2.1 Cranioandibular 3D Geometric Morphometrics

3D geometric morphometric analyses using a combination of fixed and sliding/semi-landmarks (see Section 4; Extended Data Figures S1 and S2) revealed that in shape, the first principal component (PC1) explains almost half the variation in the cranium and a third in the mandible (Figure 1a,b). In both structures, PC1 is strongly correlated with age and size (Figure 1). The remaining principal components (PC) show relatively little variation and seem to be associated with minor differences between individuals of the same age classes.

The pattern of change with age in cranial shape conforms to the typical ontogenetic pattern in mammals, comprising an initial stage of very rapid change, usually associated with lactation, a second stage of reproductive or pubertal development, and an asymptotic phase at sexual maturity (von Bertalanffy growth coefficient = K = 0.20; Figure 1c) (Bogin 1999). The significantly lower growth rate and a delay in the stabilization of the mandibular shape stands out (K = 0.12; Figure 1d). This mandibular development during maturity highlights the presence of another underlying biological phenomenon, which departs from classical ontogenetic patterns in mammals and has a high impact on shape variability. The growth in size of both the cranium (K = 0.21; Figure 1e) and the mandible (K = 0.22; Figure 1f) show a standard pattern, with more than half of the accumulated change in the immature stage, and a last phase of change in values and degree of variability in the pubertal stage before sexual maturity.

Shape and size in both crania and mandibles are highly correlated (OLSR, R² = 0.92, p < 0.0001; R² = 0.81, p < 0.0001, respectively) (Extended Data Figure S3a,b). In the case of the mandible, the allometry is noteworthy, with slower and more gradual variation in shape by age than in the cranium (Figure 1d,e). The allometric pattern present in the mandible is peculiar, fitting significantly better to a quadratic function than to a linear correlation (Linear regression model Akaike IC = 4.1 < Quadratic model Akaike IC = 6.2) (Extended Data Figure S3b). Interestingly, despite these differences in the rate of variation with respect to age, there is a greater correlation of shape against size for the mandible than the cranium (Extended Data Figure S3c,d). Strongest correlation is between the size of the crania and mandibles (OLSR, R² = 0.99, p < 0.0001; Extended Data Figure S3c), although in the case of shape the degree of correlation is also high (OLSR, R² = 0.86, p < 0.0001; Extended Data Figure S3d).

The series of craniomandibular morphological modifications in the ontogeny of this species can be observed in Extended Data Figures S4, S5, S6, S7, S8, and S9, and Extended Data—Supporting Information S20: 3D Models. These phenotypic changes are mainly focused on the relative proportion of the neurocranium to the chewing apparatus (maxillae, premaxillae and mandibles in general) and the overgrowth of the superstructures (apophyses, crests and other craniomandibular projections), going from very globular crania with very high neurocranial proportions in juveniles/newborns to extremely antero-posteriorly elongated crania in adults. Other notable changes with ontogeny are the great projection of the orbits, the occipital condyles, the occipital superstructures, the zygomatic processes, the mandibular angles and, in general, the alveolar areas of the entire anterior dentition (incisors and canines).

2.2 Modeling and Meta-Analysis

The ontogenetic patterns presented by each trait can be easily divided according to their growth coefficient in the von Bertalanffy growth curve. In those models with a coefficient greater than 0.15, there is a tendency for an abrupt change early in life (immature stage), which gives way to an attenuated change during puberty to reach the asymptotic stage practically coinciding with the onset of adulthood (Figure 2). This is the case for cranium shape and size, mandible size, height, or cranial capacity. The same pattern is observed in the degree of fusion of the mandibular symphysis (Extended Data Figure S10d) (Laws 1968). In the majority of these cases, the sharpest change in values occurs mainly in the suckling period, which never exceeds 2 years of age (Martino et al. 2024b).

On the contrary, in the traits with von Bertalanffy coefficients lower than 0.15, more gradual change is observed throughout ontogeny, with practically no rate changes between the life-history stages mentioned above (Figure 2). In some cases, this pattern is related to a late stage of phenotypic change that is mainly observed in males and is related to the display of secondary sexual characters, as in the case of the weight of the mandible and lower canines (Extended Data Figure S11) (Laws 1968; Shannon et al. 2021). In the case of the gonads this pattern is especially different from the general ontogenetic pattern (see original data in Extended Data Figure S10l,m) and it is very likely that the data fit more appropriately to another model of change due to the very late onset of expression of the hormones that determine the changes in these organs (Emerson 2000). More particular is the case of the change in total body length and mass of individuals, which despite not showing a pronounced sexual dimorphism (Shannon et al. 2021), present new stages of change later in the life of the organisms, moving away from the typical generalized mammalian growth pattern (Figure 2). Interestingly, the growth pattern of total body length and mass of individuals seems to be the most similar to the one we observed in the case of mandibular shape (Figure 2).

