A new viewpoint on antlers reveals the evolutionary history of deer (Cervidae, Mammalia)

Recent molecular phylogeny of deer revealed that the characters of antlers previously focused on are homoplasious, and antlers tend to be considered problematic for classification. However, we think antlers are important tools and reconsidered and analysed the characters and structures to use them for classification. This study developed a method to describe the branching structure of antlers by using antler grooves, which are formed on the antlers by growth, and then projecting the position of the branching directions of tines on the burr circumference. By making diagrams, comparing the branching structure interspecifically, homologous elements (tines, beams, and processes) of the antlers of 25 species of 16 genera were determined. Subsequently, ancestral state reconstruction was performed on the fixed molecular phylogenetic tree. It was revealed that Capreolinae and Cervini gained respective three-pointed antlers independently, and their subclades gained synapomorphous tines. We found new homologous and synapomorphous characters, as the antler of Eld’s deer, which has been classified in Rucervus, is structurally close to that of Elaphurus rather than that of Rucervus, consistent with molecular phylogeny. The methods of this study will contribute to the understanding of the branching structure and phylogeny of fossil species and uncover the evolutionary history of Cervidae.

www.nature.com/scientificreports www.nature.com/scientificreports/ terminology was either not given or confused. In this study, we devised methods to analyse the branching structure of antlers.
Antlers grow extensionally from the tip 39-41 like a plant stem. Plant stems may sometimes twist when growing, but it is possible to recognise twisting from streaks on the epidermis, and we can correctly understand the structure of leaves branching from the stem 42 . Similar to plant stems, by paying attention to the streaks on antlers, which are called "antler grooves" (Sup. Inf. 1- Fig. 5), we think that it is possible to recognise this twisting and understand the branching structure of antlers.
Antler grooves are composed of fine streaks and thick grooves running parallel to them. Fine streaks are thought to be formed by the extension force when the antler grows and can be observed at the tip of a velvet antler (Sup. Inf. 1-Fig. 3). The thick grooves are often observed in the proximal part, which runs almost parallel to the streaks, are traces of arteries, as mentioned in previous studies [43][44][45] (Sup. Inf. 1-Fig. 4). The streaks and arterial impressions are often parallel, but sometimes the impressions may be oblique to the streaks, in which case the fine streaks may indicate the true growth direction). In other words, to know the branching structure based on the growth of the antlers, it may be appropriate to follow the antler grooves, that is, the fine streaks formed by the tensile force of growth and the trace of the arteries parallel to them.
From this new point of view, analysing the branch structure of antlers and determining the homology of tines is expected to link the morphology of antlers and molecular phylogeny and enable the classification of fossil species by the antlers.

Methods
Descriptive methodology. In this study, we analysed the branching structure of antlers by the following method using the grooves.
Branching directions of tines. First, the point of the fork where the target tine branches off the other tine (the point of the maximum curvature on the ridge) is located. Next, the cross section of the target tine taking the opposite point to the fork (the point dividing the circumference into two) is also located; this point indicates the branching direction of the tine. Then, this point is drawn along the antler grooves to the burr, and the branching direction is projected onto the burr. Figure 1A,B shows an example of tine a 2 of Cervus nippon, and Fig. 1C shows the application of multiple tines of the same species. The branching direction of each tine is drawn down to the burr along the antler grooves. It can be projected onto the circle, presenting the cross section of the burr (Fig. 1D).
Positions of forks. The positions of forks where the tines bifurcate can be determined using the antler grooves. The point of the fork (at the maximum curvature point) is traced along the maximum curvature points of the antler grooves running parallel to the ridge of the fork. Then, when it reaches the antler grooves toward the burr, it is drawn down to the burr along them (Fig. 2B). Figure 2A shows an example of a 1 /p 1 , the fork where tines a 1 and p 1 branch ("/" represents a fork), and an example of a 2 /p 2 . For one fork, the line can be placed on the lateral or medial side; therefore, in the cross section of the burr, two positions are determined for one fork (Fig. 2C).
