Convergent Evolution of Locomotory Modes in Euarchontoglires

The research of phenotypic convergence is of increasing importance in adaptive evolution. Locomotory modes play important roles in the adaptive evolution of species in the Euarchontoglires, however, the investigation of convergent evolution of the locomotory modes across diverse Euarchontoglire orders is incomplete. We collected measurements of three phalangeal indices of manual digit III, including metacarpal of digit III (MC3), manus proximal phalanx of digit III (MPP3), and manus intermediate phalanx of digit III (MIP3), from 203 individuals of 122 Euarchontoglires species representing arboreal (orders Scandentia, Rodentia, and Primates), terrestrial (orders Scandentia and Rodentia), and gliding (orders Dermoptera and Rodentia) locomotory modes. This data can be separated into seven groups defined by order and locomotory mode. Based on combination of the three phalangeal indices, the Principle component analyses (PCA), phylomorphospace plot, and C-metrics analyses clustered the arboreal species of Scandentia, Rodentia, and Primates together and the terrestrial species of Scandentia and Rodentia together, showing the convergent signal in evolution of the arboreal (C1 = 0.424, P < 0.05) and terrestrial (C1 = 0.560, P < 0.05) locomotory modes in Euarchontoglires. Although the gliding species from Dermoptera and Rodentia did not cluster together, they also showed the convergent signal (C1 = 0.563, P < 0.05). Our work provides insight into the convergent evolution of locomotory modes in Euarchontoglires, and reveals that these three indices contribute valuable information to identify convergent evolution in Euarchontoglires.


INTRODUCTION
The relationship between morphology, locomotory modes and habitat has been receiving greater attention in evolutionary studies of phenotypic variation (Losos and Sinervo, 1989;Losos, 1990;Bonine and Garland, 1999;Vanhooydonck and Van Damme, 1999; Thorington and Santana, 2007;Goodman et al., 2008;Irschick et al., 2008;Pfaff et al., 2015;Kawashima et al., 2018). It has been observed that differences in habitat trigger divergence in the evolution of the morphology of locomotory systems across many different taxonomic taxa (Losos, 1990;Runestad and Ruff, 1995;Irschick et al., 2005;Calsbeek and Irschick, 2007;Sargis et al., 2007;Samuels and Van Valkenburgh, 2008;Schmidt, 2008), and that convergent evolution of locomotory modes for the same habitat niche adaptation in phylogenetically unrelated taxa can lead to similar morphological traits (Lemelin and Grafton, 1998;Kingsolver and Huey, 2003;Samuels and Van Valkenburgh, 2008;Gleiss et al., 2011;Losos, 2011;Edwards et al., 2012;Chen and Wilson, 2015;Morris et al., 2018;Grossnickle et al., 2019Grossnickle et al., , 2020. The morphological convergence provides clues to better understanding functional adaptation and how distantly related species evolved in a similar way. The mammalian superorder Euarchontoglires is comprised of five orders and includes Lagomorpha (rabbit, pika, and hare), Rodentia (mouse, rat, squirrel, and beaver), Dermoptera (flying lemur), Scandentia (tree shrew), and Primates (human, monkey, and ape) (Murphy et al., 2001;Kriegs et al., 2006;Vander Linden et al., 2019). A great diversity in locomotory modes is observed among and within the Euarchontoglire orders, including fossorial, ricochetal, arboreal, terrestrial, and gliding (Macphee, 1993). Interestingly, the same locomotory modes are found in different orders of Euarchontoglire, including the arboreal mode in Scandentia, Rodentia, and Primates, the terrestrial mode in Scandentia and Rodentia, and the gliding mode in Dermoptera and Rodentia (Kirk et al., 2008;Meng et al., 2017;Vander Linden et al., 2019;Grossnickle et al., 2020; Figure 1). The sharing of locomotory modes among orders makes Euarchontoglires an intriguing group for the investigation of morphological convergence with locomotory modes. Previous studies with Euarchontoglires have focused on the convergence of forelimb evolution with several locomotory modes, but these studies only examined species within Rodentia (Samuels and Van Valkenburgh, 2008) or on skeletal convergence with the gliding mode in Dermoptera and Rodentia (Grossnickle et al., 2020). Therefore, knowledge on the convergent evolution of locomotory modes across Euarchontoglire orders is very limited and it remains unclear whether the underlying skeletal morphology of terrestrial and arboreal species from the different Euarchontoglire orders have converged.
