Disconnecting Bones Within the Jaw‐Otic Network Modules Underlies

Home , Maotherium, Quadrate bone, Squamosal bone

Journal of Anatomy

J. Anat. (2019) 235, pp15--33 doi: 10.1111/joa.12992

Disconnecting bones within the jaw-otic network modules underlies mammalian middle ear evolution Aitor Navarro-Dıaz,1 Borja Esteve-Altava2 and Diego Rasskin-Gutman1

1Paleobiology and Theoretical biology (Theoretical Biology), Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Valencia, Spain 2Institute of Evolutionary Biology (UPF-CSIC), Department of Experimental and Health Sciences, University Pompeu Fabra, Barcelona, Spain

Abstract

The origin of the mammalian middle ear ossicles from the craniomandibular articulation of their synapsid ancestors is a key event in the evolution of vertebrates. The richness of the fossil record and the multitude of developmental studies have provided a stepwise reconstruction of this evolutionary innovation, highlighting the homology between the quadrate, articular, pre-articular and angular bones of early synapsids with the incus, malleus, gonial and ectotympanic bones of derived mammals, respectively. There are several aspects involved in this functional exaptation: (i) an increase of the masticatory musculature; (ii) the separation of the quadrate bone from the cranium; and (iii) the disconnection of the post-dentary bones from the dentary. Here, we compared the jaw-otic complex for 43 synapsid taxa using anatomical network analysis, showing that the disconnection of mandibular bones was a key step in the mammalian middle ear evolution, changing the skull anatomical modularity concomitant to the acquisition of new functions. Furthermore, our analysis allows the identification of three types of anatomical modules evolving through five evolutionary stages during the anatomical transformation of the jawbones into middle ear bones, with the ossification and degradation of Meckel’s cartilage in mammals as the key ontogenetic event leading the change of anatomical modularity. Key words: anatomical network analysis (AnNA); Meckel’s cartilage; modularity; synapsida.

– Introduction (Watson, 1953; Hopson, 1966; Carroll, 1988, chapters 17 18; Rubidge & Sidor, 2001; Sidor, 2001, 2003; Luo, 2011; Han Several structures of the middle ear of mammals evolved et al., 2017; Luo et al., 2017) and recent advances in devel- from the lower jaw of basal synapsids through a series of opmental biology (Luo, 2011; Anthwal et al., 2013, 2017; anatomical changes that started about 315 million years Ramırez-Chaves et al., 2016; Urban et al., 2017). ago. The now extinct non-mammal synapsids had a lower The origin of the mammalian middle ear goes back to the jaw made of up to eight bones, which articulated to the mandibular and jaw joint arrangement of primitive synap- rest of the skull through the articular-quadrate jaw joint; sids. According to the Reichert–Gaupp theory, based on their only auditory bone was the stapes (Crompton & Par- comparative anatomy and development, the ectotympanic, ker, 1978; Sidor, 2003; Meng et al., 2011; Urban et al., gonial, malleus and incus bones of the ear in mammals are 2017). In contrast, modern mammals have a single jawbone, homologous to the angular, pre-articular, articular and the dentary, which articulates to the squamosal, and four quadrate bones of the jaw in reptiles (Rich et al., 2005; Luo, closely connected auditory bones: ectotympanic, malleus, 2011; Meng et al., 2011; Maier & Ruf, 2016; Han et al., incus and stapes. Many aspects of the origin and evolution 2017). At the same time, the evolutionary history of the of the mammalian middle ear are now better understood synapsid lower jaw suggests a trade-off between an thanks to the richness of the synapsids fossil record increase in the area of muscle attachment and a reduction of some mandibular bones to improve sound transmission (Hopson, 1966; Kermack et al., 1973; Fourie, 1974; Cromp- ton & Parker, 1978; Kemp, 1979, 2007; Carroll, 1988, pp. Correspondence 393–395; Wang et al., 2001; Sidor, 2003; Soares et al., 2011; Diego Rasskin-Gutman, Paleobiology and Theoretical Biology (Theo- retical Biology), Cavanilles Institute of Biodiversity and Evolutionary Ramırez-Chaves et al., 2016; Lautenschlager et al., 2017, Biology, University of Valencia, C/ Catedratico Jose Beltran n°2, 2018). During the Permian, synapsids evolved a new power- 46980 Paterna, Valencia, Spain. E: [email protected] ful adductor musculature attached to jawbones (Kemp, Accepted for publication 6 March 2019 1969, 1979; Reisz, 1972; Fourie, 1974; Crompton & Parker, Article published online 12 April 2019 1978). Such innovation enabled early synapsids to feed

larger preys and to increase their body mass (Watson, 1953; 2003; Kemp, 2007; Meng et al., 2011; Anthwal et al., 2013). Reisz, 1972; Carroll, 1988, pg. 363; Kammerer, 2011), leading The reduction in size and the cranial disconnection of this to an increase of the dentary surface that allowed the inser- bony chain increased its vibrational mobility and airborne tion of new-developed muscle fibers that changed biting sound sensitivity (Kermack et al., 1981; Laurin, 1998; Kemp, mechanisms (Carroll, 1988, p. 393; Lautenschlager et al., 2007; Luo, 2011; Meng et al., 2011). Mesozoic mammals 2017, 2018). evolved a new dentary-squamosal jaw joint (Hopson, 1966; After the Permian-Triassic and the Triassic-Jurassic mass Romer, 1970; Crompton & Parker, 1978; Luo & Crompton, extinction events, non-mammalian therapsids reduced their 1994; Kemp, 2007; Luo, 2011; Anthwal et al., 2013; Han body size (Hopson, 1966; Frobisch,€ 2007; Kemp, 2007; Sig- et al., 2017) and relocated the primitive quadrate-articular urdsen et al., 2012; Huttenlocker, 2014). The first small jaw joint to the middle ear after the novel ossification of an insectivorous mammaliaforms originated during the Triassic embryological mandibular element, the Meckel’s cartilage (Fourie, 1974; Carroll, 1988, pp. 401–402; Sidor, 2001) and (Wang et al., 2001; Meng et al., 2011; Urban et al., 2017). had a nocturnal lifestyle (Kermack et al., 1981; Luo, 2011; The loss of bones and the disconnections in the lower jaw Han et al., 2017). Living in the shadow of the large archo- (Sidor, 2001; Luo, 2011; Meng et al., 2011; Han et al., 2017; saurs that dominated this period, any anatomical changes Urban et al., 2017) produced novel patterns of anatomical on sensory organs that favored avoiding predation, as well organization in the jaw-otic complex. Authors recognize as detecting and capturing smaller prey, would have posed three distinct patterns of organization or configuration a selective advantage (Kermack et al., 1981; Luo, 2011; types in synapsids: (i) the mandibular middle ear type, with Urban et al., 2017). Triassic mammaliaforms had a fully reor- post-dentary bones attached to the dentary and a func- ganized musculoskeletal mandibular complex with some tional quadrate-articular jaw joint (Fig. 1A–D); (ii) the tran- jawbones having distinct new roles (Hopson, 1966; Kermack sitional mammalian middle ear type, with the middle ear et al., 1973; Kemp, 1979; Meng et al., 2011; Anthwal et al., bones indirectly connected to the mandible by a link to the 2013; Han et al., 2017; Lautenschlager et al., 2017, 2018). ossified Meckel’s cartilage (Fig. 1E); and (iii) the definitive Thereby, the enlarged dentary assumed all the masticatory mammalian middle ear type, with the middle ear bones muscular insertion in the new Mesozoic mammals, while fully disconnected from the dentary and isolated from par- the smaller post-dentary bones evolved an auditory role ticipating in any chewing action (Fig. 1F; Luo, 2011; Meng that improved sound transmission from the lower jaw to et al., 2011; Ramırez-Chaves et al., 2016; Anthwal et al., the inner ear through a bony chain between the angular, 2017; Han et al., 2017; Luo et al., 2017). These configura- articular, quadrate and stapes bones (Hopson, 1966; Ker- tions can also be seen as changes in the topological mack et al., 1973, 1981; Crompton & Parker, 1978; Sidor, arrangement of the ear bones throughout their evolution.

Fig. 1 Representation of the lower jaw transition throughout the synapsid evolution. Notice the enlargement of the dentary bone, concomitant to the reduction of post-dentary bones until their disconnection as new mammalian middle ear bones. Lower jaws from (A) to (D) illustrate the mandibular middle ear type, with post-dentary bones fully attached to the dentary; the lower jaw arrangement of (E) illustrates the transitional mammalian middle ear type, with the ear bones indirectly connected to the dentary by the ossified Meckel’s cartilage; and (F) illustrates the definitive mammalian middle ear type, with the ear bones totally disconnected from the dentary bone. (A) Medial view of the lower jaw of the primitive synapsid Dimetrodon (modified from Sidor, 2003). (B) Medial view of the lower jaw of the gorgonopsian therapsid Aelurognathus (modified from Broom, 1913). (C) Medial view of the lower jaw of the cynognathian cynodont Diademodon (modified from Hopson, 1966). (D) Medial view of the lower jaw of the mammaliaform Morganucodon (modified from Kermack et al., 1973). (E) Medial view of the lower jaw of the eutriconodont mammal Yanoconodon (modified from Luo et al., 2007). (F) Lateral view of the lower jaw and middle ear bones of the marsupial mammal Mon- odelphis (Wible, 2003; Luo, 2011). Mandibles are not to scale.

