Increase on environmental seasonality through the European Early Pleistocene inferred from dental enamel hypoplasia

An in-depth study of the Early Pleistocene European remains of Hippopotamus has allowed the first detailed description of the incidence and types of dental alterations related to palaeopathologies and potentially linked to climatic and environmental factors. The results of a long-term qualitative and quantitative assessment highlight the importance of nutrient deficiencies on the development of dental enamel hypoplasia in Hippopotamus. Glacial cyclicity and the resulting changes in humidity and plant community structure conditioned the local environments critical for the survival of this taxon. Two main intervals of putative constrained nutritionally restrictions were detected at ca. 1.8 Ma and ca. 0.86 Ma (i.e., MIS63 and MIS21, respectively). Statistical comparisons show an increase in the frequency of dental hypoplasia between these two chronological periods, thus reinforcing the idea of increased seasonality in the circum-Mediterranean environments during the Early Pleistocene.


Background on Dental Enamel hypoplasia and Hippopotamus paleoecology
Dental Enamel Hypoplasia (DEH) is a disruption in enamel secretion by ameloblasts during amelogenesis, which is linked to physiological stresses 8 .There are three broad categories of DEH according to FDI (Federation Dentaire International 9 ) and, according to Hillson 10 and Hillson and Bond 11 , DEH displays a wide array of expressions, comprising of vertical or horizontal grooves, pits, or broad bands of missing or incomplete enamel.In large mammals, such disturbances are often associated with periods of environmental and/or nutrient stresses, resulting in a metabolic stress that causes this pathological state 8,12 .Many studies of the relationships between DEH and the environmental conditions in which the individuals lived have been published [13][14][15] .The position, width, and depth of DEH can provide information on the age of occurrence of anomaly, the period of stress event, and the severity of the stressors 16,17 .
Among the large mammals that inhabited Europe during the Pleistocene, the large-sized Hippopotamus spp.had several environmental requirements (such as the presence of permanent bodies of water or the abundance of grassland or aquatic macrophytes) making them relevant for climatic studies 18 .Their climatic range has been considered narrow, conditioning the use of their presence as a clear indicator of warm and humid climates 19 .However, preferential diet of individuals of Early Pleistocene European populations of (ca.3200 kg) Hippopotamus antiquus is still debated.Specifically, the discussion continues between a grazing behaviour similar to the dietary preferences of extant Hippopotamus amphibius 20 and a diet centred on aquatic macrophytes, as suggested by biogeochemical data from the late Early Pleistocene site of Venta Micena 21 .
Despite the specific ecological requirements related to water availability and the characteristic continuous growth of their anterior dentition, DEH in hippopotamuses has been scarcely reported 20,22 and no comprehensive study has been published so far, especially in relation to changes in the biotic or abiotic factors of the environment.In any case, no comprehensive study of DEH in hippopotamuses or any long-term projection of this phenomenology against changes in the biotic or abiotic environment has been carried out so far.
Following the observation of a large representation of DEH in the anterior dentition of the European hippopotamuses (H.antiquus) found in several Early Pleistocene layers at different sites in Europe (Fig. 1), this paper evaluates the question of whether this pathological state may be associated with changes in climatic condition that may resulted in nutritional stress.

Chronological and geographical distribution of Enamel Hypoplasia
The analysis of the sample of incisors and canines showed 105 pathologies related to the production of dental enamel among the 310 specimens analysed (30 among 74 Minimum Number of Individuals; Table S1).Twelve of the 23 Early Pleistocene sites studied around Southern and Central Europe document specimens with enamel hypoplasia, although with different incidence (Figs. 1, 2 and S1; Table S1).Of these sites, only the samples of Upper Valdarno and Vallparadís Section have enough specimens to be significative (see "Material and methods").
In the case of the Upper Valdarno, 52.8% of the observed specimens had enamel pathologies (Fig. 4a1).These pathologies were mainly found in the lower canines (Fig. 4a2; Table S1).On the other hand, 39.5% of the specimens showed enamel pathologies in the post-Jaramillo layers of Vallparadís Section (Fig. 4b1; Table S1).In this sample, pathologies in the upper canines stand out compared to those observed in the lower ones (Fig. 4b2).
The evaluation of the frequency of episodes of LEH in the upper canines (Table S2) resulted in a sample of 7 specimens with 34 measurable oscillations in the Upper Valdarno sample, and 15 specimens with 102 oscillations in the sample from the post-Jaramillo layers of the Vallparadís Section.Oscillations with a mean amplitude of 16.7 mm (Fig. 4a7-9) were found in the Upper Valdarno.In contrast, a mean of 14.8 mm (Fig. 4b7-9) was found in the post-Jaramillo Vallparadís Section.Comparison of the means using the non-parametric Kruskal-Wallis test provided a p-value of 0.04463, showing the presence of significant differences between the two samples, which are indicated as subscripts under the error bars in the boxplots shown in Fig. 3a12, b15.Also striking are the extreme cases, with a minimum of 8.4-9.6 mm and a maximum of 26.2-28.5 mm.