In the present study, we collected three phalangeal indices of manual digit III from 203 individuals of 122 Euarchontoglires species that represent the arboreal (orders Scandentia, Rodentia, and Primates), terrestrial (orders Scandentia and Rodentia), and gliding (orders Dermoptera and Rodentia) locomotory modes to investigate the convergent evolution of locomotory modes across these Euarchontoglire orders. The three phalangeal indices include metacarpal of digit III (MC3), manus proximal phalanx of digit III (MPP3), and manus intermediate phalanx of digit III (MIP3), which represent the inherent properties of the manus (Corruccini, 1995;Hamrick, 2001;Bloch et al., 2007;Kirk et al., 2008) and have been used to illustrate the locomotory modes of mammals (Lemelin, 1999;Hamrick, 2001;Meng et al., 2006Meng et al., , 2017Bloch et al., 2007;Kirk et al., 2008;Samuels and Van Valkenburgh, 2008;Venkataraman et al., 2013;Zheng et al., 2013;Chen and Wilson, 2015;Grossnickle et al., 2020;Young and Chadwell, 2020). Our morphometric analyses based on quantitative methods provide insights into the convergent evolution of locomotory modes in Euarchontoglires, and moreover, evaluate the potential of these indices for studies on the convergent evolution of locomotory modes in Euarchontoglires.

Data Collection
Publicly available data for the three phalangeal indices of manual digit III, including MC3, MPP3, and MIP3, from a total of 152 individuals from 122 Euarchontoglire species belonging to the orders Scandentia, Rodentia, Dermoptera, and Primates were collected from published sources (Kirk et al., 2008;Samuels and Van Valkenburgh, 2008;Chen and Wilson, 2015;Meng et al., 2017). These species represent arboreal (99 individuals from 83 species of Scandentia, Rodentia, and Primates), terrestrial (33 individuals from 28 species of Scandentia and Rodentia), and gliding (20 individuals from 11 species of Dermoptera and Rodentia) locomotory modes. In addition, three phalangeal indices of manual digit III were measured from 51 individuals of the gliding complex-toothed flying squirrel (Rodentia, Trogopterus xanthipes) that were bred in the Runxing Planting and Breeding Cooperative, Liquan County, Xi'an, Shaanxi Province in the present study and were added to the relatively limited data available to represent the gliding locomotory mode. The three phalangeal indices measured using the electronic digital indicator for Trogopterus xanthipes were shown in Supplementary Figure S1. Chordal nodes for each bone element were measured from the nearest to the farthest end, and from the parallel to the midline of the axis on the sagittal plane based on the methods of Kirk et al. (2008). In total, three phalangeal indices from 203 individuals of 122 species were collected for the present study (Supplementary Table S1). Individuals were divided into seven groups for analyses according to their taxonomic order and their locomotory mode. The groups are: Rodentia Gliding (RG; 64 individuals of 9 species), Dermoptera Gliding (DG; 7 individuals of 2 species), Scandentia Terrestrial (ST; 8 individuals of 3 species), Rodentia Terrestrial (RT; 25 individuals of 25 species), Scandentia Arboreal (SA; 6 individuals of 4 species), Rodentia Arboreal (RA; 34 individuals of 27 species), and Primate Arboreal (PA; 59 individuals of 52 species). For the present analysis, the MC3, MPP3, and MIP3 indices used were the percentage of the length of each indices of the total combined length for these three elements, i.e., MC3(p), MPP3(p), and MIP3(p), as previously used (Corruccini, 1995;Elissamburu and Vizcaino, 2004). Statistical differences within the 3 indices among the different groups were detected using the Wilcoxon test.

Morphometric Analyses
To avoid the possible bias of morphological variations resulted from the unbalanced individual numbers for each species, we calculated the average values per species to perform the morphometric analyses based on species level. First, a principle component analyses (PCA) was conducted using the FactoMineR and factoextra packages in R (Version 3.6.2) to visualize the overall distribution of the groups in the morphospace.
