal theories holds that insect wings derive from crustacean exites or

lateral lobes extending from proximal leg segments, which were incorporated into the body wall

in insects (Bruce & Patel, 2020). With exites as the basal character in the paramorph hierarchy,

many structures considered wing serial homologues, such as the larval gin trap and lateral lobes,

would be exite paramorphs (Bruce & Patel, 2020). Within that inclusive underlying homology,

however, these structures do not necessarily share more derived and specific character identities—

i.e., they are not wings. Thus, wings could still be novel characters, in the sense of novel

individualized branches in the character tree, relative to their branching from ancestral exite-like

structures, while still being serially homologous with other structures derived from exites.

33
 Similarly, a recent dual-origin theory holds that wings and other body wall outgrowths are

derived products of an ancestral developmental network that regulates three-dimensional

patterning of cuticularized bilayered epithelia (Fisher et al., 2021). In this interpretation, the basal

character would be “bilayered epithelial outgrowths” and the associated ChIM would be the

ancestral regulatory network that patterns all such structures. Derived characters such as insect

wings, the diverse gills of crustaceans, mayflies, and chelicerates, treehopper helmets, abdominal

gin traps in beetle pupae, and other marginal outgrowths would be serially homologous (and likely

all paramorphic) as bilayered epithelial outgrowths, but would not necessarily share more derived

character identities (e.g., wings) (Fisher et al., 2021). An important question for evaluating this

proposal is whether the proposed basal “character” is a developmentally individualized structure

(i.e., a character), rather than a character state or structure of some other character (see Section 3.2;

Table 1). A trait description like “bilayered epithelial outgrowth” would need to be traceable to a

monophyletic evolutionary origin. In addition, it is not clear that all wing serial homologues use

this module. Dipteran halteres are paramorphic with forewings but are not bilayered epithelia and

therefore cannot have a ChIM for that structure. In this case, “bilayered epithelial outgrowth”

might not be the most appropriate description for the basal character in the paramorph hierarchy.

These hypotheses illustrate how simple pairwise comparisons of structures will often be

inadequate to resolve serial homology relationships or determine whether the concept of serial

homology is applicable. For example, to determine whether treehopper helmets are novel or

serially homologous to wings (Kudla et al., 2022; Prud'homme et al., 2011), these structures must

be compared at the appropriate organizational level and, if they are homologous, assessed for

whether their homology is embedded in an evolutionary character tree or paramorph hierarchy. It

is possible that helmets are serially homologous to wings because they belong to an exite-derived

34
character family or the character family of some other basal character without helmets being wings.

Indeed, this seems to be the case (DiFrisco et al., 2022; Fisher et al., 2021). Thus, despite the name

“wing serial homologues,” which is commonly used to describe research in this area, wings are

evidently too derived to serve as the most evolutionarily informative character for defining the

character tree of pleural and tergal outgrowths. The name reflects the order of scientific discovery,

which started with developmental-genetic studies of Drosophila forewings, rather than the order

of evolutionary change. Accordingly, the more informative terminology would refer to the node

in the paramorph hierarchy that connects wings and helmets, if it exists: for instance, exite serial

homologues.

In most cases, knowledge of underlying ChIMs and transformation series will be

incomplete, as it is with insect wing origins. In this situation, an advantage of the hierarchical view

is that serial homology and novelty between any two structures can be established without a

complete mapping of the paramorph hierarchy. As we saw, the non-homology of treehopper

helmets and wings, and the status of the helmet as a novelty, is consistent with their serial

homology as potential exite paramorphs. By contrast, if novelty was a matter of non-homology

with any and all existing characters, it would not only be exceedingly rare as a natural category

(see Section 4) but also something that could only be established empirically after the origins and

transformation series are fully known. Our proposed framework not only avoids this discouraging

implication but also connects the status of a character as a novelty or homology to underlying

developmental differences that are informative about the capacities of a character for quasi-

independent variation and evolutionary change. In this way, a hierarchical perspective on serial

homology and novelty is a more useful and inferentially rich framework that maximizes the

contrastive resolution of incomplete data.

35
6. Paired Gnathostome Appendage Serial Homology

In recent years, Rui Diogo and collaborators have mounted a challenge to the concept of

serial homology (e.g., Miyashita & Diogo, 2016). This challenge concentrates on two prominent

hypotheses of serial homology in vertebrate anatomy: the mandibular arch as a serial homologue

of the pharyngeal arches (“gill arches”), and the serial homology of forelimbs and hindlimbs. Of

these two hypotheses, a particular popular scenario has been challenged for the question of the

origin of the mandibular arch. However, this does not affect the concept of serial homology as

such. In contrast, the case of tetrapod forelimbs and hindlimbs has been advanced as a direct

challenge to the concept of serial homology and therefore we focus on it here.