In most of the data evaluated, we found the typical generalized mammalian pattern of change between age classes, being especially significant between the pre- and post-weaning stages, notably between late infancy and puberty and with some between puberty and adulthood, which should not show significant changes throughout its duration (Figure 3). Specific cases such as combined testes weight, which does not change significantly until puberty, deviate from this pattern. Other peculiar cases are those of mandibular and canine weights, which show change during adulthood (at least statistically significant in the case of the mandible) and a large bimodality in late adulthood (Figure 3; Extended Data Figure S11). This bimodality is easily explained by sexual dimorphism in the development of secondary sexual characters. Interestingly, we also found significant changes within adulthood in the case of mandibular shape, total body length, and body mass, in this case without associated data bimodality (Figure 3; Extended Data Figure S11). For a detailed summary of the factors assessed in the consideration of the ontogenetic pattern, see Extended Data Table S1.

3.1 Cranial Phenotype and Life-History Traits Across the Ontogeny in Hippos

Our results highlight a dramatic change in craniomandibular anatomy along with other body parameters (such as body size or, especially, cranial capacity; Laws 1968; Shannon et al. 2021; Boisserie 2002; Weston and Lister 2009; Martino et al. 2024b) in the early stages of infancy, between birth and 3 years of age. A relationship of this early reconfiguration of cranial structure has been attributed to factors such as weaning, which occurs before the age of 2 years (Martino et al. 2024b) and mainly between 6 and 8 months (Eltringham 1999; Lang 1975). Indeed, as observed in our results, many of the modifications in the craniomandibular anatomy at this stage of development (e.g., enlargement of the muscle insertion zones in the temporals, reconfiguration of the alveolar space in the mandible or the projection of the gonial angle) seem to be related to the structure and functionality of the masticatory apparatus (Hiiemae 1978; Weijs 1980), so this potential relationship seems to be well founded.

A marked increase in cranial capacity occurs during infancy (Weston and Lister 2009), which is accompanied by a rapid change in facial structures such as the relative projection of the orbits or the alveolar areas of the canines and incisors. Early structuring of the central nervous system is a typical and well-known process in mammals (Bogin 1999) and is often linked to the sudden need for independence of the offspring after weaning for survival (Herring 1985). More specific is the splachnocranial projection, linked to positive craniofacial allometry (CREA) following the suggestion that larger species/individuals/ontogenetic stages in mammals tend to become long-faced as they diverge in size (Cardini and Polly 2013; Andrea 2019). In turn, these anatomical changes may be related to the semi-aquatic lifestyle of hippopotamuses (Reidenberg 2007). Although there are multiple hypotheses that assign characteristics more suited to an aquatic lifestyle to infant individuals than to adults (e.g., greater buoyancy or webbed feet) (Coughlin and Fish 2009; van der Geer et al. 2015). Indeed, Verheyen (Verheyen and Verheyen 1954) reported observations of infant individuals actually swimming. Likewise, the elevation of the orbits and the projection of the nasals with their involvement in the elevation of the nostrils, which in turn define the adult cranial shape of the common hippopotamus, are features that have been typically related to a more aquatic behavior (Boisserie et al. 2011; van der Made et al. 2015).

The second moment in the life cycle of hippopotamuses that raises several questions is the correlation between phenotypic changes and the attainment of sexual maturity (Martino et al. 2024b). The exact age of sexual maturation in hippopotamuses is indeed highly variable among individuals, especially if their development has occurred in different contexts (Dittrich 1976; Sayer and Rakha 1974). Even so, it seems clear that in captivity first offspring can be observed in both females and males at an average age of 5 years (Dittrich 1976). However, if we focus on hippopotamus individuals inhabiting nature reserves, everything seems to point to an onset of puberty around 6–7 years of age in the wild with the attainment of maturity and conclusion of puberty in all individuals at around 15 years of age (Sayer and Rakha 1974; Marshall and Sayer 1976; Smuts and Whyte 1981). Very similar information is provided by data on the weight and production of the testes and ovaries, which begin their pronounced development in 5 years old hippopotamus individuals (Smuts and Whyte 1981). In the case of the testes, the onset of spermatogenesis coincides with this age and growth changes speed at the end of puberty, whereas in the ovaries, growth continues with the same trend throughout the life of the individual.