In general, changes in the topological arrangement of 1974; Crompton & Parker, 1978; Carroll, 1988, pp. 393–395; Luo & bones may result from a variety of factors, including semi- Crompton, 1994; Rybczynski, 2000; Wang et al., 2001; Martinelli & independent integration (i.e. modularity) of parts with a Rougier, 2007; Luo, 2011; Huttenlocker & Abdala, 2015; Lauten- schlager et al., 2017, 2018). The embryonic Meckel’s cartilage was common developmental origin, growth pattern and/or probably persistent in the adult lower jaw of some synapsid taxa functional co-dependences (Esteve-Altava, 2017). Most (Kermack et al., 1973; Sues, 1986; Rougier et al., 1996; Wang et al., likely, the evolution of the jawbones of early synapsids and 2001; Rich et al., 2005; Kemp, 2007; Luo, 2011; Luo et al., 2016; the origin of the mammalian middle ear entailed a compro- Ramırez-Chaves et al., 2016; Han et al., 2017), as could be inferred mise of all such factors. by the Meckelian groove or sulcus in the lingual side of some lower Here, we have analyzed changes in anatomical modular- jaws (Kermack et al., 1973; Sues, 1986; Bonaparte et al., 2003, 2005; ity of the mammalian middle ear associated to the evolu- Rich et al., 2005; Kemp, 2007; Meng et al., 2011; Ramırez-Chaves et al., 2016; Anthwal et al., 2017; Luo et al., 2017; Urban et al., tionary disconnection of the ear bones from the lower jaw. 2017). However, connectivity only can be reliably inferred in fossils Our working hypothesis is that the developmental recruit- when the element is present as hard fossilized tissue; thus, to regis- ment of jawbones to form the middle ear produced key ter the connectivity of Meckel’s cartilage, we have only coded it evolutionary changes in the anatomical modules of the when newly ossiﬁed in early mammals. jaw-otic complex within the skull. To this end, we carried out an anatomical network analysis (AnNA) of the topology Phylogenetic context of the jaw-otic region of the skull (see Materials and methods for details). First, we built anatomical network models To map the evolution of modularity patterns on the phylogeny, we for the jaw-otic complex for 39 extinct synapsids and four assembled a tree for the 43 taxa studied, following consensus phy- extant mammals. Then, we performed a network-based logenies for the Synapsida crown group (Rubidge & Sidor, 2001; modularity search for each taxon. After identifying the net- Sidor, 2003), basal synapsids (Spindler, 2015; Brocklehurst et al., 2016), non-mammalian therapsids (Frobisch€ & Reisz, 2008; Hutten- work modules for the jaw-otic region of interest, we evalu- locker et al., 2011; Kammerer, 2011, 2016, 2017; Huttenlocker & ated whether the network modules of the middle ear of Smith, 2017), non-mammalian cynodonts (Martinelli & Soares, 2016; mammals are more specialized than the ancestral lower Martinelli et al., 2017) and mammals (Luo, 2011; O’Leary et al., jaw’s network modules of their synapsid ancestors. Finally, 2013; Bi et al., 2014; Benton et al., 2015; Han et al., 2017). We built we characterized the structural role of bones within the the phylogenetic tree using R packages ape (Paradis et al., 2004), mandibular and otic regions by quantifying their connectiv- phytools (Revell, 2012), paleotree (Bapst, 2012) and strap (Bell & ity patterns in a phylogenetic context. We focused on Lloyd, 2015), using the time data reported in the Paleobiology Data- base (Peters & McClennen, 2016), Fossil Calibration Database changes in the organization of morphological modules (Ksepka et al., 2015) and the International Chronostratigraphic linkedtothekeyeventsunderlying the evolution of the Chart (v2017/02) of the International Commission on Stratigraphy jaw-otic complex in synapsids, such as the cranial disconnec- (Cohen et al., 2014) for calibration (see Supporting information for tion of the quadrate bone (Kemp, 1979, 2007; Kermack more details about the building of the phylogenetic tree). et al., 1981; Luo & Crompton, 1994; Laurin, 1998; Luo, 2011), the ossiﬁcation of the Meckel’s cartilage (Wang Network modeling et al., 2001; Meng et al., 2011), and the separation of the post-dentary bones from the lower jaw (Rich et al., 2005; We built network models of the lower jaw and middle ear complex, Luo et al., 2007; Luo, 2011; Meng et al., 2011; Anthwal in which each node codes for one bone and each link connecting et al., 2013; Han et al., 2017; Urban et al., 2017). two nodes codes for a physical contact between two bones (following Rasskin-Gutman & Esteve-Altava, 2014). Networks were mod- eled as binary adjacency matrices (A)ofdimensionN 9 N,whereN = Materials and methods is the number of bones. Aij 1 if there is a physical contact between bones i and j; Aij = 0 if there is none. All analyses were performed in R (R Core Team, 2017) using functions of the package Data collection igraph (Csardi & Nepusz, 2006).

We gathered information on the anatomy of the lower jaw and middle ear for 43 synapsid taxa, including species of basal synapsids, Modularity analysis non-mammalian therapsids, non-mammalian cynodonts, Mesozoic mammals and extant mammals (see Table S1 for details). We We delimited anatomical modules by optimization of the network included in our analysis those bones of the head that articulate to, modularity parameter Q, as implemented in the function clus- or share a muscular attachment with, bones of the lower jaw or ter_optimal. This algorithm maximizes the modularityhi measure P 2 ¼ m ks ds middle ear (see Table S2 for the complete list of bones). This deﬁned by Newman & Girvan (2004), as Q s¼1 K 2K ,where allowedustoconsidercranialbonesrelatedbymuscularattach- m is the number of modules of the partition, ks is the number of ments to the lower jaw as an intrinsic part of the anatomical system links within module s,andds is the total number of links of nodes under study; thus, following the idea that the increase of the jaw in s (both inside and outside s). We calculated the expected error of adductor musculature is one of the main reasons for the synapsid’s Q using a jackknife algorithm where each connection is an indepen- lower jaw evolution (Watson, 1953; Hopson, 1966; Kemp, 1969, dent observation (Newman & Girvan, 2004; see details of this analy- 1979, 2007; Romer, 1970; Reisz, 1972; Kermack et al., 1973; Fourie, sis in Table S3).

Topological characterization of the jaw and otic Data and code availability modules The data and code that support the ﬁndings of this study are avail- We characterized the topology within the jaw and otic modules able from Figshare at https://figshare.com/articles/Network_Modular by measuring their number of nodes (N ), number of connections m ity_Middle_Ear_Evolution/7173125. (Km), density of connections (Dm), mean path length (Lm), mean

cluster coefficient (Cm) and heterogeneity of connections (Hm; Ras- skin-Gutman & Esteve-Altava, 2014). Nm measures the total num- Results ber of nodes in the module: the total number of anatomical elements that are part of the module. K measures the number m Modularity of the jaw-otic complex of physical contacts among the bones of the module, thus capturing the intra-module integration. Dm measures the density of con- The jaw-otic complex of synapsids can take one of three nections into the module, that is, the number of connections in alternative network modular configurations: (i) mandibular; the module with respect to the maximum possible, as (ii) transitional otic-mandibular; and (iii) otic. The mandibu- 2Km Dm ¼ ð Þ; it captures the structural organization and serves as Nm Nm 1 lar module includes only bones of the lower jaw: dentary, a proxy for morphological complexity. Lm measures the characteristic path length, the proximity of each node to other nodes splenial, coronoids, angular, surangular, pre-articular, articular and the ossified Meckel’s cartilage (green modules in within theP module in terms of number of connections, as 1 Lm ¼ dn ;n ,whered is the shortest distance in number of – Nm 1 i j Figs. 2 4). The transitional otic-mandibular module includes connections between the nodes ni and nj; this parameter mea- both mandibular and middle ear bones: the above-referred sures the functional efficiency as the interdependence between jawbones plus the quadrate/incus and stapes bones (orange parts (i.e. proxy of integration). C measures the cluster coeffi- m modules in Fig. 5). The otic module includes the middle ear cient, the number of interconnections between the neighbors of bones ectotympanic, malleus, incus and stapes, and a node into the module,P that is, the triangular loops between 3- P s 1 i excludes the only bone that forms the lower jaw in mam- nodes, as Cm ¼ ,wheresi is the number of links Nm Ki ðKi 1Þ among the neighbors of node i; this parameter captures the co- mals, the dentary (golden modules in Fig. 6). dependency between parts and serves as another proxy of inte-