Diogo’s challenge to the serial homology of gnathostome paired appendages has two

aspects, one pertaining to general issues beyond the specific example and another pertaining to the

empirical evidence relevant to claims of serial homology between the pectoral and pelvic paired

appendages. Regarding the first aspect, there are three general issues: the relationship between

homology and similarity, the co-option of molecular mechanisms, and the relativity of homology

claims to a level of organization. We have already discussed the first at length (Section 3.1) and

therefore only reiterate that arguments against serial homology based on a lack of similarity fail

because homology is not a relationship of similarity but rather identity.

The second issue concerns how to assess the similarity of developmental mechanisms

among parts as evidence for or against serial homology. Miyashita and Diogo (2016) correctly

point out that some of the mechanisms underlying paired appendage bud development are shared

with other body appendages, and these have not been proposed as serial homologues to paired fins

and limbs (e.g., barbs in catfish, medial fins, and the amniote phallus). These mechanisms include

36
the role of Shh from the zone of polarizing activity in causing anterior-posterior polarity and

sustaining appendage bud outgrowth in conjunction with FGFs from the apical ectodermal ridge

(Zeller et al., 2009). The broad distribution of this regulatory module among diverse body parts

suggests it is a generic appendage outgrowth module rather than a ChIM. As we have argued

elsewhere (DiFrisco et al., 2022), mechanistic similarity in the development of different body parts

can occur for several reasons. The similarity could be due to (i) “plesiomorphic” features that

existed in body regions where the new structures are formed, (ii) co-option, or (iii) a transfer of

character identity (e.g., duplication, or homeotic transformation). Co-option refers to the situation

where parts of an existing gene regulatory network are recruited in order to perform a

developmental function in a different region, such as appendage outgrowth or epithelial folding

(DiFrisco et al., 2022). Character duplication happens when the ChIM of one character is recruited

to be expressed in another part of the body and thus re-instantiates the character identity in a novel

location (DiFrisco et al., 2022). The latter would create a serial homologue by character

duplication. Consequently, the question of whether shared molecular mechanisms support the

hypothesis of serial homology between the pectoral and pelvic paired appendages depends on the

identification of a ChIM for paired appendages.

Paired appendage development is more complex than the Shh/Gli3/Fgf signaling module.

It also includes interactions between lateral plate mesoderm and both dorsal and ventral

compartments of the ectoderm, which help to endow the appendage bud with dorsal-ventral

polarity (Delgado & Torres, 2017). This does not exist in medial fins. To our knowledge, a putative

ChIM for paired appendage identity that distinguishes paired appendages from other body

outgrowths has not been established. By implication, no ChIM has been shown to be shared

between pectoral and pelvic appendages. Hence, assessing the claim of serial homology for paired

37
appendages in vertebrates is an outstanding research question rather than something already

established.

The third general issue is the relativity of homology claims to a level of organization. Given

that fins and limbs are complex, hierarchically structured body parts, it is necessary to explicitly

state which level or body part an assertion or denial of serial homology refers to. As our theoretical

articulation suggests (Section 4), it is possible that parts of limbs can be serially homologous,

whereas the pelvic and pectoral appendages themselves may not be serially homologous. For

example, body hair on the torso and legs can be the same, even though these body regions are not

serially homologous. This point is crucial to account for the otherwise puzzling fact that, after the

fin-limb transition, the forelimb and the hindlimb became more similar than the corresponding

pectoral and pelvic fins were in the stem lineage of tetrapods.

Another aspect of this issue is whether the autopod (hand or feet) and the mesopodium

(wrist or ankle) of the tetrapod limb are novel. There is some developmental evidence suggesting

that the distal parts of the fin are already distinguished from the rest of the fin, reminiscent of

autopodium development (Letelier et al., 2021; Metscher et al., 2005; Nakamura et al., 2016).

Some interpret this as evidence that the autopod is not novel (Letelier et al., 2021), though there

remain differences that support an interpretation of its novelty (Wagner & Chiu, 2001). If an

autopodial developmental module already exists in pelvic and pectoral fins, then this would

suggest that they are serially homologous and the simultaneous transformation of the two

appendages would be plausible due to shared developmental regulatory mechanisms. However,

even if we assume that pectoral and pelvic fins are not serially homologous and the autopod is a

novelty, as Miyashito and Diogo (2016) do, the question remains whether the “autopod” units—

and not the appendage as a whole—are serially homologous. The similarity of gene regulatory

38
networks underlying autopod identity would support such a conclusion (Petit et al., 2017), as well

as the near simultaneous origination of autopodial structures in the forelimbs and hindlimbs of

stem tetrapods (Clack, 2002). Hence, serial homology may apply to some components of the paired

appendages of tetrapods, regardless of whether the appendages themselves are serially

homologous.