Enamel hypoplasia in Early Pleistocene Hippopotamus
Few cases of DEH associated with trauma have been reported, as most pathological specimens show a clear periodicity in the DEH episodes, which allows discarding one-off events in them (see Fig. 3h and Kierdorf and Kahlke 20 ).Congenital DEH, which is the pathology least reported in previous studies, shows characteristic typologies that differentiate it from most of the cases observed in the sample studied here 8 .The wide geographical and chronological extension of the sample analysed here depicting these pathologies allows to discard an intoxication aetiology as the cause of the observed DEH 22,23 .Given the characteristic repetitive periodicity of hypoplasia episodes assessed in the pathological anterior dentition, the most likely cause of most pathologies observed in the sample studied here appears to be the repetitive and cyclical lack of nutrients.This interpretation is in line with that inferred in other studies, including those involving DEH in extant hippopotamuses 20,22 and other ungulates 14,15,24 .

Hypoplasia frequency in Early Pleistocene Hippopotamus
Lower canine growth was studied especially in the extant common hippopotamus, with direct observations of a ratio of 13-14 mm/year in an old (48 years old) female and 28-30 mm/year in a young (8 years old) one 25 .
On the other hand, estimates of growth ratios by Uno et al. 26 provide values of 33.5-74.7 mm/year for the lower canines and 19.4 ± 3.1 mm/year for the upper ones.Harris et al. 27 considered 30 mm/year as a standard measure of growth for the lower canines, while Souron et al. 28 highlighted that these values can reach 39.1 mm/year for the lower canines and 30 mm/year for the upper ones.In our case, the estimated frequencies of LEH events in the mean measurements obtained for the upper canines were 16.7 mm and 14.8 mm (Fig. 4a9, b9).The sample analysed comes from individuals of the extinct species H. antiquus, which according to estimates made by Palmqvist et al. 21could reach a size twice as large as that observed by Nowak and Walker 29 for the living common hippo (ca.3200 kg vs. ca.1500 kg).With the available data, it is difficult to ensure a specific seasonality to the observed episodes of hypoplasia, but everything points to a periodicity of approximately 1-2 episodes per year.Thus, it would be reasonable to consider that these episodes mark the season(s) with most severe weather, which would limit the access of hippopotamuses to their food resources, or with recurrent aridity periods that prevented the access to lacustrine systems with a water table of enough depth.