gration. Higher values of Dm and Cm are indicative of high com- Evolutionary stages plexity, while Lm is inversely proportional to it (Esteve-Altava et al., 2013; Rasskin-Gutman & Esteve-Altava, 2014). Finally, H m The three alternative modular configurations of jaw-otic quantifies the heterogeneity or variance of connections of the rKm complexes map into five evolutionary stages on the phy- nodes in the module, as Hm ¼ l ,whererKm and lKm are the Km logeny of synapsids (Fig. 7). This definition of evolutionary standard deviation and mean of Km, respectively; thus, capturing the structural disparity in the number of links per node. This last stages is based on: (i) the type of anatomical module parameter is a proxy for anisomerism, a property related to (mandibular, transitional otic-mandibular and otic); (ii) the structural specialization by the heterogeneity of body parts: type of jaw joint (quadrate-articular, surangular-squamosal, the less similar they are, the more specialized (Gregory, 1934, dentary-squamosal or a shared mixture of these); and (iii) 1935a,b; Esteve-Altava et al., 2013; Rasskin-Gutman & Esteve- the presence of a de novo ossification of the Meckel’s Altava, 2014). cartilage. The first evolutionary stage (S1) is found in basal synap- Role of lower jaw and middle ear bones sids and all major groups of therapsids except eucynodonts (plus the eucynodont family Tritylodontidae, represented We characterized the topological role of the dentary and middle by Kayentatherium). Their anatomical networks included ear bones within their modules by measuring their degree (Ki)and two mandibular modules that connect to the cranial mod- betweenness centrality parameter (BCi). Ki is the total numberP of connections for a given element in the network, Ki ¼ Aij ,and ules by the reptilian jaw joint between the quadrate and serves as a proxy of the total amount of functional and develop- articular bones (Fig. 2B,D). mental dependences of this element within the anatomical system The second evolutionary stage (S2) is found in advanced (Rasskin-Gutman & Esteve-Altava, 2014; Arnold et al., 2017). BCi non-mammalian cynodonts, or eucynodonts. A single measures the number of geodesics (shortest paths) in the network mandibular module characterizes their anatomical networks passing through a given node, where each shortest path is the mini- mum distance in number of links that connects two nodes in the due to the fusion of dentary and splenial bones in the sym- P ð Þ Ljp i physis. The connections between the mandibular module network. BCi ¼ ¼6 ¼6 2 ,whereLjp is the total number of short- j i p N Ljp est paths from node j to node p,andLjp (i) is the number of those and the cranial modules match the double jaw joint present

paths that pass through the node i. Nodes with higher BCi may have in eucynodonts (Fig. 3B): the primitive jaw joint quadrate- considerable inﬂuence within a network because they are more articular, and a new complementary jaw joint surangular- involved in the information passing through the nodes (Brandes, squamosal (Fig. 3D), or dentary-squamosal (Fig. 3E). 2001; Dos Santos et al., 2017). We have measured these parameters The third evolutionary stage (S3) is found in mammali- for the dentary, malleus, incus and stapes bones, excluding the gonial and ectotympanic bones because they disappeared by fusing aforms and some early mammals such as the haramiyidian to other structures in some groups, becoming untraceable as inde- Vilevolodon (Fig. 7), which present a mandibular module pendent elements. with the auditory bones joined to the dentary. S3 includes a

Fig. 2 Cranial anatomy and anatomical networks featuring the first evolutionary stage. (A) Skull and lower jaw of the primitive synapsid Vara- nosaurus in lateral (above), medial (below) and occipital (right) views (modified from Berman et al., 1995). (B) Anatomical network of the primitive synapsid Varanosaurus. (C) Skull and lower jaw of the early cynodont Procynosuchus in lateral (above), medial (below) and occipital (right) views (modified from Brink, 1961; Kemp, 1979; Allin & Hopson, 1992). (D) Anatomical network of the early cynodont Procynosuchus. The bones of evolutionary interest in this study have been drawn as follows: yellow for the dentary, red for angular/ectotympanic, light blue for pre-articular/gonial, blue for articular/malleus, green for quadrate/incus, turquoise for stapes, pink for the rest of the post-dentary bones (coronoids, splenial and surangular), orange for the ossified Meckel’s cartilage (Figures 4 and 5), metallic blue for squamosal, purple for the otic bones (opisthotic, prootic and their bone fusions), and gray for the remaining cranial bones according to their module identity. Cranial bones in white have not been included in this study. In the anatomical networks, square nodes represent the middle ear bones, 3D spherical nodes (Figures 4 and 5) represent the ossified Meckel’s cartilage, and the green areas represent the mandibular modules. Abbreviations: paired bones have been indicated in the nodes by ‘.r’ and ‘.l’ for the right and left sides, respectively; a = angular, ac = anterior coronoid, ar = articular, as = alisphenoid, BC = basicranium, bo = basioccipital, bs = basisphenoid, co = coronoid, d = dentary, ec = ectopterygoid, eo = exoccipital, ep = epipterygoid, go = gonial, in = incus, ip = in- terparietal, j = jugal, ma = malleus, mx = maxilla, oc = occipital, OMC = ossified Meckel’s cartilage, op = opisthotic, os = orbitosphenoid, ot = periotic, p = parietal, pa = pre-articular, pbs = parabasisphenoid, pc = posterior coronoid, pe = petrosal, pl = palatine, po = post-orbital, pp = post-parietal, pr = prootic, ps = pre-sphenoid, pt = pterygoid, q = quadrate, qj = quadratojugal, s = stapes, sa = surangular, so = supraoc- cipital, sp = splenial, sph = sphenoid complex, sq = squamosal, st = supratemporal, t = tabular, te = temporal, ty = ectotympanic, z = zygomatic. new bone within the mandibular modules in these mam- from the lower jaw. The post-dentary bones connect to the mals, the ossified Meckel’s cartilage (orange spherical node dentary indirectly, through the ossified Meckel’s cartilage OMC, ossified Meckel’s cartilage, in Fig. 4D,E). In this stage, (Fig. 5B). As a result, S4 differs from S3 in presenting two the mandibular network has split into two contralateral transitional otic-mandibular modules. This new pair of mod- modules due to an unfused symphysis, while keeping a ules includes the middle ear bones, quadrate/incus and double jaw joint shared between the reptilian quadrate- stapes, in addition to the jawbones (orange modules in articular and the new mammalian dentary-squamosal Fig. 5D,E). The primitive jaw joint quadrate-articular is now (Fig. 4B,C). part of these transitional modules. The fourth evolutionary stage (S4) is found in Mesozoic Finally, the fifth evolutionary stage (S5) is found in mammals that have the middle ear bones partially detached derived mammals that have a definitive middle ear

Fig. 3 Cranial anatomy and anatomical networks featuring the second evolutionary stage. (A) Skull and lower jaw of the probainognathian cynodont Probainognathus in lateral (above) and medial (below) views (modified from Carroll, 1988). (B) Double jaw joint in the probainognathian cynodont Probainognathus in medial (above) and ventral (below) views (modified from Carroll, 1988). (C) Skull of the probainognathian cynodont Lumkuia in occipital view (modified from Hopson & Kitching, 2001). (D) Anatomical network of the probainognathian cynodont Ecteninion. (E) Anatomical network of the tritheledontid cynodont Riograndia. Nomenclature of abbreviations, colors and shape of nodes indicated in Fig. 2.

with its ear bones totally disconnected from the Evolutionary trends in complexity and specialization mandible, with the dentary as the only jawbone (Fig. 6A–C). The ossiﬁed Meckel’s cartilage is absent in Network parameters (Table S4) show that through the S5 and ear bones organize in a new type of module, mammalian middle ear evolution the otic module in S5 has

the otic module (golden colored modules in Fig. 6D, lost half of the bones (Nm) present in the ancestral E); the two otic modules include the middle ear bones mandibular module in S1, with a tendency toward decreas-

only. At this later stage, the modularity-searching ing the module connectivity (Ks). In S5 the characteristic

algorithm separates modules preserving the known path length (Lm) increases their values in the otic module,

homology between the angular, pre-articular, articular while both density of connections (Dm) and clustering coef-

and quadrate bones of basal synapsids with the ecto- ﬁcient (Cm) increase. This suggests a reduction in the struc- tympanic, gonial, malleus and incus bones of mam- tural complexity, or a simpliﬁcation, of anatomical modules

mals, respectively. through evolution. The heterogeneity (Hm)valuesalso

Fig. 4 Cranial anatomy and anatomical networks featuring the third evolutionary stage. (A) Skull and lower jaw of the mammaliaform Mor- ganucodon in lateral (above) and medial (below) views (modified from Kermack et al., 1973, 1981). (B) Skull of the mammaliaform Morganu- codon in occipital view showing the new dentary-squamosal mammalian jaw joint (modified from Kermack et al., 1981). (C) Double jaw joint of the mammaliaform Morganucodon in ventral view (modified from Kermack et al., 1981). (D) Anatomical network of the mammaliaform Morganucodon. (E) Anatomical network of the mammaliaform Sinoconodon. Nomenclature of abbreviations, colors and shape of nodes indicated in Fig. 2.

increase through stages, suggesting a greater specialization reducing the number of connections or disconnecting the of the jaw-otic complex. middle ear bones in S2 and S3. In S4, the quadrate/incus and stapes bones keep the same values of K as in S3, while the articular/malleus remains disconnected. The K for the Topological changes in bone role dentary and middle ear bones is the lowest in the last evo- The analysis of connectivity (K) and betweenness centrality lutionary stage S5, where the trend of bone disconnection (BC) in the dentary and the middle ear bones showed dif- produces the modern configuration in mammals: a group ferences throughout the five evolutionary stages (Table S5). of middle ear bones separated from a single lower jaw Figure 8 compares the values of K and BC for each bone bone. and evolutionary stage. Regarding bones centrality, the BC values for the den- Regarding bones’ connectivity, the middle ear bones – tary and the middle ear bones are similar to K (Fig. 8B; articular/malleus, quadrate/incus and stapes – present their Table S5). The articular and quadrate bones have the high- highest number of connections in S1 (Fig. 8A; Table S5). est values of BC in S1 (Figs 8B and 9A). However, the high From this first evolutionary stage, there is a trend toward BC values of these bones decrease in S2 (Fig. 8B), where

