Discussion

We applied 3D geometric morphometrics on humerus and cranial osteology in crocodile newts, a group of largely understudied Asian salamandrid salamanders with polymorphic reproductive strategies, in order to reveal patterns of SD and how it is linked to the different mating behaviours. Our study does add new insights into the evolutionary forces leading to morphological differences between the sexes in urodeles. We confirmed our hypothesis that the different reproductive ecologies of crocodile newts lead (at least partly) to different patterns of SSD and SShD, as well as evolutionary trajectories of SD.
In salamandrids, it has been shown that allometric trajectories differ between taxonomic units allowing higher flexibility for shape evolution (Ivanović et al., 2007, Cvijanović et al., 2014, Ivanović & Arntzen, 2017). Even within populations of a species, differences in allometries can be traced (Ivanović & Kalezić, 2012). Most of the pleurodeline newt species showed similar allometric trajectories but some were able to evolve into a different direction. Although there was a high flexibility in allometries across different genera, similar allometric shape changes were observed throughout the pterygoid, quadratum and the decrease of the occipital region (Ivanović et al., 2012). Apparently the anterior skull shape was therefore more flexible. We found similar allometric trajectories between males and females in all species. In some populations of selected European newts, allometries differ between males and females (Ivanović et al., 2008) whereas in the most basal true salamander this is not the case (Pogoda et al., 2020). In general put on record, if selection acts on size in one sex, this can lead to shape differences due to allometric shape changes during growth (Ivanović & Kalezić, 2012, Pogoda & Kupfer, 2020). Complex interactions of different allometric trajectories between species and sexes make it complicated to pinpoint individual selective mechanisms. The different allometric trajectories between species likely lead at least partly to differences in shape as well as size differences in cranium and humerus occur in the different newt species. Thus, SShD-patterns can vary based on the allometric shape changes and the degree of SSD in a species. Further, some body parts in urodeles are rather less strongly ossified right after metamorphosis e.g., including the cranium, metacarpals and –tarsals but they ossify during ontogenetic growth. This could explain further shape variation in larger crania. Crocodile newts exhibit elaborated bony ridges and ornamentation on the cranial skeleton which may increase due to ongoing ossification during aging and growth, leading to more elaborated squamosal bony ridge at the posterior part. Increasing connection with size of the maxillary with the quadratum and pterygoid can likely be attributed also to the ongoing ossification during life.
Sexual shape dimorphisms in the cranium of other newts and salamanders comprise differences in vomer length, quadrate position, occipital region size, skull width and snout shape (Ivanović et al., 2008, Ivanović & Kalezić, 2012, Alarcón‐Ríos et al., 2017, Pogoda et al., 2020). These shape changes were also present in crocodile newts, but they additionally differ in the extent of the squamosal bone. Different shape changes from males to females between populations were already indicated in other salamandrids like Lissotriton (Ivanović & Kalezić, 2012), Ichthyosaura (Ivanović et al., 2009),Salamandra (Alarcón‐Ríos et al., 2017) and Salamandrina(Romano et al., 2009, Pogoda & Kupfer, 2020). Different ecological selective forces each population of a species endures at its specific site, likely play a major role on the maintenance of subtle shape differences between species (Kalezić et al., 1992, Schäuble, 2004, Angelini et al., 2015). Crocodile newts inhabit a large distribution area from Nepal to Japan, including a variety of habitats from lowland rainforest to temperate, mountainous grasslands (Hernandez, 2016, Wang et al., 2018) forcing specific adaptations to those contrasting environments. On the other hand, this group of salamanders showed high conservatism in their macro-ecological differentiation (Hernandez et al., 2018) likely leading to little shape differentiation between species as well (Pogoda et al., 2020).