The specific historical scenario for the origin of paired appendages remains contested.

There is agreement that the first paired appendages with endoskeleton were found in a group of

ostracoderms within the gnathostome stem lineage ~430 mya (Coates, 2003; Janvier, 1996, 2001;

Wilson et al., 2007). Placodermi, an extinct gnathostome crown-group clade, is recognized to be

the oldest group with two paired appendages. At face value, this is compatible with a duplication

event; for example, the pelvic paired appendage of crown-group gnathostomes could be a duplicate

of the pectoral appendage of ostracoderms. The fossil remains of pelvic appendages in basal

placoderms do not show much similarity with fossilized pectoral fins (Coates, 2003). However,

we do not have information about what the pelvic fins looked like shortly after their origin. Thus,

we do not know whether the morphological differences between pectoral and pelvic fins observed

in fossil remains are due to adaptive divergence or independent origin (i.e., non-homology).

However, we do know that the pelvic fin was initially much smaller than the pectoral fin and also

was repeatedly lost, as evidenced in specimens that lack pelvic fins but still have a pelvic girdle

(Zhu et al., 2012). It is thus not clear whether the pelvic fin performed an important function or

was just a rudimentary organ, one that may not have had a specific function immediately after

origination. We therefore conclude that existing paleontological evidence cannot resolve the issue

of whether paired appendages are serial homologues.

39
 Despite the inability of paleontological evidence to decide the question of the origin of

pectoral and pelvic paired appendages, existing developmental evidence provides a plausible

scenario that supports serial homology of pectoral and pelvic fins. There are two developmental

differences that distinguish the body wall of lamprey, a cyclostome which never had paired

appendages, and extant gnathostomes with two pairs of fins or limbs. First, the posterior lateral

plate mesoderm of lamprey is not differentiated into splanchnic (visceral) and somatic (body wall)

layers, and the dorso-ventral compartment boundary in the ectoderm is located dorsally of the

lateral plate mesoderm (i.e., it faces the somites and not the lateral plate) (Tanaka & Onimaru,

2012). In contrast, the gnathostome paired fins and limbs derive from the somatic layer of the

lateral plate mesoderm and the identity of the somatic layer is considered essential for the

development of paired appendage bud mesoderm. Also, the appendage bud develops at the place

where the dorso-ventral compartment boundary is opposite the lateral plate. This boundary

becomes the apical ectodermal ridge, an essential signaling center for appendage bud development

and outgrowth. These facts and the plausible assumption that the differentiation of the lateral plate

mesoderm in stem gnathostomes proceeded from anterior to posterior is consistent with the fact

that paired appendages first appeared in the pectoral region and only later on the posterior body.

Importantly, this scenario does not address the question of why there are only two pairs of

appendages in gnathostomes.

Another interesting observation related to paired appendages is that the pectoral and the

pelvic appendages became more similar during the fin-limb transition than they were as fins

(Miyashita & Diogo, 2016; Roth, 1988). The distal parts of the limb (i.e., elements of the autopod)

are anatomically more similar between forelimb and hindlimb than the corresponding parts of the

pectoral and pelvic fins (Miyashita & Diogo, 2016). Additionally, the gene regulatory networks in

40
the distal parts of both forelimb and hindlimb are more similar to each other than what is found in

their proximal parts (Petit et al., 2017). Miyashito and Diogo (2016) interpret these facts as

showing that the forelimb and hindlimb are similar due to evolutionary convergence and therefore

should not be considered serial homologues. However, this conclusion is questionable for two

reasons. First, it depends on the notion that serial homology is similarity due to derivation from an

ancestral unduplicated character. As discussed (Section 3.1), homology is not a relation of

similarity but rather identity—in spite of differences in form and function. Second, the finding that

serially homologous characters can become more similar over evolutionary time relative to their

initial appearance should not be surprising. This would be expected if the characters are under

selection for a common function, such as locomotion. Moreover, if serial homologues possess

shared underlying developmental mechanisms, evolution in these mechanisms can result in

correlations between serial homologues beyond their initial similarity.

More importantly, even if numerous structures commonly regarded as serial homologues

do not exhibit an evolutionary trend of initial similarity followed by later divergence, as Siomava

and colleagues (2020) show, the precise pattern of these trajectories does not establish or

undermine their status as serial homologues. In fact, most of the traits used to measure similarity

in this study are character states—shapes, sizes, and numbers of parts—rather than character

identities. The expectation that serial homologues should conform to a particular evolutionary

pattern (e.g., ancestral similarity [isomerism], derived dissimilarity [anisomerism]) only makes

sense if homology is understood as similarity due to common descent (Mayr, 1969, see above,

Section 3.1), rather than as a question of character identity.