According to our results, a cranial phenotype very similar to that of the adult is structured in the pubertal stage, reaching stabilization of shape and size with the definitive maturity of all individuals. This phenomenon does not extend to the mandible, which seems to follow a similar dynamic of morphological change to that found in infancy. Individual shoulder height seem to respond in a very similar way to the cranium, reaching adult size by the end of the pubertal stage (Shannon et al. 2021; Marshall and Sayer 1976). As in most mammals (Kilborn et al. 2002), it is during puberty and around the final establishment of maturity that the full development of much of the musculoskeletal system is reached, with the fusion of most of the epiphyses and sutures of the post-cranial skeleton, with the exception of some vertebral discs and bony sutures of the girdles (Extended Data Figure S12) (Weston 1997). All this information points to a high correlation between the development of reproductive traits and musculoskeletal system that enables independence and reproduction in individuals between 5 and 15 years of age (taking into account the large amplitude of this interval due to the incorporation of data from zoo and wild individuals). At the social level, these still young males are structured in subgroups within the same group (bachelors) as long as they maintain a non-confrontational profile with the territorial males, delaying the age of the first courtship displays (Eltringham 1999).

With the onset of adulthood, dental eruption and the fusion of the mandibular symphysis reach the adult phenotypical stage (Laws 1968; Boisserie 2002). In many species, sexual dimorphism develops late in ontogeny and mainly in size differences between the sexes (Lindenfors et al. 2007; Pérez-Barbería et al. 2002; Roylance-Casson 2021). This phenomenon is strongly related to the expression of different development rates (e.g., in the rate of body growth or in the development of hypertrophy of an anatomical element) during ontogeny between females and males (Isaac 2005). In hippopotamuses, this difference in the development of most body structures between sexes does not seem to occur, with sexual dimorphism mainly reflected in the bimodality of canine and mandibular weights in adults [for more information see Shannon et al. (2021)]. It is expected that the greatest development of sexual dimorphism in canines and mandibles occur at the same age when young adult males acquire territorial behaviors and intentions to dominate the groups in which they have grown up (Eltringham 1999). If these observations were confirmed, it would be very interesting to test the relationship between the triggering of canine hypertrophy by late hormone expression and the onset of territorial ethological patterns (Emerson 2000).

In the general pattern of ontogenetic change in mammals (Bogin 1999; Shea 1983, 1985a, 1985b; Leigh and Shea 1995), and specifically in hippopotamuses (Weston 2003), much of the process of phenotypic growth modification has an allometric explanation. It is precisely the different allometry patterns between the elements of the craniomandibular system that are most affected during life-history and evolutionary processes, generating the diversity of forms that we currently find in mammals (Lanzetti et al. 2022a, 2023; Goswami and Prochel 2007). In our case, particularly striking are the differences in the variation rate in the shape and size observed between the mandible and the cranium. This result suggests the presence of a mismatch in the growth rate between these two structures. This discrepancy could be related to the pronounced sexual dimorphism of the mandible and, in particular, of the lower canines. Surprisingly, when looking at the data on mandible shape and size, it is not possible to discern any sexual bias or bimodality in the distribution of the data at any ontogenetic stage (see Figure 3 and Extended Data Figure S11). In the case of the mandible, the quadratic model to which its allometry fits could be explained by a first phase in its ontogeny in which the variation in size did not entail any change in its shape. This process could be explained, in part, by a phase of growth that was not associated with any dental emergence event or its associated morphological reconfiguration (Laws 1968; Weston 1997; Avedik and Clauss 2023). Likewise, in the mandibular shape, a final stage of change seems to emerge before the stabilization of the mature adult phenotype. This last stage of change is not observed in the shape and size of the cranium nor in the height of the individuals. Instead, a similar pattern is observed in total body length or body mass. It is easy to relate the continuous change during adulthood of these latter parameters to the pattern of food intake (Lui and Baron 2011; Mumby et al. 2015). If we look in more detail, the main anatomical changes that occur in the mandible during adulthood are centered on the increased development of the muscular attachments of the masseter muscles that produce an extension of the gonial angle. These morphological changes have major implications for the masticatory biomechanics that are progressively reconfigured with the continuous modification of dental occlusion (Avedik and Clauss 2023). It is important to mention intraspecific agonistic behavior as another possible biological feature related to this morphological trait. Although territorial males are the main source of agonistic behavior, females readily respond to it. Both males and females have constantly growing canines and incisors that they use in fights, and this ethological phenomenon may be an important driver of mandibular morphology and its change over time (Herring 1985; Avedik and Clauss 2023).