Then, taking into the phylogenetic effect, the principal components of the Procrustes coordinates were projected in the phylomorphospace plot by using the phytools package in R (Version 3.6.2). A well-accepted mammalian species tree of 122 species in our study from TimeTree database 1 was used as the input tree (Figure 1). This plot could detect the convergence based on the length and direction of branches in relation to their ancestral states (Feijó et al., 2020).
To statistically quantify the magnitude of convergence for the three locomotory modes from different orders, i.e., the gliding (RG and DG), arboreal (RA, PA, and SA) and terrestrial (ST and RT) modes, the distance-based metric (C1-C4) were calculated (Stayton, 2015) by using the "convrat" function from convevol package in R (Version 3.6.2). C1 is the distance between the convergent tips of the taxa of interest (Dtip) divided by the maximum distance between any ancestral nodes (Dmax), which represents the proportion of the maximum distance between two lineages. This change of this value from 0 to 1 means the degree of convergence signal from non-convergence to convergence. C2 is obtained by Dtip minus Dmax and represent the magnitude of convergent change. C3 and C4 are the standardized value of C2 which were calculated by dividing C2 by the total morphological branch length of convergent taxa of interest (C3) and the total amount of branch length in the entire clade (C4) (Stayton, 2015), respectively. The significance of the metrics was assessed by the "convratsig" function of the same package with 1000 iterations.
A PCA of the three indices [MC3(p) + MPP3(p) + MIP3(p)] showed that PC1 explained 73.6% of variation and separated the terrestrial species of Scandentia and Rodentia (ST and RT groups) from the arboreal and gliding species (RA, PA, SA, RG, and DG groups) (p < 0.05). PC2 explained 26.4% of variation and separated the gliding species of Dermoptera (DG group) from the other species examined (p < 0.05) (Figure 2).
As seen from the phylomorphospace plot, the terrestrial species from Scandentia (ST group) and Rodentia (RT group) clustered together, suggesting that their similar phalangeal morphology may contribute to adapt to the same ground habitat. Likewise, the arboreal species from Scandentia (SA group), Rodentia (RA group), and Primate (PA group) clustered together, suggesting that their similar phalangeal morphology help them adapt to their grasping behavior. These results indicate the similarity of these morphological traits in the terrestrial and arboreal locomotory modes in Euarchontoglires. Interestingly, we found that the gliding species from Rodentia (RG group) were located between the arboreal species (RA, PA, and SA groups) and the gliding species from Dermoptera (DG group), consistent with the observation that the RG group species have evolved their morphological characteristics by not only adapting to tree-dwelling and branch-grasping life, but also to the gliding movement (Samuels and Van Valkenburgh, 2008;Grossnickle et al., 2020). The gliding species from Dermoptera are most distantly separated from the other species. Although the gliding species of Dermoptera (DG group) did not cluster with those from Rodentia (RG group) in the phylomorphospace plot (Figure 2), these two groups possess the same gliding locomotory mode and show a trend of closer proximity.
The C-metrics analyses of all three locomotory modes showed statistically significant convergence [gliding (DG and RG), C1 = 0.563, P < 0.05], arboreal (RA, PA, and SA), C1 = 0.424, P < 0.05) and terrestrial (ST and RT), C1 = 0.560, P < 0.05)], which indicate that evolution has reduced the distances in the gliding species and terrestrial species by showing more than 50% similarity than expected by chance, and in the arboreal species by more than 40% (Table 1). In addition, we also explore  The numbers of species used in each locomotory modes were given in parentheses. The "*" represents significant P-values among each locomotory modes with P < 0.05 generated from 1,000 simulations.
the potential species that show more signals of convergence for the three locomotory modes, as used by the method of Stayton (2015). The phylomorphospace plot analyses and the C-metrics analyses identified and quantified at least 72 species in total showing more signals of convergence for the three locomotory modes, respectively (Supplementary Table S3 and Supplementary Figure S2). Therefore, our morphometric analyses based on quantitative methods provide insight into the convergent evolution of terrestrial, arboreal, and gliding locomotory modes in Euarchontoglires. In our analyses, the terrestrial species (ST and RT groups) demonstrate a relatively long metacarpal (MC3) and short phalange (MPP3 and MIP3) compared with the arboreal and gliding species (P < 0.05) (Figure 3), which would help increase their speed for hunting or feeding on the ground (Peng et al., 1991;Kirk et al., 2008;Samuels and Van Valkenburgh, 2008;Fuchs and Corbach-Söhle, 2010). In addition, previous studies have reported the possible genetic mechanisms at the gene level involving muscle contraction and skeletal morphogenesis for locomotory adaptation of terrestrial in Scandentia group (Fan et al., 2014). Our study not only confirmed the long metacarpal and short phalange in the terrestrial species of both Scandentia and Rodentia, but also demonstrated convergent evolution with the terrestrial locomotory mode in these two orders of Euarchontoglire.