41
7. The Value of Theoretical Articulation: Beyond Serial Homology

Toward the end of the 19th century, Ray Lankester reflected on the impact of Owen’s

(1843, 1848) theoretical articulation in relation to homology.

It is not easy to exaggerate the service rendered by Owen to the study of zoology by the introduction
of this apparently small piece of verbal mechanism; …And, though the conceptions of “archetypal
morphology,” to which it had reference, are now abandoned in favour of a genetic morphology, yet
we should remember, in estimating the value of this and of other speculations which have given
place to new views in the history of science” (Lankester, 1888, 808).

The ability to simultaneously recognize Owen’s achievement in the midst of his error is significant.

This tells us something important about good working concepts in the sciences. One of their chief

assets is fruitfulness, the capacity to engender new forms of scientific inference or research,

especially in the context of new empirical discoveries that disrupt current biological understanding.

Concepts can accomplish this despite not being fully adequate or having aspects that require

further modification. Owen’s categories of special, serial, and general homology exemplify this

fruitfulness and provide an emulable historical case of theoretical articulation in comparative

biology.

The pervasiveness of gene sharing in evolution (Piatigorsky, 2007) and the increasing

discovery of so-called deep homology across the tree of life (Shubin et al., 1997) disrupted

expectations among biologists.

The remarkable similarity of the genetic regulation of development in distant organisms has
heralded a new conception of evolution. It was a big surprise when evolutionary conservation of
the Krebs cycle, the genetic code, and classes of structural proteins was extended to regulation of
development. The diversity of organisms had fooled everyone into thinking that the evolution of
completely different regulatory processes, or at least completely different uses of the same genes,
was likely to be responsible for evolutionary change (Scott, 2000, 33).

This disruption generates an unresolved conceptual problem: Given the fact of common descent,

why is everything not homologous with everything else? Since most characters exhibit genetic and

developmental commonality―what some have labeled “partial” homology―the distinction

42
between homology and non-homology seems threatened. This threat calls into question a large

range of comparative morphological knowledge and makes it difficult to ascertain the origin of

evolutionary novelties, if there even is such a thing.

If we apply the criterion of fruitfulness to the notion of “homology as similarity,” then we

quickly see the difficulty (Section 3). It is a stopping point of inquiry; it does not enable further

investigation or hypothesis generation. Likewise, “partial homology,” while ostensibly oriented at

biological realism, also diminishes the fruitfulness of general comparative reasoning and

hypothesis-generation based on homology. To address the problem that arises from the conceptual

disruption caused by empirical discoveries at the center of contemporary evo-devo, a different

form of theoretical articulation is necessary. Herein we used successful patterns of reasoning that

have coalesced in talking about homology at the level of gene sequence in combination with

important desiderata (e.g., homologues must form equivalence classes and therefore involve

“sameness”) to offer a hierarchical, phylogenetic characterization of serial homology in terms of

a five-fold theoretical classification (Section 3.3, Table 1). Our proposal was grounded in earlier

work that elucidates what it means for a part to have identity as a biological character (DiFrisco et

al., 2020) and is amply illustrated by findings over the past decade in studies of the origin and

evolution of cell types (Arendt et al., 2016). This theoretical elaboration meets the criterion of

fruitfulness and offers biologists a richer and more precise vocabulary, opening new avenues of

inquiry on recalcitrant questions about insect wing serial homologues and vertebrate paired

appendages. More generally, it is especially helpful for comparative analyses focused on the origin

of evolutionary novelties through concepts such as “paramorph hierarchies.” It demonstrates that

the identification of novelty as such does not demand a complete knowledge of precursors or a

43
step-by-step transformation series. Thus, this understanding of novelty can fruitfully guide

research as it proceeds rather than being only applicable when empirical matters are fully resolved.

In recognition of the ongoing task of theoretical articulation, it is important to end on open

questions where further empirical and conceptual work are needed. For example, the concept of a

“character swarm” picks out a phenomenon that appears to fall between homomorphs and

paramorphs (Wagner, 2014). Feather tracts appear to be paramorphs—serially homologous

features that are sufficiently individualized to take on distinct identities and functions in the head,

wing, and tail locations on birds. However, they retain a tendency to vary as a unit independent of

their positional indexing and thereby evolve in concert, which seems to clash with their purported

evolutionary individuality. This is displayed in “transgressive variation” when a unique feature of

one feather tract, such as coloration in wing feathers used in mating displays, also appears in low

frequency in other feather tracts where this function is irrelevant. Feather tracts appear to hover

between distinct, individualized character identities and a single, globally coordinated character

with multiple states.