3.2 General Growth Patterns

In general, four distinct patterns of change during the lifespan of individuals could be characterized in the hippopotamus body structures assessed. The first is that of the internal organs and skeletal system, with a rapid initial phase of change followed by attenuated change during puberty and a final asymptotic phase in adulthood. This pattern is common for such body systems in all mammals (Tanner et al. 2010; Bogin 1999; Shea 1986; Galatius et al. 2011; Mattson et al. 2006; Fortune et al. 2020; Shea 1983, 1985a, 1985b) and, as expected, fits most hippo traits evaluated in the meta-analysis. The development of the nervous system seems to fit this model best, mainly due to the aforementioned early stabilization of the encephalic structure (Weston and Lister 2009). Directly connected to the latter is cranial development (Lyras 2019).

The second pattern occurs in the elements associated with secondary sexual characteristics and presents a pattern with changes in infant age and puberty that are slower than in the previous case. An extra stage of morphological change is also visible within adulthood, the latter characterized by bimodality in the data reflecting sexual dimorphism. This phenomenon is typical of mammals that show high rates of sexual dimorphism in the later stages of sexual maturity (Shea 1986; Emerson 2000; Lindenfors et al. 2007). The origin of the latter is rooted in the differential selection patterns experienced by female and male individuals (Badyaev 2002; Isaac 2005). These patterns are theoretically more pronounced in organisms with polygynous mating systems that occupy open habitats (Pérez-Barbería et al. 2002; Jarman 1974) which provide environmental conditions more suitable for this new phase of energetic investment (German and Hochberg 2020). As we observed, in the case of H. amphibius this pattern is characteristic of the lower canines but not of the body size, highlighting its ecological peculiarities compared to other ungulates (Shannon et al. 2021).

The third case is present in body sizes and characteristics that are neither skeletal nor closely associated with internal organ maturity, showing ontogenetic changes similar to the second case, but without expressing bimodalities in the data at any stage. This phenomenon is due to a gradual variation in the characteristics of all individuals throughout their lives (Bogin 1999; von Bertalanffy 1957; Jeglinski et al. 2012), with these changes being mainly associated with an increased intake or accumulation of fat in tissues (Lui and Baron 2011). This phenomenon is documented in the post-maturity body mass gain in large mammals with competitive social interaction such as Asian elephants (Mumby et al. 2015) and it is not surprising that hippopotamuses, with their semigraviportal nature (Hutchinson and Pringle 2024), exhibit similar characteristics. More curious is the presence of a similar pattern in mandibular morphology, which in this case we assimilate to changes in the biomechanics of mastication.

The last of the patterns is presented by the gonads, which do not begin to develop until late in puberty and continue to undergo substantial changes throughout the individual's life. Unsurprisingly, this dynamic is associated with a large increase in physiological pathways related to sex hormones production (Bogin 1999; Emerson 2000).

In conclusion, the analysis of a large cranial dataset has provided extensive information on the ontogenetic pattern of hippopotamuses, the only large semigraviportal social mammal with semi-aquatic habits. Specifically, we gain insight into the main stages that characterize the ontogenetic change in Hippopotamus amphibius. As in the vast majority of social mammals, the first years of life are the key for the establishment of the adult craniomandibular structure. These changes are strongly conditioned by the development of the central nervous system and changes in feeding and ethological habits. The sexual maturation phase in turn leads to the stabilization of most phenotypic characteristics of adult individuals, including cranial morphology, full skeletal development, and maturity of the reproductive organs. One of the most striking results obtained is the discovery of an extended/additional late phase of modification of the mandibular phenotype that, interestingly, is not related to the development of sexual dimorphic characteristics in males (i.e., hypertrophy of the lower canines). It is possibly linked to occlusal reconfiguration in adulthood due to dental wear and to an increased food intake, leading to a marked development of the insertions of the temporalis and masseter muscles. This pattern can also be seen in the change in mass and total length of individuals during adulthood. These results are particularly important since hippopotamuses are a peculiar component of the present, as well as the past, megafauna. Future comparative work including these data together with studies of phylogenetically close groups (such as the pygmy hippopotamus, Choeropsis liberiensis) or their extinct representatives (such as hippopotamuses or anthracotheriids, which have a good fossil record) will allow to rigorously address issues such as the ontogenetic origin of adaptations to aquatic life in mammals or the impact of changes in global climate on niche modifications in megafauna through their development.