In addition, areas bounded by forks are taken as the "zone" of the tines (Fig. 2D). This is because the tines in which certain antler grooves reach distally from the burr are bound differently by the fork. This zone can be represented as shown in Fig. 2E in the cross section of the burr.
When overlapping with a more proximal tine. It is no problem when branching directions of tines or positions of forks come down to the burr, but often, they overlap with more proximal tines and do not reach the burr. In this case, we focused on the antler grooves on the proximal tine that overlaps. There are two flows of antler grooves in the proximal direction on the tine, one running to the burr and the other running to the other tine. In addition, a flow boundary exists. We named this the "groove-boundary" (Fig. 3). When overlapping (example of the branching direction of a tine in Fig. 3A), moving the point α(before) to the new point α(after) is prorated by centring the groove-boundary in the following manner: [length between the ridge and α(before) = x]: [length between α(before) and groove-boundary = y] = [length between the opposite side of the ridge and α(after) = X]: [length between α(after) and the groove-boundary = Y]. Then, α(after) is taken down to the burr. In the case of forks, this was the same (Fig. 3B).
The purpose of centring is to ensure that even when the overlapping proximal tine disappears, the branching direction of the tine or the positions of the fork will come down to the same points on the burr. The reason for performing proportional allotment is because it is presumed that antler grooves leading from the burr to the tine will fold back to more distal tines. Sup. Inf. 1- Fig. 6 is an example of this. In these antlers, the antler grooves leading to tine G and tine H turn back to the more distal tines. Thus, it is presumed that antler grooves leading to a certain tine pass through the tine, turning back and running to more distal tines. Therefore, we think this proportional allotment is appropriate so that the branching direction of the tine or the positions of the fork come down to the same points on the burr.
Positional relation with the skull. Antlers grow from the pedicles, protrusions of the posterior lateral area of the frontal bone, which is a part of the skull. Therefore, to determine the true branching structure of the antler, it is necessary to know how it is positioned on the skull; that is, to clarify the relative positional relation between the branching direction of the tines and positions of the forks, and some structural indicators of the skull. Therefore, several structures were used as indices.
In the frontal area (Fig. 4A), the ridge extending from the supraorbital region to the burr on the pedicle can be used as an index. We named it the supraorbital ridge on the pedicle (SR). In addition, the medial branch of the frontal branch of the superficial temporal artery, sending blood to the antler 46,47 , often makes an impression on the pedicle, which can also be used as an index. This is abbreviated as IFST.
www.nature.com/scientificreports www.nature.com/scientificreports/ In the temporal area (Fig. 4B), the ridge extending from the retro-orbital region to the burr on the pedicle can be used as an index. We named it the temporal ridge on the pedicle (TR). In addition, the frontoscutular muscle and the interscutular muscle adhere at intervals from the temporal ridge on the pedicle, with their lower edges bordering the temporal line 48 . Furthermore, considering the branching directions of the tines and the positions of the forks to extend along the impressions of the nerves 49 , thin streaks run on the pedicle. In the case of a specimen of Cervus nippon shown in Fig. 4B, a 1 /p 1 is extended along the nerve impressions to the superior-posterior margin of the orbit, which is abbreviated as SPO. In the same way, a 2 /p 2 is extended through the lateral margin of the origin of the frontoscutular muscle, to the posterior margin of the orbit, which is abbreviated as LMOF. p 1 is extended to the posterior extremity of the origin of the interscutular muscle (PEOI).
In addition, the nerves running to the antler on the surface of the pedicle, which are derived from the trigeminal nerve, come from two areas: the supraorbital area and the temporal area. Nerves from the supraorbital area are through the medial side and those from the temporal area are through the lateral side 49 (Fig. 4C). The boundary between them (hereafter abbreviated as BN) is on the posterior side of the pedicle (Fig. 4D), and is determined by the impression of nerves. Figure 4E shows the position of the indices with relation to the skull on the burr cross section.