Fig. 5 Cranial anatomy and anatomical networks featuring the fourth evolutionary stage. (A) Skull of the symmetrodont mammal Maotherium in lateral view (modified from Rougier et al., 2003). (B) Lower jaw of the eutriconodont mammal Yanoconodon in medial (above) and ventral (below) views showing the medio-lateral separation of the middle ear bones from the dentary by the curvature of the ossified Meckel’s cartilage (modified from Luo et al., 2007). (C) Lower jaw (above) and middle ear bones (below) of the eutriconodont mammal Liaoconodon showing the Meckelian groove on the dentary surface (modified from Meng et al., 2011). (D) Anatomical network of the eutriconodont mammal Liaoconodon. (E) Anatomical network of the symmetrodont mammal Maotherium. Red areas in the anatomical networks represent the transitional otic-mandibular modules. Nomenclature of abbreviations, colors and shape of nodes indicated in Fig. 2.

the emergence of the double jaw joint increases the BC in Discussion the dentary, surangular and squamosal bones that form the new complementary jaw joint (Fig. 9B). S3 also follows Evolution of the modular organization of the jaw- this pattern, the dentary and squamosal bones are now otic complex the ones with the highest values of BC, after acquiring a new role as main elements of the double jaw joint in mam- Our analysis of the jaw-otic complex using anatomical net- maliaforms (Fig. 9C). S4 shows a slight decrease of BC for works delimits three alternative modular conﬁgurations for the dentary (Fig. 8B), but it is still the node with the high- the bones involved in the mammalian middle ear evolution: est BC value within the new transitional otic-mandibular (i) mandibular; (ii) transitional otic-mandibular; and (iii) otic module, in which the middle ear bones keep their low cen- modules. The mandibular module is analogous to the trality (Figs 8B and 9D). However, the ear bones slightly mandibular middle ear complex described in non-mamma- increase their BC values within the otic module in S5; here, lian synapsids and mammaliaforms; the transitional otic- the auditory bony chain is completely disconnected from mandibular module occurs in the intermediate stage of the the dentary, which in turn decreases its BC value (Figs 8B mammalian middle ear evolution and is analogous to the and 9E). transitional mammalian middle ear described in some

Fig. 6 Cranial anatomy and anatomical networks featuring the fifth evolutionary stage. (A) Skull of the marsupial mammal Monodelphis in lateral view (modified from Wible, 2003). (B) Otic region of the marsupial mammal Monodelphis in ventral view (modified from Wible, 2003). (C) Middle ear bones of the marsupial mammal Monodelphis (left; modified from Luo, 2001), and the haramiyidan mammal Arboroharamiya (right; modified from Han et al., 2017). (D) Anatomical network of the haramiyidan mammal Arboroharamiya. (E) Anatomical network of the monotreme mammal Ornithorhynchus. Golden areas in the anatomical networks represent the otic modules. Nomenclature of abbreviations, colors and shape of nodes indicated in Fig. 2.

Mesozoic mammals, in which the ear bones connect only (Fig. 7). All mandibular modules group jawbones, but taxa partially to the lower jaw; and the otic module is analogous differ in their anatomical organization; for example, a dif- to the definitive mammalian middle ear described for ferent number of bones involved in the mandibular symph- extant mammals (Wang et al., 2001; Luo et al., 2007, 2016; ysis, or some bony fusions in the articular complex (see Ji et al., 2009; Luo, 2011; Meng et al., 2011; Anthwal et al., Supporting information figures). In addition, the more 2017; Han et al., 2017; Urban et al., 2017). The different derived non-mammalian eucynodont Kayentatherium,from organizational steps of these network modules allowed us the family Tritylodontidae, has its two mandibular rami to divide the evolution of the mammalian ear into five evo- grouped in a single mandibular module, perhaps due to a lutionary stages (Fig. 7). low number of mandibular elements forming its lower jaw The first evolutionary stage (S1) shows two mandibular (Figure S27). This configuration is typical of S2 for eucyn- modules (in green, Fig. 2B,D), which connect to the cranial odonts (see below), so it is not surprising that the trity- modules by the plesiomorphic quadrate-articular jaw joint lodontid Kayentatherium hasittoo.Infact,theonly (Kermack et al., 1973, 1981; Crompton & Parker, 1978; Car- mandibular feature to classify Kayentatherium in S1 is its roll, 1988, p. 395; Han et al., 2017). S1 is conserved during reversion to the reptilian quadrate-articular jaw joint (Sues, the evolution of synapsids, with a few variations in basal 1986), trait of focus of many discussions about the origin of synapsids, non-cynodont therapsids and epicynodonts mammals (Carroll, 1988, pp. 388–392; Luo, 1994; Bonaparte

Fig. 7 Phylogenetic tree of synapsids included in the study. The ﬁve evolutionary stages have been color-mapped on the tree. Major clade labels: A, Synapsida; B, Therapsida; C, Biarmosuchia; D, Dinocephalia; E, Anomodontia; F, Gorgonopsia; G, Therocephalia; H, Cynodontia; I, Eucynodon- tia; J, Mammaliaformes; K, Mammalia; L, Eutriconodonta; M, Allotheria; and N, Theria.

et al., 2003; Liu & Olsen, 2010; Martinelli et al., 2017; Bona- the definitive mammalian jaw joint (Luo, 2011; Han et al., parte & Crompton, 2018). 2017), which places tritheledontids and brasilodontids A single mandibular module characterizes the second eucynodonts close to the origin of the mammalian crown evolutionary stage (S2) of eucynodonts (green module in group (Bonaparte et al., 2003, 2005; Martinelli & Rougier, Fig. 3D,E). This module connects to the cranium by a double 2007; Liu & Olsen, 2010; Soares et al., 2011; Bonaparte & jaw joint (Fig. 3B), a feature related to an improved masti- Crompton, 2018). catory system in non-mammalian eucynodonts (Jasinoski & As a novelty, the third evolutionary stage (S3) presents Abdala, 2017; Lautenschlager et al., 2017, 2018), similar to the emergence of a newly ossified Meckel’s cartilage within that developed in mammals (Fourie, 1974). The mandibular the two mandibular network modules (orange spherical symphysis is often fused in these taxa (Fig. 3D), which con- nodes,OMC,intothegreen modules in Fig. 4D,E), here con- fers a greater pressure bite (Ivakhnenko, 2008). The primi- nected themselves by an unfused dentary symphysis. The tive reptilian jaw joint was reinforced with a new potential existence of the Meckel’s cartilage in the previous surangular articulation on the squamosal bone (Fig. 3B,D) evolutionary stages can be inferred by the presence of that would have provided resistance to the dislocation of Meckelian grooves in the lower jaw of some non-mamma- the lower jaw during biting because of the reduction in size lian eucynodonts (Kermack et al., 1973; Sues, 1986; Bona- of the quadrate and articular bones (Luo & Crompton, parte et al., 2005; Kemp, 2007); nonetheless, this new 1994; Han et al., 2017). The double jaw joint improved endochondral ossification constitutes a developmental nov- when the enlarged dentary began articulating to the squa- elty and a crucial step in the mammalian middle ear evolu- mosal (Fig. 3E), thus excluding the surangular from the joint tion (Sidor, 2001; Anthwal et al., 2017). The earliest record (Fig. 3D). This novel dentary-squamosal articulation formed of an ossified Meckel’s cartilage is found in small

Fig. 8 Boxplots of the connectivity (K) and betweenness centrality (BC) values of the dentary and middle ear bones (articular/malleus, quadrate/incus and stapes) in each evolutionary stage. (A) Connectivity results. Black bar is either the statistical mode or mean, as indicated in Table 1. (B) Betweenness Centrality results. mammaliaforms such as Morganucodon (Fig. 4A,B), which et al., 2001; Luo, 2011; Meng et al., 2011; Ramırez-Chaves retained the feeble, but still functional, quadrate-articular et al., 2016; Urban et al., 2017). Their lower jaws comprise jaw joint by reinforcing it with a stronger dentary-squamo- only the dentary bone, with some vestiges of splenial and sal articulation (Fig. 4B,C; Luo & Crompton, 1994), thus coronoid bones (Luo et al., 2007, 2017; Han et al., 2017), allowing mammals to have a more effective control of the and there is no trace of connections to post-dentary bones adductor musculature and jaw articulation (Kermack et al., inferred by the absence of post-dentary trough (Meng 1973). Once the mammalian dentary-squamosal articulation et al., 2011; Han et al., 2017; Luo et al., 2017). However, ear was in place, the ancestral quadrate-articular jaw joint was bones were not completely isolated, remaining indirectly free to evolve as a sound transmission structure (Hopson, connected to the mandible via the ossiﬁed Meckel’s carti- 1966; Luo, 2011; Anthwal et al., 2013). lage (Fig. 5B), which now serves as a stabilizing bridge for The fourth evolutionary stage (S4) corresponds to a transi- the migration of ear bones from the jaw to the base of the tional state of the mammalian middle ear evolution. S4 is skull (Luo et al., 2007; Meng et al., 2011). In S4, the exclu- characteristic of those Mesozoic mammals having a Mecke- sively mandibular modules of S1 to S3 include now also the lian groove on the dentary surface, as a trace of attachment auditory ossicles quadrate/incus and stapes (orange mod- for a persistent ossiﬁed Meckel’s cartilage (Fig. 5C; Wang ules in Fig. 5D,E), thus showing these new transitional otic-