4.1 Materials

To representatively assess the change in the craniomandibular phenotype of Hippopotamus amphibius during its life cycle, 198 individuals (183 crania and 172 jaws) have been digitized in 3D. This sample includes individuals of all ontogenetic stages, mainly wild individuals, although some come from zoos. It also includes individuals whose sex in life is known (24 known females and 24 known males) and others for which it has not been possible to obtain this information. The specimens have been digitized by various specialists, using photogrammetric methodologies, structured light scanners, laser scanners, or computed tomography. The materials come from 20 institutions of importance in the study of the natural sciences: D'Arcy Thompson Zoology Museum (Dundee, Scotland, UK), Darwin Museum (Moscow, Russia), Institut Català de Paleoecologia Humana i Evolució Social (Tarragona, Spain), Lapworth Museum of Geology (Birmingham, England), LSA Museum of Zoology University of Michigan (Michigan, USA), Museo de Anatomía Comparada de Vertebrados (Madrid, Spain), Museum of Natural History Wroclaw (Wroclaw, Poland), Museum of Zoology (Cambridge, England), Museum of Vertebrate Zoology (Berkeley, UK), Museu de Ciències Naturals de Barcelona (Barcelona, Spain), Museo di Storia Naturale di Firenze (Florence, Italy), The Natural History Museum (London, England), Muséum National d'Histoire Naturelle (Paris, France), National Museum of Nature and Scien (Tokyo, Japan), Universidade Nova (Lisbon, Portugal), Royal Museum for Central Africa (Tervuren, Belgium), Staatliches Museum für Naturkunde Stuttgart (Stuttgart, Germany), Museo Anatómico Universidad de Valladolid (Valladolid, Spain), University of Wyoming Museum of Vertebrates (Wyoming, USA), Museum für Naturkunde (Berlin, Germany). The data associated with each specimen included in the study can be found in Extended Data Tables S2 and S6.

Some of the specimens included in the analyses showed a certain degree of fragmentation, which has been corrected by merging the symmetrical anatomical part, if preserved (Lautenschlager 2016; Moya-Costa et al. 2019). Blender v.3.1.2 software was used for the symmetrisation of the specimens. The alignment and fusion of the fragments were performed in CR Studio v.1.7.5.086 software. The age estimation of each individual was carried out following the criteria of emergence and dental wear proposed by Laws (Laws 1968), as long as the mandible was present, and those proposed by Martínez-Martinez-Navarro et al. (2022) for isolated crania. The analyses included individuals from the 20 ontogenetic stages defined by Laws (Laws 1968).

4.2 3D Geometric Morphometrics

4.3 Life-History Data

The proxies of the life-history traits have been obtained from different bibliographic sources and databases, which will be presented below. All the data extracted from the different sources of information and used in the analyses discussed here can be consulted in Extended Data Table S5.

General body parameters are represented by body length, height, and body mass, measured between 1961 and 1966 in Queen Elizabeth National Park (Uganda) by Prof. Richard Laws' team (Laws 1968) and published by Shannon et al. (2021).

Skeletal maturity data, such as fusion of mandibular symphysis and loss of the first deciduous premolar (dp1,) are taken from data from Queen Elizabeth National Park (Uganda) reported in Laws (Laws 1968). Estimates of bone fusion moments are taken from observations made by Weston (Weston 1997) on individuals from the National Museums of Kenya (Kenya).

Cranial capacity data are from the measurement of specimens in different European museums by Weston & Lister (Weston and Lister 2009).

The jaw weight and canine weight data, used and previously published by Shannon et al. (2021) to discuss sexual dimorphism, are part of the Queen Elizabeth National Park (Uganda) database generated by Prof. Richard Laws' team (Laws 1968).