Diagram. We created a diagram to represent the branching structure information of antlers obtained by the above methods in one figure. Determining the homologous tine and terminology. In this study, we compared, among the species, the positional order of the branching directions of tines, the positions of forks, the branching hierarchical www.nature.com/scientificreports www.nature.com/scientificreports/ positions of tines, and indices that determined the positional relation with the skull, on the diagrams. Tines at the same position on the diagram (have the same structure) were determined to be homologous.
We gave terminologies to the homologous tines. Given terminologies were based on previous literature as much as possible; however, new names were made for the tines with no terminologies in the literature. "-tine" and "-beam" were given to those whose length is normally more than twice the basal diameter, and "-process" was used for those normally less than twice the basal diameter. Although "-tine" and "-beam" are named, they are not conceptually distinguished in this study, because a tine inevitably becomes a beam if the top of it bifurcates. www.nature.com/scientificreports www.nature.com/scientificreports/ Reconstructing ancestral character states. After determining the homology of tines, we reconstructed the ancestral states of all tines, treating them as characters. The reconstruction was performed using the fixed topology of the most recent molecular phylogenetic tree, the Bayesian inference analyses of the combined molecular data set by Heckeberg 12 .
We set the character states 2, 1, and 0. When the tine was observed in 80% or more of the observed specimens of the species, the character state was 2. When observed in less than 80%, the character state was 1. When observed in no specimens, the character state was 0.
Between character states 0 and 1, reconstruction was performed by Dollo parsimony [50][51][52] . This parsimony is the method under the constraint that derived characters are gained only once, respectively, based on Dollo's law 53 that complicated derived characters do not evolve twice and it is never gained again once it is lost. In the evolution of antlers in Cervidae, it is presumed that a homologous tine with the same complex structure (having the same www.nature.com/scientificreports www.nature.com/scientificreports/ branching direction, fork position, and branching hierarchical position and at the same position in the indices determining the positional relation with the skull) is not doubled. This is because it is presumed that the genes expressing the homologous tine may not be acquired twice, but on the other hand, it can be lost forever if the gene is lost in the population.
Between character states 1 and 2, reconstruction was performed using standard parsimony (ACCTRAN). In evolutionary history, a tine that was newly acquired due to a mutation in genes is, if it is adaptive, likely to spread rapidly within the population after acquisition (character state 1 → 2). In contrast, if the tine is no longer adaptive in the offspring population, the rate within the population will decrease (character state 2 → 1). If a new tine is acquired by mutation and is adaptive, it should spread in a short time in the population. Therefore, when 0 → 1 occurs, it is likely that 1 → 2 occurs immediately after. It is unlikely that 1 → 2 will evolve many times in evolutionary history (though this is not without possibility that it will happen several times). In contrast, 2 → 1 is likely to occur independently after the offspring diverges when the tine is no longer adaptive. Therefore, we applied ACCTRAN. www.nature.com/scientificreports www.nature.com/scientificreports/ Materials. The species we observed in this study are listed in Table 1. The ancestral reconstruction was performed using adult specimens, the numbers of which are also shown in Table 1. Sixteen out of the eighteen extant genera 54 are included. Only Rucervus schomburgki is currently extinct. Hydrophotes intermis, which is known to have no antlers, was also included in the samples to ensure that it had none. Species names are based on Wilson and Reeder 54 , except for Eld's deer, which was revealed to belong to the different clade from Rucervus duvaucelii and R. schoburgki by multiple molecular phylogenic studies [8][9][10][11][12] . Therefore, we used Panolia 8,55 for the genus containing Eld's deer in this study. All specimens examined belonged to public museum/university collections. The specimen numbers and collection sources are listed in Sup. Inf. 9 (and 6).

Results
Homologous elements. In this study, we determined more than 40 homologous elements (tines, beams, and processes). The main conclusions are as follows (See Fig. 6).