Empirical puzzles of this type in morphology do not constitute a counterexample to the

proposed theoretical articulation surrounding serial homology offered here. As we have argued

elsewhere (DiFrisco et al., 2020), our investigative methodology is a form of conceptual

engineering (Brigandt, 2011; Wilson, 2018; Wimsatt, 2007), which is applied to evolved, complex

systems whose history is often inaccessible. Therefore, variation and exceptions are expected. To

understand these systems, scientists must adopt strategies and heuristics that yield partial models

and incomplete perspectives. Conceptual engineering concentrates on fashioning concepts with

the goal of generating theoretical understanding that exhibits fruitfulness for future research. Our

proposed re-engineering of serial homology works toward this goal by accounting for why

44
everything is not homologous to everything else and why there is no contradiction in saying that

genuine novelty has evolved at different levels of organization out of a variety of precursors in the

history of life.

45
Bibliography

Abouheif, E. (1999). Establishing homology criteria for regulatory gene networks: prospects and
challenges. In G. R. Bock & G. Cardew (Eds.), Homology (pp. 207–225). Chichester: John
Wiley & Sons.

Abouheif, E. (2021). Ant caste evo-devo: it’s not all about size. Trends in Ecology & Evolution,
36, 668–670.

Arendt, D. (2008). The evolution of cell types in animals: emerging principles from molecular
studies. Nature Reviews Genetics, 9, 868–882.

Arendt, D., Bertucci, P. Y., Achim, K., & Musser, J. M. (2019). Evolution of neuronal types and
families. Current Opinion in Neurobiology, 56, 144–152.

Arendt, D., Musser, J. M., Baker, C. V. H., Bergman, A., Cepko, C., Erwin, D. H., . . . Wagner, G.
P. (2016). The origin and evolution of cell types. Nature Reviews Genetics, 17, 744–757.

Blackburn, D. G. (1991). Evolutionary origins of the mammary gland. Mammal Review, 21, 81–
96.

Brigandt, I. (2011). Natural kinds and concepts: A pragmatist and methodologically naturalistic
account. In J. Knowles & H. Rydenfelt (Eds.), Pragmatism, Science and Naturalism (pp.
171–196). Frankfurt am Main: Peter Lang.

Brigandt, I., & Love, A. C. (2012). Conceptualizing evolutionary novelty: Moving beyond
definitional debates. Journal of Experimental Zoology (Mol Dev Evol), 318B, 417–427.

Brooks, D. S., DiFrisco, J., & Wimsatt, W. C. (Eds.). (2021). Levels of Organization in the
Biological Sciences. Cambridge: MIT Press.

Bruce, H. S., & Patel, N. H. (2020). Knockout of crustacean leg patterning genes suggests that
insect wings and body walls evolved from ancient leg segments. Nature Ecology &
Evolution, 4, 1703–1712.

Buchmann, K. (2014). Evolution of Innate Immunity: Clues from Invertebrates via Fish to
Mammals. Front Immunol, 5, 459.

Chipman, A. D. (2015). Hexapoda: Comparative Aspects of Early Development. In A. Wanninger

(Ed.), Evolutionary Developmental Biology of Invertebrates 5: Ecdysozoa III: Hexapoda
(pp. 93–110). Vienna: Springer Vienna.

Clack, J. A. (2002). Gaining Ground: The Origin and Evolution of Tetrapods. Bloomington, IN:
Indiana University Press.

Clark-Hachtel, C. M., & Tomoyasu, Y. (2016). Exploring the origin of insect wings from an evo-
devo perspective. Current Opinion in Insect Science, 13, 77–85.

46
Coates, M. I. (2003). The Evolution of Paired Fins. Theory in Biosciences, 122, 266–287.

de Beer, G. (1971). Homology, an unsolved problem. London: Oxford University Press.

de Pinna, M. C. C. (1991). Concepts and tests of homology in the cladistic paradigm. Cladistics,
7, 367–394.

Delgado, I., & Torres, M. (2017). Coordination of limb development by crosstalk among axial
patterning pathways. Developmental Biology, 429, 382–386.

DiFrisco, J. (2021). Toward a theory of homology: development and the de-coupling of

morphological and molecular evolution. British Journal for the Philosophy of Science.

DiFrisco, J., Love, A. C., & Wagner, G. P. (2020). Character identity mechanisms: a conceptual
model for comparative-mechanistic biology. Biology & Philosophy, 35, 44.

DiFrisco, J., & Wagner, G. P. (2022). Body Plan Identity: A Mechanistic Model. Evolutionary
Biology, 49, 123–141.

DiFrisco, J., Wagner, G. P., & Love, A. C. (2022). Reframing research on evolutionary novelty
and co-option: character identity mechanisms versus deep homology. Seminars in Cell &
Developmental Biology.

Ereshefsky, M. (2012). Homology thinking. Biology & Philosophy, 27, 381–400.