Fig. 9 Anatomical networks representation of the BCi results for each cranial bone varying throughout the five evolutionary stages. The greater size and heat color of the nodes, the greater value of their BCi; square nodes represent the middle ear bones. (A) BCi results on the network of the primitive synapsid Eohaptodus of the first evolutionary stage. (B) BCi results on the network of the probainognathian cynodont Chiniquodon of the second evolutionary stage. (C) BCi results on the network of the mammaliaform Sinoconodon of the third evolutionary stage. (D) BCi results on the network of the symmetrodont mammal Maotherium of the fourth evolutionary stage. (E) BCi results on the network of the placental mammal Homo of the fifth evolutionary stage. mandibular modules a mix of mandibular and ear bones. otic-mandibular network modules found in S4 to the otic The structure of the anatomical network changed in S4 and network modules found in S5, generating the definitive so did its modularity. The mammalian dentary-squamosal mammalian middle ear. articulation is already fully functional in S4, freeing the Various lineages independently acquired the mam- quadrate-articular joint to, together with the pre-articular malian middle ear by homoplasy (Luo, 2011) , just like and angular, evolve for transmitting airborne sounds our modularity analysis has shown for S5 (Fig. 7). Placen- (Wang et al., 2001; Kemp, 2007; Han et al., 2017). This tran- tal and marsupial mammals have different cellular and sitional mammalian middle ear fills the morphological gap apoptosis mechanism involved in degradation of the between the ancestral mandibular middle ear found in S1 Meckel’s cartilage (Urban et al., 2017), and there is evi- to S3 and the derived definitive mammalian middle ear dence for the independent acquisition of a middle ear found in S5 (Meng et al., 2011). separation in monotremes and some Mesozoic mam- Finally, a middle ear fully disconnected from the lower malian clades (Rich et al., 2005; Luo, 2011; Han et al., jaw (now only the dentary) characterizes the fifth evolu- 2017). The modularity changes concomitant to the evolutionary stage (S5; Fig. 6A–C). All extant mammals belong to tion of the mammalian middle ear that we reported S5, as well as those mammals of the fossil record whose also evolved convergently. dentaries lack the Meckelian and post-dentary grooves on their surface. The ossified Meckel’s cartilage disappears in Disconnection of the middle ear bones as a devel- S5, while the angular, pre-articular-articular and quadrate opmental evolutionary trigger bones form the definitive middle ear bones ectotympanic, malleus and incus, respectively (Fig. 6A–C; Meng et al., Our comparison of the number of connections and 2011). The otic modules defining S5 (golden modules in betweenness centrality for key bones (Table S5; Fig. 8) sug- Fig. 6D,E) show a full disconnection of the middle ear bones gests that there was an evolutionary trend toward bone dis- from their mandibular origin. Such evolutionary transition connection in the jaw-otic complex underlying the above- needed the resorption of Meckel’s cartilage before its ossifi- described changes of modularity. cation during development (Luo, 2011; Anthwal et al., S1 presents the highest values of K and BC for the bones 2013, 2017; Maier & Ruf, 2016; Urban et al., 2017). Ontoge- of the middle ear (Fig. 8). While the larger size of the netically, the proximal part of Meckel’s cartilage is medially articular and quadrate bones of basal synapsids explains displaced from the mandible and ossifies forming the incus their high values of K, high BC values of these bones and malleus bones (Fig. 5B), which display a negative allom- appear because they are the only bones that connect the etry compared with the size of the dentary and the skull cranium to the mandibular complex (Fig. 9A). For the (Luo, 2011). The dentary grows until it meets the squamosal stapes, its high K value is also due to its massive size at in a functional jaw joint (Luo, 2011; Anthwal et al., 2013, this stage (the larger the bone, the more contacts it can 2017). At this point in development, processes of apoptosis make), playing a structural role supporting the braincase and chondroclast activity degrade the Meckelian cartilagi- by linking a fenestra ovalis bounded by up to four bones nous matrix before the onset of osteogenesis (Anthwal in non-mammalian synapsids: the prootic, opisthotic, et al., 2013, 2017; Urban et al., 2017). This distal resorption basioccipital and parabasisphenoid (Fig. 2A; Mendrez, of the Meckel’s cartilage breaks the connection between 1974; Crompton & Parker, 1978; Carroll, 1988, pp. 363, the middle ear and the lower jaw, allowing the formation 394; Reisz et al., 1992; Clack, 1998, 2002; Sigurdsen et al., of the mammalian middle ear (Anthwal et al., 2013, 2017; 2012). The mammalian middle ear bones could have func- Urban et al., 2017). In fact, the absence of chondroclast tioned transmitting low-frequency sounds from the activity generates the full endochondral ossification of ground to the inner ear in S1 (Reisz et al., 1992; Ivakh- Meckel’s cartilage in mice and opossum mutants, resulting nenko, 2008; Huttenlocker & Sidor, 2016); in fact, our in a physical joint of the middle ear bones with the lower results suggest that the amplitude of vibrational move- jaw that mirrors the ancestral phenotype of S4 (Luo, 2011; ments could have been constrained not only by their size,

Luo et al., 2016; Anthwal et al., 2017; Urban et al., 2017). but also by their many connections (Karticular = 4,

Thus, the complete disconnection of the middle ear bones Kquadrate = 6, Kstapes = 5; Fig. 8A; Table 1). from the lower jaw was the key evo-devo event that trig- From the ancestral S1 conﬁguration, the middle ear gered the second modularity change, from the transitional bones began a trend of disconnection from the mandibular

Table 1 Summary of the features that characterize the ﬁve evolutionary stages of the mammalian middle ear evolution.

Stage 1 Stage 2 Stage 3 Stage 4 Stage 5

Clades Non-eucynodont Non-mammalian Mammaliaformes Mesozoic mammals Mammals without synapsids eucynodonts and Mesozoic with Meckelian Meckelian groove mammals with groove (extant mammals) post-dentary trough Jaw joint Quadrate-articular Double jaw joint: Double jaw joint: Dentary-squamosal Dentary-squamosal quadrate-articular quadrate-articular and surangular- and dentary- squamosal or squamosal dentary-squamosal ID modules Two mandibular A single mandibular Two mandibular Two transitional Two otic modules modules module modules otic-mandibular modules Bones in the Dentary, splenial, Dentary, splenial, Dentary, coronoid, Dentary, splenial, Ectotympanic, modules anterior coronoid, coronoid, angular, angular, surangular, coronoid, ossified malleus, incus posterior coronoid, surangular, articular and ossified Meckel’s cartilage, and stapes angular, surangular, pre-articular and Meckel’s cartilage ectotympanic, pre-articular and articular gonial, articular malleus, incus and stapes Mode of Dentary = 6 Dentary = 9 Dentary = 6 Dentary = 3 Dentary = 2 connections Articular = 4 Articular = 5* Articular = 4 Malleus = 2 Malleus = 2* of the dentary Quadrate = 6 Quadrate = 5 Quadrate = 3 Incus = 3 Incus = 2* and middle Stapes = 5 Stapes = 3 Stapes = 2 Stapes = 2 Stapes = 2 ear bones Type of middle ear Mandibular Mandibular Mandibular Transitional Definitive middle ear middle ear middle ear mammalian mammalian middle ear middle ear

*When bones have a multimodal distribution, we took the mean of connections (see Table S5 for information on the number of con-

nections, Ki).

complex in S2 and S3 (Fig. 8A). The new connection of the frequency sounds (Kermack et al., 1981; Carroll, 1988, pp. squamosal with the surangular bone in S2 (Fig. 9B), and 387, 393–394; Laurin, 1998; Kemp, 2007). At the same time, with the dentary in S3 (Fig. 9C), increased their BC values at the quadrate developed a syndesmotic joint within the squa- these two stages above the highest BC values that the mosal recess, instead of being attached to it by a suture quadrate-articular joint had at S1. This replacement of the (Fig. 3C), thus freeing it from some constraints and allowing BC value could indicate that the new mammalian dentary- it to achieve a more vibrational mobility (Kemp, 1979, 2007; squamosal connection joint is now assuming the compres- Luo & Crompton, 1994; Luo, 1994). Finally, the stapes was sive load transmission carried out by the ancestral quadrate- also freed from its supporting role and performed a better articular jaw joint; the miniaturization of the mandible at sound transmission to the inner ear (Kemp, 1979; Carroll, these stages could have facilitated this new functional role 1988, p. 394; Clack, 2002). The key event that boosted these (Lautenschlager et al., 2018). In turn, the quadrate and bony disconnections was the development of a new jaw articular bones had mostly a masticatory role in S2 and S3, joint. In stages S2 and S3, the quadrate-articular joint became making eucynodonts and mammaliaforms less sensitive to simpler (Luo & Crompton, 1994), but still played a dual func- high-frequency sounds (Kermack et al., 1981; Kemp, 2007; tion maintaining the jaw joint and transmitting low-fre- Manley & Sienknecht, 2013), presumably because their quency sounds (Kermack et al., 1981; Kemp, 2007; Meng ear bones still had too many connections to the cranium et al., 2011; Han et al., 2017; Lautenschlager et al., 2018).