At the second fork of Rusa, Axis, Cervus, and Dama, a 2 is posterior to the TR and p 2 reaches the PEIO. a 2 /p 2 reaches the LMOF through the pedicle. Therefore, a 2 and p 2 are homologous among these genera. a 2 is named the trez tine based on many previous studies 16,19,22,23,25,26,[28][29][30] . p 2 is called the beam or main beam in previous studies 14,23,26,[28][29][30] , and we named it the lower beam. The trez tine was observed in almost all specimens in Rusa, Axis, Cervus, and Dama, and rarely in Elaphurs.
At the second fork in Capreolus and Rangifer, a 2 is posterior to SR and is slightly anterior to lateral a 1 /p 1 (=brow tine/lower beam). p 2 is slightly medial to the BN. The lateral a 2 /p 2 reaches the LMOF through the pedicle, and the medial a 2 /p 2 is at the same position as the medial a 1 /p 1 (=brow tine/lower beam). Therefore, a 2 and p 2 are homologous among these two genera. In contrast, a 2 and p 2 of these genera have different structures from the above-described a 2 (=trez tine) and p 2 (=higher beam) of Rusa, Axis, Cervus, and Dama; they have different positional orders of the branching direction and fork positions, as well as differences in the indices determining the positional relations with the skull. (For example, a 2 of Rusa, Axis, Cevus, and Dama is posterior to lateral a 1 / p 1 [=brow tine/lower beam], whereas a 2 of Capreolus and Rangifer is anterior to lateral a 1 /p 1 [=brow tine/lower beam]). Therefore, they are not homologous. We gave new names to a 2 and p 2 of Capreolus and Rangifer. a 2 was named the frontal tine based on previous usage 56 , and p 2 was named the upper beam, because several examples in the literature 15,28,56 termed this part the "beam" in Rangifer. In addition, in Blastocerus, a 1 is between the SR and TR, p 1 is slightly medial to the BN, and lateral a 1 /p 1 reaches LMOF through the pedicle. Therefore, a 1 and p 1 of Blastocerus are the same as a 2 and p 2 of Caprelus and Rangifer, respectively. That is, a 1 and p 1 of Blastocerus are termed the frontal tine and upper beam, respectively. The frontal tine and upper beam were observed in almost all specimens in Rangifer, Capreolus, and Blastocerus, and very rarely in Odocoileus.
In Cervus canadensis and Cervus elaphus, x is between the TR and a 1 /p 1 (=brow tine/lower beam). x/p 1 (=x/ brow tine) is at the same position as x. There are no other tines that have this structure in the typical antlers of the other species. This tine was named the bez tine based on its frequent usage in the literature 14,19,22,[28][29][30] . The bez tine was observed in all specimens of C. elaphus and C. canadensis, and rarely in Dama and Rusa unicolor.
In Panolia eldii, p 2 is at the BN and lateral a 2 /p 2 is at the same position as p 1 (= lower beam). This structured tine is not seen in the typical antlers of other species. This tine is called the medial tine. The medial tine was observed in 100% of P. eldii and rarely in Elaphurus davidianus.
In Rangifer, p 3 is at the BN and a 3 is at the same position as a 2 (=frontal tine). In Blastocerus, p 2 is at the BN and a 2 is at the same position as a 1 (=frontal tine). Therefore, p 3 of Rangifer and p 2 of Blastocerus are homologous; it was named the rear tine, based on previous usage 56 . This tine was observed in Odocoileini.
In this way, the homology of most antler elements (tines, beams and processes) was determined and the terminologies of tines were provided. The various types of antlers observed in this study are shown in Sup. Inf. 3. A www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ detailed explanation of each homologous tine is provided in Sup. Inf. 4. A comparison between the homologous tines identified in the study and those described by Pocock 33 , as well as a comparison between the names given to tines in this study and those used previously in the literature, is shown in Sup. Inf. 5. The matrix of existence or nonexistence of the tines, beams, and processes of all specimens observed in this study is provided in Sup. Inf. 6. The matrix of percentages of the antler elements (tines, beams, and processes) of the species is shown in Sup. Inf. 7.