Fisher, C. R., Kratovil, J. D., Angelini, D. R., & Jockusch, E. L. (2021). Out from under the wing:
reconceptualizing the insect wing gene regulatory network as a versatile, general module
for body-wall lobes in arthropods. Proceedings of the Royal Society B: Biological Sciences,
288, 20211808.

Fusco, G. (2022). Serial homology. Biological Theory, 17, 114–119.

Ganley, A. R. D., & Kobayashi, T. (2007). Highly efficient concerted evolution in the ribosomal
DNA repeats: Total rDNA repeat variation revealed by whole-genome shotgun sequence
data. Genome Research, 17, 184–191.

Gehring, W. J. (1998). Master Control Genes in Development and Evolution: The Homeobox
Story. New Haven, CT: Yale University Press.

Goodwin, B. C. (1994). Homology, development, and heredity. In B. K. Hall (Ed.), Homology:

The Hierarchical Basis of Comparative Biology (pp. 230–249). San Diego: Academic
Press.

Grimaldi, D., & Engel, M. S. (2005). Evolution of the Insects. Cambridge: Cambridge University
Press.

Haag, E. S. (2014). The same but different: worms reveal the pervasiveness of developmental
system drift. PLoS Genet, 10, e1004150.

47
Haag, E. S., & True, J. R. (2021). Developmental system drift. In L. Nuño de la Rosa & G. B.
Müller (Eds.), Evolutionary Developmental Biology: A Reference Guide (pp. 99–110).
Cham: Springer International Publishing.

Hall, B. K. (2003). Descent with modification: the unity underlying homology and homoplasy as
seen through an analysis of development and evolution. Biological Reviews of the
Cambridge Philosophical Society, 78, 409–433.

Hall, B. K. (Ed.) (1994). Homology: The Hierarchical Basis of Comparative Biology: Academic
Press.

Hartenstein, V., & Martinez, P. (2019). Phagocytosis in cellular defense and nutrition: a food-
centered approach to the evolution of macrophages. Cell and Tissue Research, 377, 527–
547.

Harvey, P. H., & Pagel, M. D. (1991). The Comparative Method in Evolutionary Biology. Oxford
& NY: Oxford University Press.

Hughes, C. L., & Kaufman, T. C. (2002). Exploring the myriapod body plan: expression patterns
of the ten Hox genes in a centipede. Development, 129, 1225–1238.

Janvier, P. (1996). Early Vertebrates. Oxford: Clarendon Press.

Janvier, P. (2001). Ostracoderms and the shaping of the gnathostome characters. In P. E. Ahlberg
(Ed.), Major Events in Early Vertebrate Evolution: Palaeontology, Phylogeny, Genetics
and Development (pp. 172–186). London and New York: Taylor and Francis.

Kristan, W. B. (2016). Early evolution of neurons. Current Biology, 26, R949–R954.

Kudla, A. M., Miranda, X., & Nijhout, H. F. (2022). The roles of growth regulation and appendage
patterning genes in the morphogenesis of treehopper pronota. Proceedings of the Royal
Society B: Biological Sciences, 289, 20212682.

Kuklová-Peck, J. (1987). New Carboniferous Diplura, Monura, and Thysanura, the hexapod
ground plan, and the role of thoracic side lobes in the origin of wings (Insecta). Canadian
Journal of Zoology, 65, 2327–2345.

Lankester, E. R. (1870). On the use of the term homology in modern zoology, and the distinction
between homogenetic and homoplastic agreements. Annals and Magazine of Natural
History, 6, 34–43.

Lankester, E. R. (1888). Zoology. In Encyclopaedia Brittanica, Ninth Edition (Vol. 24 (URA-

ZYM), pp. 799–820). Edinburgh: Adam and Charles Black.

Letelier, J., Naranjo, S., Sospedra-Arrufat, I., Martinez-Morales, J. R., Lopez-Rios, J., Shubin, N.,
& Gómez-Skarmeta, J. L. (2021). The Shh/Gli3 gene regulatory network precedes the
origin of paired fins and reveals the deep homology between distal fins and digits.
Proceedings of the National Academy of Sciences, 118, e2100575118.

48
Lev, O., & Chipman, A. D. (2021). Development of the pre-gnathal segments in the milkweed bug
Oncopeltus fasciatus suggests they are not serial homologs of trunk segments. Frontiers in
Cell and Developmental Biology, 9, 695135.

Lev, O., Edgecombe, G. D., & Chipman, A. D. (2022). Serial Homology and Segment Identity in
the Arthropod Head. Integrative Organismal Biology, 4, obac015.

Love, A. C. (2006). Evolutionary morphology and evo-devo: Hierarchy and novelty. Theory in
Biosciences, 124, 317–333.

Love, A. C. (2007). Functional homology and homology of function: biological concepts and
philosophical consequences. Biology and Philosophy, 22, 691–708.