(Karticular = 5, Kquadrate = 5, Kstapes = 3, for S2; and A partial decoupling of feeding and hearing functions

Karticular = 4, Kquadrate = 3, Kstapes = 2, for S3; Fig. 8A, occurred at stage S4, in which the new mammalian den- Table 1). However, unlike S1, the post-dentary bones lose tary-squamosal jaw joint presents higher BC values than the sutural contacts on the dentary surface at these stages, primitive quadrate-articular jaw joint (Fig. 9D). The dentary becoming sometimes isolated as a single rod suspended over and middle ear bones are more disconnected (Fig. 8A), the dentary trough by connective tissue, thus allowing vibra- resulting in a transitional otic-mandibular module whose tional movements under the inﬂuence of intense low- ear bones could function more efﬁciently transmitting

© 2019 Anatomical Society Network modularity of the mammalian middle ear evolution, A. Navarro-Dıaz et al. 29 airborne sounds, although still constrained by the chewing Changes in complexity, specialization and bone function because of their mandibular connections with the contribution to the jaw-otic complex ossiﬁed Meckel’s cartilage (Fig. 5B,C; Meng et al., 2011; Han et al., 2017). In stage S5, when the Meckel’s cartilage is Morphologically, the single dentary of mammals is a ontogenetically degraded and the middle ear bones are more specialized bone than the compound lower jaw of completely disconnected from the dentary, feeding and their synapsid ancestors (Anthwal et al., 2013). Through- hearing functions were ﬁnally fully dissociated. Due to out its evolution, the dentary has developed a multitude these events of bone disconnections in S4 and S5, the den- of bony processes for the insertion of different muscles tary decreased both its K and its BC values (Fig. 8). Similarly, to perform all the chewing functions once shared by mul- the middle ear bones also displayed the progressive reduc- tiple bones in the lower jaws of non-mammalian synap- tion of K values in S4 (Karticular = 2, Kquadrate = 3, Kstapes = 2) sids (Fig. 1; Anthwal et al., 2013; Lautenschlager et al., and reached their lowest values in S5 (Karticular = 2, 2018). As suggested by Williston’s Law, evolution tends

Kquadrate = 2, Kstapes = 2; Fig. 8A, Table 1). Because the incus to reduce the number of cranial elements whilst increas- and stapes are the central components of the auditory ing the specialization of the remaining ones, a process chain in S5, it is not surprising that their BC values are called anisomerism (Williston, 1914; Gregory, 1934, 1935a; slightly higher than in S4 (Figs 8B and 9E). The otic network Esteve-Altava et al., 2013; Rasskin-Gutman & Esteve- conﬁguration of S5, that of a deﬁnitive mammalian middle Altava, 2014). Heterogeneity (Hm), a network proxy for ear, allows ear bones to reach a greater vibrational mobility anisomerism, captures this phenomenon showing that making them more sensitive to high-frequency airborne otic modules in S5 have fewer bony elements (Nm)and sounds (Hopson, 1966; Kermack et al., 1981; Luo, 2011; are more specialized anatomical units than the mandibu- Manley & Sienknecht, 2013). lar modules from which they evolved (Table S4).

Fig. 10 Summary of features of the mammalian middle ear evolution throughout the ﬁve evolutionary stages described in this study, according to the types of jaw joint, middle ear, anatomical modules, the ossiﬁcation of the Meckel’s cartilage, and the measured number of connections (Ki)of the dentary, articular/malleus, quadrate/incus and stapes.

However, our results for the within-module complexity ontogenetic degradation of the ossiﬁed Meckel’s cartilage show that the mammalian middle ear evolution led to an in modern mammals broke the physical connection otic module that is simpler than the ancestral mandibular between the dentary and the newly middle ear bones, thus

module from which it evolved (DS5 < DS1; CS5 < CS1; LS5 > LS1; generating the transition to the otic module in S5 (Fig. 10). Table S4). The loss and the massive disconnection of the Both modularity changes were necessary to decouple feed-

bones involved in the jaw-otic modules (Km values in ing and hearing functions that the middle ear bones shared Table S4) caused the evolutionary simplification of the whole in ancestral stages of synapsids. anatomical region, resulting in a bony chain that improves Finally, our findings on the evolution of anisomerism and the transmission of sound. This result opposes to the general complexity suggest that the emergence of the mammalian trend observed for the overall mammalian skull, whose dentary-squamosal jaw joint supposed the starting point of anatomical transformations led to an increase in anatomical a semi-independent evolution between middle ear bones complexity accompanying the reduction in bone number and the rest of the skull: whilst the mammalian skull (Esteve-Altava et al., 2013, 2014; see also Sidor, 2001, and increased its structural complexity (Esteve-Altava et al., McShea & Hordijk, 2013 for a different interpretation). This 2013, 2014), the jaw and middle ear simplified themselves would suggest that the mammalian middle ear and the over- and became more specialized. all skull evolved semi-independently, following decoupled evolutionary trends of morphological complexity. Acknowledgements

Concluding remarks This project was funded by the Spanish Ministerio de Economıa y Competitividad (BFU2015-70927-R) to DR-G. BE-A has received Our anatomical network analysis of the mammalian mid- financial support through the Postdoctoral Junior Leader Fellowship dle ear aligns with current views on the processes leading Programme from ‘la Caixa’ Banking Foundation (LCF/BQ/LI18/ 11630002) and thanks the support of the Unidad de Excelencia to its evolution, while providing new sources of informa- Marıa de Maeztu (MDM-2014-0370). tion to characterize and trace its evolutionary patterns. We identified three types of anatomical modules that explain how the middle ear bones evolved semi-indepen- Conflict of interest dently within the skull through five evolutionary stages The authors declare no conflict of interest. (Fig. 10). This finding allows us to pinpoint three key developmental events that triggered the anatomical changes, transforming the modularity of the mammalian Author contributions jaw-otic complex: (i) the emergence of the new dentary- AN-D, BE-A and DR-G designed the study. AN-D collected squamosal articulation; (ii) the evolutionary trend toward the data and built the network models. All authors post-dentary disconnection; and (iii) the Meckel’s cartilage analyzed the data, interpreted the results and wrote the ossification and degradation. manuscript. Although the new dentary-squamosal articulation did not change the modular structure of the lower jaw, the change of values of BC from S1 to S3 highlights how the References emergence of this newly mammalian jaw joint played a key Allin EF, Hopson JA (1992) Evolution of the auditory system in role in reducing the participation of the quadrate and artic- Synapsida (“mammal-like reptiles” and primitive mammals) as ular bones in it, boosting their functional exaptation seen in the fossil record. In: The Evolutionary Biology of Hear- toward vibrational sound transmission in S4 and S5. These ing. (eds Webster DB, et al.), pp. 587–614. New York: results correlate with the idea that the dentary-squamosal Springer. jaw joint reduced the compressive load transmission of Anthwal N, Joshi L, Tucker AS (2013) Evolution of the mammalian middle ear and jaw: adaptations and novel structures. the quadrate bone during mastication (Lautenschlager J Anat 222, 147–160. et al., 2018). Anthwal N, Urban DJ, Luo Z-X, et al. (2017) Meckel’s cartilage To attain the characteristic mammalian middle ear config- breakdown offers clues to mammalian middle ear evolution. uration, the jaw-otic complex underwent two modularity Nat Ecol Evol 1,1–6. changes linked to an evolutionary trend toward bone dis- Arnold P, Esteve-Altava B, Fischer MS (2017) Musculoskeletal connection (Fig. 10). The two modularity changes have a networks reveal topological disparity in mammalian neck evo- – common key trigger: the transformation of Meckel’s carti- lution. BMC Evol Biol 17,1 18. Bapst DW (2012) Paleotree: an R package for paleontological lage. The Meckel’s cartilage showed a neomorphic ossifica- and phylogenetic analyses of evolution. Methods Ecol Evol 3, tion in S3, serving as a stabilizing structure in the 803–807. disconnection of post-dentary bones from the dentary in Bell MA, Lloyd GT (2015) Strap: an R package for plotting phylo- S4, then changing the ancestral mandibular module to a genies against stratigraphy and assessing their stratigraphic new transitional otic-mandibular module. Finally, the novel congruence. Palaeontology 58, 379–389.