Ancestral states and evolution. Sup. Inf. 8 shows the results of the reconstruction of ancestral character states, from the fixed topology of Heckeberg 12 , of the respective elements (tines, beams, and processes) based on the homology determined in this study. Figure 7 shows the reconstructed ancestral states (depicted as silhouettes) at the nodes and the evolutionary history of the antlers.
The most recent common ancestor (MRCA) of the extant species had two-pointed antlers with the brow tine and the lower beam. In Mutiacus, the primitive two-pointed antlers were retained.
The common ancestor of Cervini possessed a brow tine, lower beam, trez tine, higher beam, and brow process. Axis retained primitive three-point antlers. Rucervus lost its trez tine and higher beam and gained vertical and back beams and distal tines. The common ancestor of Dama, Elaphurus, Panolia, Rusa, and Cervus gained the potential ability to express a crown-inner tine, crown-outer tine, and crown-back tine, which form a trifurcation ("crown") at the end of the higher beam. Dama gained a guard tine and palmation of the distal portion of the higher beam. The clade of Elaphurus davidianus and Panolia eldii gained a medial tine and pre-cacuminal tines. Panolia eldii gained a cacuminal tine and Elaphurus gained pre-trez tines and the bifurcation of the brow tine. Cervus makes different forms for each species depending on the manner of existence of the distal tines on the higher beam. C. nippon lost its bez tine.
In Capreolinae, the common ancestor was typically three-pointed with the brow tine, lower beam, frontal tine, and upper beam, and Capreolus almost retained this form. Hydropotes lost their antlers completely. The clade of Alceini and Odocoileini gained the ability to express terminal-anterior/posterior tines. Alces lost its brow tine and frontal tine. Odocoileini gained a rear tine. In Odocoileus, the brow and frontal tines were not normally expressed, and upper-1 st , 2 nd , and 3 rd tines were obtained.

Discussion
Arrangement of the terminology of homologous elements. As a result of this study, the correspondence between homologous elements and previous terminology was clarified. For example, the "posteromedial tine" 27 or "back-inner tine" 21 and "anterolateral tine" 27 or "front-outer tine" 21 in Rusa and Axis are homologous to the "trez tine" 16,19,22,23,25,26,[28][29][30] and posterior portion of the "beam" 14,23,26,[28][29][30] in Cervus elaphus, respectively. We named the former the trez tine and the latter the higher beam. Further, the "bez tine" 28 of Rangifer tarandus was not homologous to the "bez tine" 14,19,22,[28][29][30] of Cervus. We named the former the bez tine and the latter the frontal tine. In this study, we also gave terminologies to the elements that have not been given them before. It can be said that a measurable arrangement was performed for the identification of these antler elements in the future. (See Sup. Inf. 5 for the correspondence between terminologies in this study and those previously used).

Different three-pointed structures between Cervini and Capreolinae.
This study revealed that three-pointed structures are different between Cervini and Capreolinae. Compared to the basic three-pointed forms of Cervini and Capreolinae, the brow tine and lower beam are homologous, but the tines bifurcating at the end of the lower beam were the trez tine and higher beam in Cervini, and the frontal tine and upper beam in Capreolinae. The view of identifying the homology of the tines at the second fork between Cervini and Capreolinae 32,33,57 needs to be dismissed completely. At the same time, the MRCA of the extant species was revealed to have been two-pointed. Although it is known that primitive fossil cervids, such as Procervulus and Dicrocerus, are two-pointed, it was revealed that the antler of the MRCA of extant species was still two-pointed at that stage. The presumption that the MRCA was three-pointed 9 was contradicted in this study.
Panolia eldii has a different structured antler to Rucervus. Eld's deer belongs to Rucervus 35,54 , but all recent molecular phylogenetic topology studies [8][9][10][11][12] show that the position of this species is not in the same clade www.nature.com/scientificreports www.nature.com/scientificreports/ as Rucervus duvaucelii and R. schomburgki, but in a clade with Elaphurus davidianus; therefore, Pitra et al. 8 used the genus Panolia, and this study follows it. Analysis of the antler in this study revealed that the branching structure of the antler of Panolia eldii was completely different from that of Rucervus, and was closer to that Elaphurus, reflecting the molecular phylogeny.