Mayr, E. (1969). Principles of Systematic Zoology. New York: McGraw-Hill.

McKenna, K. Z., Wagner, G. P., & Cooper, K. L. (2021). A developmental perspective of

homology and evolutionary novelty. In S. F. Gilbert (Ed.), Current Topics in
Developmental Biology (Vol. 141, pp. 1–38): Academic Press.

Metscher, B. D., Takahashi, K., Crow, K., Amemiya, C., Nonaka, D. F., & Wagner, G. P. (2005).
Expression of Hoxa-11 and Hoxa-13 in the pectoral fin of a basal ray-finned fish, Polyodon
spathula: implications for the origin of tetrapod limbs. Evolution & Development, 7, 186–
195.

Minelli, A. (1998). Molecules, developmental nodules, and phenotypes: a combinatorial approach

to homology. Molecular Phylogenetics and Evolution, 9, 340–347.

Minelli, A. (2000). Limbs and tail as evolutionarily diverging duplicates of the main body axis.
Evolution & Development, 2, 157–165.

Minelli, A. (2002). Homology, limbs, and genitalia. Evolution & Development, 4, 127–132.

Minelli, A., & Fusco, G. (2013). Homology. In K. Kampourakis (Ed.), The Philosophy of Biology:
A Companion for Educators (pp. 289–322). Dordrecht: Springer.

Minelli, A., & Perfuffo, B. (1991). Developmental pathways, homology and homonomy in
metameric animals. Journal of Evolutionary Biology, 3, 429–445.

Miyashita, T., & Diogo, R. (2016). Evolution of serial patterns in the vertebrate pharyngeal
apparatus and paired appendages via assimilation of dissimilar units. Frontiers in Ecology
and Evolution, 4.

Moczek, A. (2014). Towards a theory of development through a theory of developmental

evolution. In A. Minelli & T. Pradeu (Eds.), Towards a Theory of Development (pp. 218–
226). Oxford: Oxford University Press.

49
Müller, G. B. (2003). Homology: The evolution of morphological organization. In G. B. Müller &
S. A. Newman (Eds.), Origination of Organismal Form: Beyond the Gene in
Developmental and Evolutionary Biology (pp. 51–69). Cambridge, MA: MIT Press.

Müller, G. B., & Wagner, G. P. (1991). Novelty in evolution: restructuring the concept. Annual
Review of Ecology and Systematics, 22, 229–256.

Nakamura, T., Gehrke, A. R., Lemberg, J., Szymaszek, J., & Shubin, N. H. (2016). Digits and fin
rays share common developmental histories. Nature, 537, 225–228.

Niwa, N., Akimoto-Kato, A., Niimi, T., Tojo, K., Machida, R., & Hayashi, S. (2010). Evolutionary
origin of the insect wing via integration of two developmental modules. Evolution &
Development, 12, 168–176.

Nixon, K. C., & Carpenter, J. M. (2012). On homology. Cladistics, 28, 160–169.

Novick, A. (2018). The fine structure of ‘homology’. Biology & Philosophy, 33, 6.

Oftedal, O. T. (2002). The mammary gland and its origin during synapsid evolution. Journal of
Mammary Gland Biology and Neoplasia, 7(3), 225–252.

Okabe, Y., & Medzhitov, R. (2016). Tissue biology perspective on macrophages. Nature
Immunology, 17, 9–17.

Owen, R. (1843). Lectures on the Comparative Anatomy and Physiology of the Invertebrate
Animals. London: Longman, Brown, Green, and Longmans.

Owen, R. (1848). On the Archetype and Homologies of the Vertebrate Skeleton. London: John van
Voorst.

Patterson, C. (1982) Morphological characters and homology. In: K. A. Joysey & A. E. Friday
(Eds.), Problems in Phylogenetic Reconstruction (pp. 21-74). London: Academic Press.

Pavlicev, M., Herdina, A. N., & Wagner, G. (2022). Female genital variation far exceeds that of
male genitalia: a review of comparative anatomy of clitoris and the female lower
reproductive tract in Theria. Integrative and Comparative Biology, icac026.

Petit, F., Sears, K. E., & Ahituv, N. (2017). Limb development: a paradigm of gene regulation.
Nat Rev Genet, 18, 245–258.

Piatigorsky, J. (2007). Gene Sharing and Evolution: The Diversity of Protein Functions.
Cambridge: Harvard University Press.

Prince, V. E., & Pickett, F. B. (2002). Splitting pairs: the diverging fates of duplicated genes.
Nature Reviews Genetics, 3, 827–837.

50
Prud'homme, B., Minervino, C., Hocine, M., Cande, J. D., Aouane, A., Dufour, H. D., . . . Gompel,
N. (2011). Body plan innovation in treehoppers through the evolution of an extra wing-like
appendage. Nature, 473, 83–86.