Benton MJ, Donoghue PCJ, Asher RJ, et al. (2015) Constraints Frobisch€ J, Reisz RR (2008) A new species of Emydops (Synap- on the timescale of animal evolutionary history. Palaeontol sida, Anomodontia) and a discussion of dental variability and Electron 18,1–106. pathology in dicynodonts. J Vertebr Paleontol 28, 770–787. Berman DS, Reisz RR, Bolt JR, et al. (1995) The cranial anatomy Gregory WK (1934) Polyisomerism and anisomerism in cranial and relationships of the synapsid Varanosaurus (Eupely- and dental evolution among vertebrates. Proc Natl Acad Sci cosauria: Ophiacodontidae) from the Early Permian of Texas USA 20,1–9. and Oklahoma. Ann Carnegie Mus 64,99–133. Gregory WK (1935a) On the evolution of the skulls of vertebrates Bi S, Wang Y, Guan J, et al. (2014) Three new Jurassic with special reference to heritable changes in proportional euharamiyidan species reinforce early divergence of mammals. diameters (anisomerism). Proc Natl Acad Sci USA 21,1–8. Nature 514, 579–584. Gregory WK (1935b) ‘Williston’s law’ relating to the evolution Bonaparte JF, Crompton AW (2018) Origin and relationships of of skull bones in the vertebrates. Am J Phys Anthropol 20, the Ictidosauria to non-mammalian cynodonts and mammals. 123–152. Hist Biol 30, 174–182. Han G, Mao F, Bi S, et al. (2017) A Jurassic gliding euharamiyi- Bonaparte JF, Martinelli AG, Schultz CL, et al. (2003) The sister dan mammal with an ear of five auditory bones. Nature 551, group of mammals: small cynodonts from the Late Triassic of 451–456. southern Brazil. Rev Bras Paleontol 5,5–27. Hopson JA (1966) The origin of the mammalian middle ear. Bonaparte JF, Martinelli AG, Schultz CL (2005) New information Integr Comp Biol 6, 437–450. on Brasilodon and Brasilitherium (Cynodontia, Probainog- Hopson JA, Kitching JW (2001) A probainognathian cynodont nathia) from the Late Triassic, southern Brazil. Rev Bras Pale- from South Africa and the phylogeny of nonmammalian cyn- ontol 8,25–46. odonts. Bull Mus Comp Zool 156,5–35. Brandes U (2001) A faster algorithm for betweenness centrality. Huttenlocker AK (2014) Body size reductions in nonmammalian J Math Sociol 25, 163–177. eutheriodont therapsids (Synapsida) during the end-Permian Brink AS (1961) A new type of primitive cynodont. Paleontol Afr mass extinction. PLoS ONE 9, e87553. 7, 119–154. Huttenlocker AK, Abdala F (2015) Revision of the first the- Brocklehurst N, Reisz RR, Fernandez V, et al. (2016) A re-descrip- rocephalian, Theriognathus Owen (Therapsida: Whaitsi- tion of “Mycterosaurus” smithae, an Early Permian eothyridid, idae), and implications for cranial ontogeny and allometry and its impact on the phylogeny of pelycosaurian-grade in nonmammaliaform eutheriodonts. J Paleontol 89,645– synapsids. PLoS ONE 11, e0156810. 664. Broom R (1913) On the Gorgonopsia, a suborder of the mam- Huttenlocker AK, Sidor CA (2016) The first karenitid (Therapsida, mal-like reptiles. Proc Zool Soc Lond 1913, 225–230. Therocephalia) from the Upper Permian of Gondwana and Carroll RL (1988) Vertebrate Paleontology and Evolution. New the biogeography of Permo-Triassic therocephalians. J Vertebr York: WH Freeman. Paleontol 36, e1111897. Clack JA (1998) The neurocranium of Acanthostega gunnari Jar- Huttenlocker AK, Smith RMH (2017) New whaitsioids (Therap- vik and the evolution of the otic region in tetrapods. Zool J sida: Therocephalia) from the Teekloof Formation of South Linn Soc 122,61–97. Africa and therocephalian diversity during the end-Guadalu- Clack JA (2002) Patterns and processes in the early evolution of pian extinction. PeerJ 5, e3868. the tetrapod ear. J Neurobiol 53, 251–264. Huttenlocker AK, Sidor CA, Smith RMH (2011) A new specimen Cohen KM, Finney SC, Gibbard PL, et al. (2014) The ICS of Promoschorhynchus (Therapsida: Therocephalia: Akidnog- International Chronostratigraphic Chart. Episodes 36,199– nathidae) from the Lower Triassic of South Africa and its 204. implications for theriodont survivorship across the Permo- Crompton AW, Parker P (1978) Evolution of the mammalian Triassic boundary. J Vertebr Paleontol 31, 405–421. masticatory apparatus. Am Sci 66, 192–201. Ivakhnenko MF (2008) Cranial morphology and evolution of Per- Csardi G, Nepusz T (2006) The igraph software package for com- mian Dinomorpha (Eotherapsida) of eastern Europe. Paleontol plex network research. InterJ Complex Syst 1695,1–9. J 42, 859–995. Dos Santos DA, Fratani J, Ponssa ML, et al. (2017) Network Jasinoski SC, Abdala F (2017) Cranial ontogeny of the Early Tri- architecture associated with the highly specialized hindlimb of assic basal cynodont Galesaurus planiceps. Anat Rec 300, 353– frogs. PLoS ONE 12,1–17. 381. Esteve-Altava B (2017) In search of morphological modules: a Ji Q, Luo Z-X, Zhang X, et al. (2009) Evolutionary development systematic review. Biol Rev 92, 1332–1347. of the middle ear in Mesozoic therian mammals. Science 326, Esteve-Altava B, Marugan-Lob on J, Botella H, et al. (2013) Struc- 278–281. tural constraints in the evolution of the tetrapod skull com- Kammerer CF (2011) Systematics of the Anteosauria (Therapsida: plexity: Williston’s Law revisited using network models. Evol Dinocephalia). J Syst Palaeontol 9, 261–304. Biol 40, 209–219. Kammerer CF (2016) Two unrecognised burnetiamorph specimens Esteve-Altava B, Marugan-Lob on J, Botella H, et al. (2014) Ran- from historic Karoo collections. Palaeontol Afr 50,64–75. dom loss and selective fusion of bones originate morphologi- Kammerer CF (2017) Anatomy and relationships of the South cal complexity trends in tetrapod skull networks. Evol Biol 41, African gorgonopsian Arctops (Therapsida, Theriodontia). Pap 52–61. Palaeontol 3, 583–611. Fourie S (1974) The cranial morphology of Thrinaxodon liorhi- Kemp TS (1969) On the functional morphology of the gor- nus Seeley. Ann S Afr Mus 65, 337–400. gonopsid skull. Philos Trans R Soc B Biol Sci 256,1–83. Frobisch€ J (2007) The cranial anatomy of Kombuisia frerensis Kemp TS (1979) The primitive cynodont Procynosuchus: func- Hotton (Synapsida, Dicynodontia) and a new phylogeny of tional anatomy of the skull and relationships. Philos Trans R anomodont therapsids. Zool J Linn Soc 150, 117–144. Soc B Biol Sci 285,73–122.