Paleoenvironmental inferences
If we focus on the paleoenvironmental context of the samples, the Upper Valdarno specimens (ca.1.9-1.7 Ma) record apparently drastic episodes for the physiology of individuals of European hippopotamus populations.In fact, in the coeval localities of this basin, seasonal drought events have been hypothesised, especially in the site of Poggio Rosso 3 .These aridity episodes fit with the proposed alternations of cold/warm phases of the European climate, which would in turn results in changes in the distribution of open/closed environments 2,30 .Ultimately, disrupting the access to the vegetation on which the hippopotamuses fed.Between 1.7 and 1.4 Ma the climate seems to have been associated with more favourable conditions, with the expansion of grasslands 31 .This environment is close to ideal conditions for the maintenance of hippopotamus populations 32 and might reflect in the absence of DEH in the samples of Venta Micena and Monte Argentario.Nevertheless, we remain cautious in this interpretation, as only two specimens from each site were evaluated.In fact, linear dental hypoplasia was found in specimens of horse Equus altidens from Venta Micena, which suggests unfavourable conditions at least for this grazing and fully terrestrial species 33 .Furthermore, previous biogeochemical studies based on the relative abundance of carbon-and nitrogen-stable isotopes retrieved from bone collagen in the late Early Pleistocene site of Venta Micena (Baza Basin, SE Spain) have shown unexpectedly high δ 15 N values in the remains of H. antiquus compared to other ungulates recorded at the site.The values were even higher than those measured in the hypercarnivores of the palaeocommunity (e.g., sabertooth cats, hyenas, and wild dogs).These high values of H. antiquus in Venta Micena were interpreted as reflecting a diet based mainly on aquatic plants, including the macrophytes and algae that grew in the oligosaline waters of the lacustrine environments of the Baza Basin 21,34 .The feeding behaviour of extant common hippopotamuses is preferentially, and almost exclusively, based on grazing on terrestrial vegetation, and browsing habits have only been reported when grasses are not available 35 .If the interpretation of the Venta Micena isotopic data was confirmed, the  feeding habits of the hippopotamuses found at this site would be particularly different from those of the extant populations and might be linked to environmental conditions that limited the proliferation of grazing areas.
In the period 1.4-1.1 Ma, a progressive increase in the amplitude of climate oscillations is recorded being associated with the beginning of the EMPT 2 , which resulted in an appreciable diffusion of Mediterranean vegetation with the intercalation of arid phases and greater seasonality 30 .The sample from this chronological range includes only five pathological specimens (10.2% of the total number of analysed specimens from this chronological range; e.g., Barranco León D; Fig. 1; Table S1).None of the 23 specimens from the Lower Unit of Cal Guardiola (MIS35) show any type of DEH.The palaeoecological reconstructions of Cal Guardiola in these chronologies point to a mixed environment, with the presence of wooded and open areas, leading to the inference of a climate with some permanent humidity 36 (Fig. 1).
The absence of DEH in the samples dated to ca, 1.0 Ma (MIS31) in the Lower Unit of Vallparadís Estació is noteworthy, although the sample from these localities is very limited (3 specimens).In other samples of the same chronology, such as Untermassfeld or Collecurti, several specimens with pathologies were reported but the prevalence and severity of DEH in them is not excessively high 20 .Palaeoenvironmental reconstructions of these localities, specifically Vallparadís Estació layer EVT12, point to the near absence of DEH in the hippopotamus remains recorded, which coincides with a period of aridity and expansion of grasslands at least in the Mediterranean Europe 36,37 .
During the chronological span comprised between MIS24 to MIS22 (ca.900 ka), a long and severe glacial phase is recorded 2,38 .The samples dated to ca. 0.90-0.86Ma include a prevalence of sites with a high incidence of DEH, such as the post-Jaramillo units of Vallparadís Section.In the later site, an increase in humidity and tree cover has been detected, which could limit the expansion of grasslands 37 .It should also be noted that these are the first records of H. antiquus after MIS24-22 marking a change in biodiversity of terrestrial communities, so it is possible that plant biodiversity also continued to be affected.
Finally, it is important to make special mention of the significant differences reported in the frequency of the episodes of upper canine hypoplasia between the samples from the Upper Valdarno (ca.1.9-1.7 Ma) and those This phenomenon highlights the presence of a marked climatic temporality and an increase in its frequency that may be in line with postulated changes in seasonality during the late Early Pleistocene around the Mediterranean Basin 39 .

Materials
Specimens of the anterior dentition of Early Pleistocene hippopotamuses (H.antiquus) from different European sites have been analysed (Table S1; Figs. 1 and S1).These dental elements have been selected for analysis for the continuous growth of the anterior dentition throughout the life of the individuals, which allows the assessment of changes in the type of enamel secretion by ameloblasts at different stages of their life cycle, with a time resolution of few days 25 .Moreover, preliminary observations of cases of dental hypoplasia in the postcanine dentition have not reported any remarkable cases of DEH.
We have selected those samples of European localities dated to the late Early Pleistocene (i.e., Late Villafranchian and Epivillafranchian), with an established presence of hippopotamuses and accessibility of data to the authors (either primary access or through the literature).For a detailed list of the localities studied and all characteristics related to each sample, see Figs. 1, S1 and Table S2.
310 tooth specimens of the anterior dentition were evaluated.The Upper Valdarno sites 40 and the post-Jaramillo levels of the Vallparadís Section 41 were treated as two unique samples in this study, since in both cases the accumulations that make up the samples are very close geographically and minimal differences in the chronology of deposition are accepted.