Ralevski, A., Apelt, F., Olas, J. J., Mueller-Roeber, B., Rugarli, E. I., Kragler, F., & Horvath, T.
L. (2022). Plant mitochondrial FMT and its mammalian homolog CLUH controls
development and behavior in Arabidopsis and locomotion in mice. Cellular and Molecular
Life Sciences, 79, 334.

Roth, V. L. (1984). On homology. Biological Journal of the Linnean Society, 22, 13–29.

Roth, V. L. (1988). The biological basis of homology. In C. J. Humphries (Ed.), Ontogeny and
Systematics (pp. 1–26). New York: Columbia University Press.

Sattler, R. (1984). Homology – a continuing challenge. Systematic Botany, 9, 382–394.

Sattler, R. (1994). Homology, homeosis and process morphology in plants. In Homology: The
Hierachical Basis of Comparative Biology. B.K Hall (ed) Academic Press (pp. 423–475).

Schmitt, S. (2017). Serial homology as a challenge to evolutionary theory — the repeated parts of
organisms from idealistic morphology to evodevo. In P. Huneman & D. M. Walsh (Eds.),
Challenging the Modern Synthesis (pp. 317–347). Oxford: Oxford University Press.

Scott, M. P. (2000). Development: the natural history of genes. Cell, 100, 27–40.

Shubin, N., Tabin, C., & Carroll, S. B. (2009). Deep homology and the origins of evolutionary
novelty. Nature, 457, 818–823.

Shubin, N., Tabin, C., & Carroll, S. B. (1997). Fossils, Genes and the Evolution of Animal Limbs.
Nature, 388, 639–648.

Shubin, N., & Wake, D. (1996). Phylogeny, variation, and morphological integration. American
Zoologist, 36, 51–60.

Simpson, G. G. (1961). Principles of Animal Taxonomy. New York: Columbia University Press.

Siomava, N., Fuentes, J. S. M., & Diogo, R. (2020). Deconstructing the long-standing a priori
assumption that serial homology generally involves ancestral similarity followed by
anatomical divergence. Journal of Morphology, 281, 1110–1132.

Spemann, H. (1915). Zur Geschichte und Kritik des Begriffs der Homologie. In C. Chun & W.
Johannsen (Eds.), Allgemeine Biologie (Vol. 3.4, pp. 163–185). Leipzig und Berlin: B.G.
Teubner.

Sterner, B., & Lidgard, S. (2014). The normative structure of mathematization in systematic
biology. Studies in History and Philosophy of Biological and Biomedical Sciences, 46, 44–
54.

51
 Stewart, T. A., Liang, C., Cotney, J. L., Noonan, J. P., Sanger, T. J., & Wagner, G. P. (2019).
Evidence against tetrapod-wide digit identities and for a limited frame shift in bird wings.
Nature Communications, 10, 3244.

Tanaka, M., & Onimaru, K. (2012). Acquisition of the paired fins: a view from the sequential
evolution of the lateral plate mesoderm. Evolution & Development, 14, 412–420.

Wagner, G. P. (2007). The developmental genetics of homology. Nat Rev Genet, 8, 473–479.

Wagner, G. P. (2014). Homology, Genes, and Evolutionary Innovation. Princeton, NJ: Princeton
University Press.

Wagner, G. P., & Chiu, C. H. (2001). The tetrapod limb: A hypothesis on its origin. Journal of
Experimental Zoology, 291, 226–240.

Wilson, M. (2018). Physics Avoidance: Essays in Conceptual Strategy. Oxford: Oxford University
Press.

Wilson, M. V. H., Hanke, G. F., & Märss, T. (2007). Paired fins of jawless vertebrates and their
homologies across the agnathan–gnathostome transition. In J. S. Anderson & H. D. Sues
(Eds.), Major Transitions in Vertebrate Evolution (pp. 122–149). Bloomington, IN:
Indiana University Press.

Wimsatt, W. C. (2007). Re-Engineering Philosophy for Limited Beings: Piecewise

Approximations to Reality. Cambridge, MA: Harvard University Press.

Zeller, R., Lopez-Rios, J., & Zuniga, A. (2009). Vertebrate limb bud development: moving
towards integrative analysis of organogenesis. Nature Reviews Genetics, 10, 845–858.

Zhu, M., Yu, X., Choo, B., Wang, J., & Jia, L. (2012). An antiarch placoderm shows that pelvic
girdles arose at the root of jawed vertebrates. Biology Letters, 8, 453–456.

Zimmer, E. A., Martin, S. L., Beverley, S. M., Kan, Y. W., & Wilson, A. C. (1980). Rapid
duplication and loss of genes coding for the alpha chains of hemoglobin. Proceedings of
the National Academy of Sciences USA, 77, 2158–2162.

52

View publication stats