Kemp TS (2007) Acoustic transformer function of the postden- Mendrez CH (1974) A new specimen of Promoschorhynchus tary bones and quadrate of a nonmammalian cynodont. J Ver- platyrhinus Brink 1954 (Moschorhinidae) from the Dapto- tebr Paleontol 27, 431–441. cephalus-zone (Upper Permian) of South Africa. Paleontol Afr Kermack KA, Mussett F, Rigney HW (1973) The lower jaw of 17,69–85. Morganucodon. Zool J Linn Soc 53,87–175. Meng J, Wang Y, Li C (2011) Transitional mammalian middle ear Kermack KA, Mussett F, Rigney HW (1981) The skull of Mor- from a new Cretaceous Jehol eutriconodont. Nature 472, 181– ganucodon. Zool J Linn Soc 71,1–158. 185. Ksepka DT, Parham JF, Allman JF, et al. (2015) The fossil calibra- Newman MEJ, Girvan M (2004) Finding and evaluating commu- tion database – a new resource for divergence dating. Syst nity structure in networks. Phys Rev E 69,1–16. Biol 64, 853–859. O’Leary MA, Bloch JI, Flynn JJ, et al. (2013) The placental mam- Laurin M (1998) New data on the cranial anatomy of Lycaenops mal ancestor and the Post-K-Pg radiation of placentals. (Synapsida, Gorgonopsidae), and reﬂections on the possible Science 339, 662–667. presence of streptostyly in gorgonopsians. J Vertebr Paleontol Paradis E, Claude J, Strimmer K (2004) APE: analyses of phyloge- 18, 765–776. netics and evolution in R language. Bioinformatics 20, 289– Lautenschlager S, Gill P, Luo Z-X, et al. (2017) Morphological 290. evolution of the mammalian jaw adductor complex. Biol Rev Peters SE, McClennen M (2016) The Paleobiology Database 92, 1910–1940. application programming interface. Paleobiology 42,1–7. Lautenschlager S, Gill P, Luo Z-X, et al. (2018) The role of minia- R Core Team (2017) R: A Language and Environment for Statisti- turization in the evolution of the mammalian jaw and middle cal Computing. Vienna, Austria: R Foundation or Statistical ear. Nature 561, 533–537. Computing. URL https://www.R-project.org/. Liu J, Olsen P (2010) The phylogenetic relationships of Ramırez-Chaves HE, Wroe SW, Selwood L, et al. (2016) Mam- Eucynodontia (Amniota: Synapsida). J Mamm Evol 17, 151–176. malian development does not recapitulate suspected key Luo Z-X (1994) Sister-group relationships of mammals and trans- transformations in the evolutionary detachment of the mam- formations of diagnostic mammalian characters. In: In the malian middle ear. Proc R Soc B Biol Sci 283,1–7. Shadow of the Dinosaurs-Early Mesozoic Tetrapods. (eds Fra- Rasskin-Gutman D, Esteve-Altava B (2014) Connecting the dots: ser NC, Sues H-D), pp. 98–130. New York: Cambridge Univer- anatomical network analysis in morphological EvoDevo. Biol sity Press. Theory 9, 178–193. Luo Z-X (2011) Developmental patterns in Mesozoic evolution of Reisz RR (1972) Pelycosaurian reptiles from the Middle Pennsyl- mammal ears. Annu Rev Ecol Evol Syst 42, 355–380. vanian of North America. Bull Mus Comp Zool 144,27–61. Luo Z-X, Crompton AW (1994) Transformation of the quadrate Reisz RR, Berman DS, Scott D (1992) The cranial anatomy and (incus) through the transition from non-mammalian cyn- relationships of Secodontosaurus, an unusual mammal-like odonts to mammals. J Vertebr Paleontol 14, 341–374. reptile (Synapsida: Sphenacodontidae) from the Early Permian Luo Z-X, Chen P, Li G, et al. (2007) A new eutriconodont mam- of Texas. Zool J Linn Soc 104, 127–184. mal and evolutionary development in early mammals. Nature Revell LJ (2012) phytools: an R package for phylogenetic com- 446, 288–293. parative biology (and other things). Methods Ecol Evol 3, 217– Luo Z-X, Schultz JA, Ekdale EG, et al. (2016) Evolution of the 223. middle and inner ears of Mammaliaforms: the approach to Rich TH, Hopson JA, Musser AM, et al. (2005) Independent ori- mammals. In: Evolution of the Vertebrate Ear – Evidence from gins of middle ear bones in monotremes and therians. Science the Fossil Record. (eds Clack JA, Fay RR, Popper AN), pp. 139– 307, 910–914. 174. New York: Springer Handbook of Auditory Research 59. Romer AS (1970) The Chanares~ (Argentina) Triassic reptile fauna. Luo Z-X, Meng QJ, Grossnickle DM, et al. (2017) New evidence VI. A chiniquodontid cynodont with an incipient for mammaliaform ear evolution and feeding adaptation in a squamosaldentary jaw articulation. Breviora 344,1–18. Jurassic ecosystem. Nature 548, 326–329. Rougier GW, Wible JR, Novacek MJ (1996) Middle-ear ossicles of Maier W, Ruf I (2016) Evolution of the mammalian middle ear: a the multituberculate Kryptobaatar from the Mongolian Late historical review. J Anat 228, 270–283. Cretaceous: implications for mammaliamorph relationships Manley GA, Sienknecht UJ (2013) The evolution and develop- and the evolution of the auditory apparatus. Am Mus Novit ment of middle ears in land vertebrates. In: The Middle Ear: 3187,1–43. Science, Otosurgery, and Technology. (eds Puria S, et al.), pp. Rougier GW, Ji Q, Novacek MJ (2003) A new symmetrodont 7–30. New York: Springer Handbook of Auditory Research 46. mammal with fur impressions from the Mesozoic of China. Martinelli AG, Rougier GW (2007) On Chaliminia musteloides Acta Geol Sin 77,7–14. (Eucynodontia: Tritheledontidae) from the Late Triassic of Rubidge BS, Sidor CA (2001) Evolutionary patterns among Argentina, and a phylogeny of Ictidosauria. J Vertebr Paleon- Permo-Triassic therapsids. Annu Rev Ecol Syst 32, 449–480. tol 27, 442–460. Rybczynski N (2000) Cranial anatomy and phylogenetic position Martinelli AG, Soares MB (2016) Evolution of South American of Suminia getmanovi, a basal anomodont (Amniota: Therap- non-mammaliaform cynodonts (Therapsida, Cynodontia). Con- sida) from the Late Permian of Eastern Europe. Zool J Linn Soc trib del MACN 6, 183–197. 130, 329–373. Martinelli AG, Soares MB, Oliveira T, et al. (2017) The Triassic Sidor CA (2001) Simpliﬁcation as a trend in synapsid cranial evo- eucynodont Candelariodon barberenai revisited and the early lution. Evolution 55, 1419–1442. diversity of stem prozostrodontians. Acta Palaeontol Pol 62, Sidor CA (2003) Evolutionary trends and the origin of the mam- 527–542. malian lower jaw. Paleobiology 29, 605–640. McShea DW, Hordijk W (2013) Complexity by subtraction. Evol Sigurdsen T, Huttenlocker AK, Modesto SP, et al. (2012) Biol 40, 504–520. Reassessment of the morphology and paleobiology of the

therocephalian Tetracynodon darti (Therapsida), and the phy- Fig. S7. Organizational modules of the anatomical network of logenetic relationships of Baurioidea. J Vertebr Paleontol 32, Secodontosaurus. 1113–1134. Fig. S8. Organizational modules of the anatomical network of Soares MB, Schultz CL, Horn BLD (2011) New information on Lobalopex. Riograndia guaibensis Bonaparte, Ferigolo & Ribeiro, 2001 Fig. S9. Organizational modules of the anatomical network of (Eucynodontia, Tritheledontidae) from the Late Triassic of Proburnetia. southern Brazil: anatomical and biostratigraphic implications. Fig. S10. Organizational modules of the anatomical network of An Acad Bras Cienc 83, 329–354. Sinophoneus. Spindler F (2015) The Basal Sphenacodontia – Systematic Revi- Fig. S11. Organizational modules of the anatomical network of sion and Evolutionary Implications. PhD thesis, 1–385. Titanophoneus. Sues HD (1986) The skull and dentition of two tritylodontid Fig. S12. Organizational modules of the anatomical network of synapsids from the Lower Jurassic of western North America. Moschops. Bull Mus Comp Zool 151, 217–268. Fig. S13. Organizational modules of the anatomical network of Urban DJ, Anthwal N, Luo Z-X, et al. (2017) A new develop- Suminia. mental mechanism for the separation of the mammalian Fig. S14. Organizational modules of the anatomical network of middle ear ossicles from the jaw. Proc R Soc B Biol Sci 284, Emydops. 20162416. Fig. S15. Organizational modules of the anatomical network of Wang Y, Hu Y, Meng J, et al. (2001) An ossiﬁed Meckel’s carti- Kombuisia. lage in two cretaceous mammals and origin of the mam- Fig. S16. Organizational modules of the anatomical network of malian middle ear. Science 294, 357–361. Arctognathus. Watson DMS (1953) The evolution of the mammalian ear. Evolu- Fig. S17. Organizational modules of the anatomical network of tion 7, 159–177. Aelurognathus. Wible JR (2003) On the cranial osteology of the short-tailed Fig. S18. Organizational modules of the anatomical network of opossum Monodelphis brevicaudata (Didelphidae, Marsupi- Sycosaurus. alia). Ann Carnegie Mus 72, 137–202. Fig. S19. Organizational modules of the anatomical network of Williston SW (1914) Water Reptiles of the Past and Present. Chi- Moschorhinus. cago: The University of Chicago Press. Fig. S20. Organizational modules of the anatomical network of Whaitsia. Supporting Information Fig. S21. Organizational modules of the anatomical network of Tetracynodon. Additional Supporting Information may be found in the online Fig. S22. Organizational modules of the anatomical network of version of this article: Thrinaxodon. Appendix S1. Phylogenetic tree. Fig. S23. Organizational modules of the anatomical network of Table S1. List of synapsid species in the study and literature used Diademodon. for data collection. Fig. S24. Organizational modules of the anatomical network of Table S2. Cranial bones coded and their anatomical role. Lumkuia. Table S3. Results of the modularity analysis. Fig. S25. Organizational modules of the anatomical network of Table S4. Results of the within-module network parameters of Chiniquodon. the jaw-otic network modules. Fig. S26. Organizational modules of the anatomical network of Table S5. Results of the network connectivity (Ki) and between- Probainognathus. ness centrality (BCi) parameters measured for the dentary (d), Fig. S27. Organizational modules of the anatomical network of articular/malleus (art), quadrate/incus (q) and stapes (s) bones. Kayentatherium. Fig. S1. Organizational modules of the anatomical network of Fig. S28. Organizational modules of the anatomical network of Ennatosaurus. Yanoconodon. Fig. S2. Organizational modules of the anatomical network of Fig. S29. Organizational modules of the anatomical network of Varanops. Vilevolodon. Fig. S3. Organizational modules of the anatomical network of Fig. S30. Organizational modules of the anatomical network of Edaphosaurus. Kryptobaatar. Fig. S4. Organizational modules of the anatomical network of Fig. S31. Organizational modules of the anatomical network of Eohaptodus. Monodelphis. Fig. S5. Organizational modules of the anatomical network of Fig. S32. Organizational modules of the anatomical network of Sphenacodon. Canis. Fig. S6. Organizational modules of the anatomical network of Fig. S33. Organizational modules of the anatomical network of Dimetrodon. Homo.