The p 2 branching direction (in Pocock 33 ) of Panolia eldii, named the medial tine in this study, is at the BN, and the branching direction of its counterpart a 2 , the lower beam, is posterior to the TR. The medial tine/lower beam (the position of the fork) (lateral side) is at the same as the branching direction of the lower beam (to the brow tine). In contrast, in Rucervus, the branching direction of p 2 (in Pocock 33 ), the back beam (in this study), is lateral to the BN, and its counterpart a 2 , the vertical beam, is anterior to the TR. The vertical beam (a 2 )/back beam (p 2 ) (the position of the fork) is not the same as the branching direction of the lower beam. Examination of the other tines does not reveal any homologous elements except for the brow tine and lower beam. The branching structure of the antler of Panolia eldii, which has been found in Rucervus owing to its unique curved antler form, was quite different, which is consistent with the molecular phylogeny.
On the other hand, it was revealed that there was plural synapomorphy between Panolia eldii and Elaphurus davidianus, which are sister taxa by molecular phylogeny.
The medial tine (100% in P. eldii and 11% in E. davidianus), pre-cacuminal tines (40% in P. eldii and 33% in E. davidianus), and loss of the trez tine (100% in P. eldii and 94% in E. davidianus) (Sup. Inf.  are synapomorphies in the two genera. crown tines of Cervus and Dama. The study also provided a new view into the distal tines of Cervus and Dama. The distal tines of Cervus elaphus are often termed "crown (tines)" 25,28,30 . This terminology is also used in Cervus canadensis 29 . However, the distinction between the tines in the crown has not yet been clarified, and an inter-specific comparison of the inside tines of the crown has not yet been performed. The most complex antler in Cervus is that of C. elaphus, whose typical antler trifurcates at the distal end of the higher beam. We named these three tines the crown-inner tine, crown-outer tine, and crown-back tine. The crown-back tine sometimes trifurcates at the distal end, and we named these three tines the crown-back-inner tine, crown-back-outer tine, and crown-back-back tine (Sup. Inf. . As a result of the comparison between C. elaphus and C. canadensis, it was revealed that the distal tines at the end of the higher beam in the typical antler of C. canadensis were the crown-inner tine and crown-back tine, and the crown-back-inner tine and crown-back-back tine distally branching from the crown-back tine (Sup. Inf. 3-Fig. 14). Furthermore, in the typical antler of Cervus nippon, it was revealed that the distal tines at the end of the higher beam were the crown-inner tine and crown-outer tine (Sup. Inf. 3- Fig. 15). As evidence, trifurcation containing the crown-back tine is sometimes seen in C. nippon (Sup. Inf. 3- Figs. 16 and 17). In Cervus, the recognition of homology of the brow tine, bez tine, trez tine (or tres tine), beam and crown (or cornet, or royal) tines has been well established 14,19,22,23,29 , but the structure and homology of the inside tines of the crown have not been studied at all. To the best of our knowledge, this study solved the complex inside structure of the crown of Cervus and the homology of the tines inside it for the first time.
In addition, this study revealed the distal structure of the antler of Dama dama. The distal portion of the adult antler of Dama dama is palmate, whereas that of the juvenile antler is not palmate. Analysis of the juvenile antler identified the crown-inner tine and crown-back tine (Sup. Inf. 3-Fig. 20). In addition, analysis of adults revealed that the palmate parts are mainly the crown-inner tine and crown-back tine (Sup. Inf. 3-Figs. 21-23). As evidence, some mutant antlers have a crown-outer tine (Sup. Inf. 3- Fig. 24). (In addition, Dama dama has an autapomorphy, termed the guard tine, at the bottom of the palmation. The terminology "guard tine" is used by the Australian Deer Association 58 ). For the antler of Dama dama, like in Cervus, wide recognition of homology of the proximal portion of the antler exists with Cervus species for the brow tine, bez tine, trez tine, and beam 16,20,24,26 , but regarding the distal portion, palmation needs to be focused upon, and homologous elements with other species are not been known. The new viewpoint of this study has revealed that the branching structure of the distal portion of Dama dama is homologous to that of Cervus. curious antlers in capreolinae: especially those in Alces and Odocoileus. Comparisons of the antlers of Alces and Odocoileus with those of Capreolus, Rangifer, and Blastocerus have been seldom made 33,37,38 but there is a poor unified view of homology and terminologies. In this study, we revealed the remarkable homology of Capreolinae antlers.