In comparison, the arboreal species (RA, PA, and SA groups) show an elongated phalange MPP3 and MIP3 compared with terrestrial species (P < 0.05) (Figure 3), which would help them grasp branches while they are moving. This phenomenon has been demonstrated in arboreal Primates (Lemelin, 1999;Hamrick, 2001;Bloch et al., 2007;Kirk et al., 2008;Young and Chadwell, 2020) and Rodentia (Kirk et al., 2008;Samuels and Van Valkenburgh, 2008). Our study not only observed this in the arboreal species of Scandentia, Rodentia, and Primates, but also demonstrated convergent evolution with the arboreal locomotory mode in these three orders of Euarchontoglire.
The gliding species (DG and RG groups) demonstrated an elongation of the phalangeal MIP3 compared with arboreal and terrestrial species (Figure 3), which is likely associated with their unique attachment of the patagia to adapt the gliding locomotory habitat (Macphee, 1993) or to their roosting behavior (Grossnickle et al., 2020). Although they both possessed patagium, but they are different in that the former shows the patagium attachment by the styliform cartilages from the wrist or elbow while the latter's patagium are attached to the digits (Jackson, 2012). Further examination of the gliding species from these two orders showed that the gliding species from Rodentia have MPP3 lengths similar to those of arboreal species and have elongated MIP3, while the gliding species from Dermoptera have the shortest MPP3 and the most elongated MIP3 compared with species of the other two locomotory modes and the gliding species from Rodentia (Supplementary Table S1). According to these above observations, we speculate that the similarities and differences between these two orders make them both distant from the other locomotory modes, but prevent their clustering like that seen for the other two locomotory modes, although they do show a trend of having closer proximity in the phylomorphospace plot and do show the significant convergence signal in C-metrics analyses. Besides the phalangeal indices of manual digit III examined here, the gliding mammals also developed other morphological modifications for gliding adaptation, including the relatively elongated neck, longer forearms, lumbar, and shorter thoracic vertebrae, hands, feet bony as well as the more sensitivity of bony labyrinth morphometry in the inner ear (Thorington and Santana, 2007;Pfaff et al., 2015;Kawashima et al., 2018). So we hypothesized that additional morphological traits, in addition to the three indices used in the present study, are also necessary to address the convergence in the gliding locomotory mode. Indeed, a recent phylogenetic comparative analysis of 31 skeletal traits from the gliding mammals has observed convergence, including the gliding Dermoptera and Rodentia (Grossnickle et al., 2020).

CONCLUSION
In the present study, we used morphometric analyses based on quantitative methods to explore the convergence of locomotory modes across diverse Euarchontoglires orders. Our results showed that the phalangeal morphology of species with terrestrial, arboreal, and gliding locomotory modes from multiple Euarchontoglire orders are convergent, supporting the convergent evolution of these three locomotory modes. This study adds evidence of phenotypic convergent evolution of the locomotory modes in Euarchontoglires by extending the investigation into the terrestrial, arboreal, and gliding locomotory modes across four Euarchontoglires orders, which had previously only been studied for the gliding mode of Euarchontoglires or for the other modes only within Primates and Rodentia. With these findings we provide a framework for future work on convergence in mammals, and the application of the phalangeal indices used here to allow the inference and interpretation of locomotory modes of unknown or fossil species.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
The animal study was reviewed and approved by the Committee on Animal Research and Ethics of Yunnan University.

AUTHOR CONTRIBUTIONS
LY designed the study. W-HG, L-FC, TZ, XW, and RL measured and collected the data. X-PW and W-HG performed the data analyses. LY, X-PW, and W-HG wrote the manuscript. CR and DI revised the manuscript. All authors read and approved the final manuscript.