Data relating to sexual maturity and changes in reproductive behavior are represented by: the percentage of pregnant and lactating females in the Luangwa valley (Zambia) between 1970 and 1971 (Marshall and Sayer 1976); the relationship between the percentage of pregnant and lactating females and the size of testes and ovaries is taken from specimens analysed between 1974 and 1975 in the Kruger National Park (South Africa) by Smuts and Whyte (1981); for the minimum age of generation of the first offspring, the data of Dittrich (1976) from the databases of various zoos have been considered; the absolute proportions of pubertal and mature individuals in a population according to age are from the Luangwa River (Zambia) and are taken from Sayer and Rakha (1974).

The graphical representation of each specific parameter has been done by means of scatter plots versus age of the individuals in Past 4.03 software (Extended Data Figure S10a–m,o) (Hammer et al. 2001), except in the case of the first offspring which have been represented as boxplots versus age (Extended Data Figure S10o). A summary chart has been generated to observe and discuss the ontogenetic coincidence of the different life-history events and processes with the developmental stages of craniomandibular morphology (Extended Data Figure S12).

4.4 Key Age Classes—Life-History Events

To assess the ontogenetic pattern of each trait under study, it was first necessary to determine key life-history events. The set of parameters related to hippopotamus reproduction assessed in this study characterizes quite clearly the three most important stages of non-human mammal development. First, in individuals of both sexes kept in captivity, the minimum and average age of reproduction are around 3 and 5 years, respectively, which indicates the ages for the onset of puberty (Extended Data Figure S10n) (Dittrich 1976). The onset of lactating or pregnancy in a small percentage of wild 5-year-old female individuals indicates the beginning of the pubertal phase, while the generalization of lactation and pregnancy in 14–15-year-old females marks the onset of adulthood (Extended Data Figure S10j,k) (Marshall and Sayer 1976; Smuts and Whyte 1981). It is not possible to observe these stages in the growth of the ovaries, which does not seem to follow a sigmoidal pattern (Figure 2; Extended Data Figure S10m) (Smuts and Whyte 1981). Males, on the other hand, do not appear to functionally generate sperm until 5 years of age (coinciding with the end of the infantile stage) and do not reach the most stable testicular size until 15 years of age (coinciding with the transition to adulthood; Figure 2; Extended Data Figure S10l) (Smuts and Whyte 1981). These stages can also be clearly seen in the pattern of fusion and maturation of the bones of the postcranial skeleton. A main skeletal maturation seems clear up to the age of 10 years and a final stage of skeletal growth closure with complete fusion of the epiphyses between the ages of 15 and 25 years (Extended Data Figure S12) (Weston 1997). In summary, the definition of the transition from infantile to pubertal age at around 5 years and the onset of adulthood at approximately 15 years seems to hold well in hippopotamuses.

4.5 Ontogenetic Modeling

Corresponding in this case L_∞ with a, K with c, and t₀ with (lnb)/c. In this paper, we focus especially on the values obtained for the growth coefficient (K in the original equation and c in the form we use here), which are comparable between the models obtained with each dataset and inform us of the dynamics of ontogenetic change presented by each character.

The datasets evaluated correspond to the cranium and mandible shape (as regression scores procrustes coordinates/age), cranium and mandible size (as CS), total body length, height, body mass, cranial capacity, jaw weight, canine weight, combined testes weight, and combined ovaries weight (Figure 2).

To assess the presence of the theoretically expected ontogenetic stages of significant change in each body parameter evaluated, statistical tests were performed by grouping individuals into different age classes. Firstly, the pre-weaning stage (0–2 years), then the post-weaning infant stage (2–5 years), followed by the pubertal stage (5–15 years), and finally the adult stage. The latter has been divided into two age classes (15–25 years and 25–43 years) to allow us to evaluate the presence of a late stage of change that departs from the classical growth pattern in mammals (Figure 3). The frequencies of each value for each age class have also been arranged, making it possible to visualize the presence of bimodal distributions of the data (Figure 3). The generality of the data did not correspond to normal and homoscedastic samples, and small samples were included in the assessment, so non-parametric statistical tests were used. For analyses of the data, the Kruskal–Wallis test was applied, together with Mann–Whitney pairwise post hoc test. The results of the tests are specified in Figure 3 and in the Extended Data Table S6.