First, a 1 of Blastocerus is the frontal tine, which is not homologous to a 1 = brow tine of Rangifer and Capreolus but was homologous to a 2 . Therefore, it was revealed that the brow tine was lost in Blastocerus.
For Alces alces, it was revealed that the palmate portion of its antler is homologous to the distal portion of that in Rangifer tarandus. Photographs of the specimens provided in Sup. Inf. 3- Fig. 41 show the similarity between the two (the distal portion of the antler of Rangifer is often palmate like Alces). We named this portion the terminal-anterior/posterior tines based on the previously used terminology, "terminal tines" 15 . Therefore, in Alces, the brow tine and frontal tine were completely lost, and only the terminal-anterior/posterior tines were enlarged. This study has identified the great peculiarity of the antler of Alces.
This study also revealed that Odocoileus has a very specific structured antler. It was difficult to recognise the homologous correspondence to the other genera from the antler of Odocoileus when we observed only the typical antler. The clue came from the NSMT-M32362 specimen of O. virginianus in the National Science Museum (Sup. Inf. . This specimen has a very abnormal left antler. The typical antler part corresponds to the posterior portion of this abnormal one. By comparing to the antler of Rangifer tarandus, it was revealed that this abnormal antler has both a brow tine and frontal tine. It almost trifurcates (strictly bifurcates twice) at the proximal part, the anterior branch is the brow tine, the middle is the frontal tine, and the posterior is the upper beam (lower beam is very short). It was revealed that the first anterior tine of a typical antler of Odocoileus, which is often called www.nature.com/scientificreports www.nature.com/scientificreports/ the "sub-basal snag 59,60 ", is homologous to the posterior branching tail-like tine in Rangifer (rear tine). Therefore, in Odocoileus, the brow tine and frontal tine are completely lost normally, the whole antler is significantly twisted, and the rear tine, which branches off posteriorly in Rangifer and Blastocerus, branches anteromedially at the relatively proximal part of the antler. It became clear that the lower beam was very short and the upper beam, from which autapomorphic tines (upper-1st, 2nd, and 3rd tines) branch off, extended greatly.
Thus, it was uncovered that Odocoileus has a very specific and surprisingly peculiar antler that loses proximal tines and is entirely twisted.
Application to the phylogeny of fossil species. In this study, we clarified the structural differences between the three-pointed antlers of Cervini and Capreolinae. Comparing the antler of a fossil species with the basic branching structure of antlers of both clades will help to determine which clade the fossil species belongs to.
Moreover, as in the case of Panolia eldii (which was traditionally classified in Rucervus based on the external form of the antler but was reconfigured into the same clade as Elaphurus by molecular phylogeny), in this study, the antler was revealed to be different in structure from Rucervus and have some synapomorphies with Elaphurus, in accordance with the molecular phylogeny. There is a great possibility that, from a new point of view using the antler grooves of this study, the systematic position of the fossil species that had been previously classified only by the external form of their antlers could be reconfigured.
As it was uncovered that the palmate portion of the antler of Alces was in fact homologous to the distal portion of that of Rangifer, by re-examining, from the viewpoint of this study, the structure of the distal portion of the fossil antlers that had not been noticed so far, the phylogenetic classification could be revisited. Naturally, the viewpoint of this study will also contribute to the taxonomy and phylogeny of newly generated fossil species. The new point of view in this study has the potential to clarify the evolutionary history of the whole Cervidae family, including fossil species.