Methods
The criteria for the detection of enamel pathologies have been standardised following the typification set out by Goodman and Rose 8 , paying special attention to LEH treated in detailed works such as Dobney and Ervynck 14 .In the specific case of LEH, Fig. 2 shows the different degrees of severity observed in each type of tooth element included in the analysis sample.
Data collection was carried out by direct observation of most samples studied and of those specimens figured in the literature (see Table S1).All cases of dental enamel pathologies found in the anterior dentition were documented in detail, noting the anatomical determination of the element and the typology of the pathologies, taking photographs and scanning the specimens at different 3D resolutions with the most accurate scanner available at each institution (Creality CR-SCAN 01, Precision: 0.1 mm, Resolution: 0.5 mm; Artec Eva, Precision: 0.1 mm, Resolution: 0.2 mm; Artec Space Spider, Precision: 0.05 mm, Resolution: 0.1 mm; Artec Micro, Precision: 0.01 mm, Resolution: 0.029 mm).All data were included in databases for individual observation, detailed analyses, and replicability of the evaluation (Tables S1, S2 and S3).
Sites where the sample was small, or in which the representation of pathologies was null or minimal, were evaluated only qualitatively.The prevalence of hypoplasia in samples from the Upper Valdarno sites and the post-Jaramillo layer of the Vallparadís Section was compared using descriptive statistics and qualitative characteristics (Figs. 3 and 4).
The upper canines were selected for the quantitative comparison of the frequency of LEH episodes in those cases where a sequential pattern was detected, because their growth rates are approximately known in the living H. amphibius [25][26][27] .Frequency measurements were standardized as the measurement of the distance from the middle of one normal enamel attachment phase to the next, measuring from the most proximal part of the canine to the most distal, in 3D models using Meshlab software (Fig. S2, Table S2).For the comparison of the frequency of the LEH episodes between the samples from the Upper Valdarno and the post-Jaramillo Vallparadís Section, box plots of the individual measurements of each episode were made (Fig. 4).A statistical comparison of the differences between the frequencies of the two samples was carried out.Firstly, the normality of the samples was checked using the Shaphiro-Wilk test, observing that both the whole data (W = 0.9455, p (normal) = 3.512*10-5) and the samples of the Upper Valdarno (W = 0.9328, p (normal) = 0.03769) and the post-Jaramillo Vallparadís Section (W = 0.9492, p (normal) = 0.00064) showed departures from normality.Given the absence of normality, the Kruskal-Wallis non-parametric test for comparison of means was performed.
Additionally, direct observations of the sample of Hippopotamus amphibius specimens available at the Natural History Museum (London), the Muséum National d'Histoire Naturelle (Paris), the Museo Nacional de Ciencias Naturales (Madrid) and the Museo Anatómico de la Universidad de Valladolid (Valladolid) were made to check the possibility of including a current comparative representation of the incidence and degree of Enamel Hypoplasia.Very few cases of enamel hypoplasia (Fig. S3) were observed in this sample, mostly related to malocclusion (Fig. S3g, i, j), congenital pathologies (Fig. S3m, n) or infections of the entire dental series (Fig. S3p, q, r).Two of the individuals with these pathologies came from zoos (Fig. S3i-f, h), while the other seven were wild or of unknown origin (Fig. S3g, i-r).In none of the cases was the degree of linear enamel hypoplasia as severe as in the Hippopotamus antiquus sample, so it was not possible to compare the two samples or to assess quantitatively differences in the frequency of hypoplasia episodes.

Figure 1 .
Figure 1.Geographical and chronological location of the samples included in this study.Chronological representation also includes epochs/ages, the marine isotopic curve and the most important climatic events in Europe during the Early Pleistocene.Upper Valdarno comprises specimens from: Upper Valdarno general area (more details unknown), Castelfranco, Cava Bacchi, Figline, Renacci, Vacchereccia, Malpasso and Il Colombaiolo.Above the known chronological ranges of each locality (lines or dots), sites with documented hypoplasia are marked in orange; in blue, sites where it was not detected.Marine Isotopic Stages information from Lisiecki and Raymo 1 .Climatic Events information from Kahlke et al. 42 .

Figure 2 .
Figure 2. Skeletal elements included in the analysis and the different degrees of severity of Linear Enamel Hypoplasia (LEH) observed in each element.Skull specimen: AC3864.1: upper incisors, 2: upper canines, 3: lower incisors, 4: lower canines.From minimum to maximum severity of LEH: periodic appearance of lines with change of enamel colouring; onset of enamel loss from the enamel-dentine contact limits; appearance of periodic bands with absence of enamel with more than half of the development of the enamel layer.Scale bar equals 20 mm.