Sexual selection and sex roles during reproduction are a main force forming body shapes (see Darwin, 1871, Fairbairn et al., 2007). Previously, we found already that the mating mode partly explains shape variation in crocodile newts (Pogoda et al., 2020). While the mating mode is explaining the different SSD and SShD-trajectories of cranial morphology between species, it does not apply to the humerus, although they differ in shape between mating modes. As both sexes of amplecting species respond differently in cranial shape changes if referenced accordingly to the mean cranial shape but not in the humerus (Fig.9) differences of SD-trajectories of the cranium but not humerus morphology might be explained. Further, as the cranium is much more complex build it provides more possibilities for shape variation than the humerus does. Only amplecting species exhibit SSD in the body parts investigated herein. Often it is tried to explain cranial shape differences with differences in food niches (Shine, 1989, Shetty & Shine, 2002, Ivanović & Kalezić, 2012, Reinhard & Kupfer, 2015) however our knowledge on food niche differences in salamanders is incomplete and a definite association of cranial shape differences with food niches is yet untested. No intersexual differences in food allocation were found inT. podichthys , the only studied crocodile newt in terms of trophic ecology (Phimmachak et al., 2015b). Size differences and slender humeri shape likely facilitate clasping (compare to Pleurodeles ) while more robust and ossified fore limbs provide better standing during circular dancing. This is contradictory to patterns found in European newts also applying a stereotypic courtship behaviour. Among European newts males regularly bear longer fore limbs (e.g., Malmgren & Thollesson, 1999, Çiçek et al., 2011, Reinhard & Kupfer, 2015). We demonstrated that different selective forces are acting differently on the crocodile newt taxa inducing interspecifically different allometric trajectories while the trajectories were constant between the sexes. We are aware that our sample size of some of the studied species and/or sex was relatively small, mainly attributable to the overwhelming male bias of crocodile newts housed in natural history collections sometimes with a male/female ratio of about 30/1 but GM was proved in being well capable of revealing even quite subtle shape differences (e.g., Blanco & Godfrey, 2006, Abdel-Rahman et al., 2009, Arendt, 2010, Ivanović & Kalezić, 2012, Pogoda et al., 2020)
We conclude that morphological adaptations between species take place in correlation with and probably adaptation to their occupied environments which include large variation from the tropics to temperate climate zones while the selective forces stay more or less constant in the entire group as sex roles do not diverge, although some aspects of reproduction and courtship adapted. The interaction of variable allometric trajectories of species and sexes might partly explain the interspecific differences in SD patterns observed. Additionally, ecological parameters might influence strength and direction of SD across populations. The mating mode is to some degree an explanatory variable of the interspecific cranial shape variation in the SD-trajectories. The humerus differed with mating mode but other selective forces shape SD-trajectories of this specific body part. However, to understand evolutionary processes future studies especially on the ecology of the enigmatic crocodile newts are needed. Numerous species were just recently identified based on molecular data leading to an enormous increase in species numbers over the last decade (Stuart et al., 2010, Shen et al., 2012, Nishikawa et al., 2013a, Nishikawa et al., 2013b, Hou et al., 2014, Nishikawa et al., 2014, Yang et al., 2014, Khatiwada et al., 2015, Le et al., 2015, Phimmachak et al., 2015a, Qian et al., 2017, Grismer et al., 2018, Grismer et al., 2019, Zaw et al., 2019, Bernardes et al., 2020, Pomchote et al., 2020, Poyarkov et al., 2021). However morphological distinct characters are hard to identify and often appear somehow descriptive summarising differences of particular body part proportions based on a few specimens without any accounting for SD. Unfortunately no information on ecology from the many recently described species is available and observations in captivity are often based on individuals of uncertain genetic identity. It will become more crucial to focus more on the intraspecific morphological differences applying an integrative approach for future taxonomic research on crocodile newts. Especially the subgenus Yaotritonincludes genetically distinct lineages not accessed accurately so far (e.g., Wang et al., 2018, Bernardes et al., 2020). Future research has to take into account larger taxonomic units with broader ecological niches to resolve evolutionary processes and mechanisms of SD on a larger scale.