The Middle Pleistocene assemblages

The Middle Pleistocene assemblages consist of two Proboscidean species: Stegodon trigonocephalus and Elephas hysudrindicus, distributed in the Kedung Brubus + Bapang, Trinil, Ngandong faunas, and from the Middle Pleistocene formations in West Java (Figure 4.13C). Comparative data of other non-Elephantoidea taxa from Sangiran and Trinil from Janssen et al., 2016 (Figure 4.15) were included to detect possible niche separation between co-occurring taxa from these localities.

The δ¹³C values of all Stegodon trigonocephalus and Elephas hysudrindicus samples from the Kedung Brubus + Bapang assemblage (between -2.6 and 1.4‰) are concentrated within the range of herbivore grazing C4 plants in a savannah vegetation, except for one E. hysudrindcus individual that falls in the range for a mixed diet. The δ¹³C values of S. trigonocephalus from younger assemblages such as the Trinil fauna (range between -2.7 and 1.4‰), West Java composite assemblage (range between -3.0 and 1.3‰) and Ngandong (between -5.9 and 2.3‰), are consistent with a range for mixed C3-C4 and predominant C4 diet in a relatively open vegetation. The grazer individuals are dominant in the Trinil and West Java faunas, while the proportion of mixed feeder individuals increases in the Ngandong fauna. Meanwhile, samples of E.

hysudrindicus are rarely encountered in the West Java assemblage (originating from the Gintung and Tambakan Formations) and Ngandong Fauna, and are even totally absent from the Trinil fauna. The δ¹³C values of two individuals from West Java (0.1‰ and 1.0‰) suggest a predominant C4 diet in an open grassland habitat, while the only analysed individual from Ngandong (δ¹³C = -2.9‰) was a mixed C₃-C₄ feeder. The δ¹³C values of S. trigonocephalus and E. hysudrindicus from different Middle Pleistocene faunal assemblages are statistically indistinguishable (p value = 0.1448,

Figure 4.15 Mean δ¹³C and δ¹⁸O values and ranges of: A. all analysed samples from the Sangiran locality, including the Sangiran and Kedung Brubus + Bapang faunas and B. all analysed samples from the Trinil H.K. locality. Faunal assemblages are indicated by different colours, while taxa are indicated by different shapes. Comparative data of cervids, bovids, suids, hominins and felids are taken from Janssen et al. (2016). Indicate in caption why orange dashed line is different for the carnivores, primate and suids.

Kruskal-Wallis test). Nevertheless, there is a trend that the δ¹³C values of the Middle Pleistocene faunas gradually become more negative in younger assemblages.

While all analysed S. trigonocephalus and E. hysudrindicus individuals from the Middle Pleistocene faunal assemblages represent open C₄-dominated vegetation to mixed C₃-C₄ grassland, the δ¹³C value of the Elephas hysudrindicus individual from Sunggu (- 12.7‰) fits in the expected range for predominant C3 diet. The δ¹³C value of the Sunggu individual, presumably represents the youngest occurrence of this species, is significantly more negative than all other analysed individuals from the Kedung Brubus + Bapang, Trinil, Ngandong and West Java assemblages.

Correspondingly, the δ¹⁸O values of all analysed samples are high to moderate (-7.6‰

to 2.3‰, with an exception of a S. trigonocephalus individual from the Bapang Formation (9.7‰), which is strikingly lower in comparison to all other individuals. In those assemblages where the two species co-occurred, the δ¹⁸O values of E.

hysudrindicus are mostly higher than for S. trigonocephalus. However, statistical analysis of the two co-occurring species can only be applied to individuals from the Kedung Brubus + Bapang Fauna (Kruskal-Wallis test p = 0.0008, Mann-Whitney test Benferroni corrected p = 9.44 E-4), since the sample sizes of E. hysudrindicus in other assemblages are insufficient for statistical analysis. Within different assemblages, the δ¹⁸O values of S. trigonocephalus from the Ngandong fauna are considered lower than for other groups, but only significantly lower compared to the individuals from the Kedung Brubus + Bapang Fauna (Kruskal-Wallis test p = 0.0051, Mann-Whitney test Benferroni corrected p = 0.0142).

4.4.3.2.1 Sangiran

Apart from Proboscidean species, the exact stratigraphic provenances of other taxa analysed by Janssen et al. (2016) are uncertain, as the specimens may either have been derive from the Sangiran Formation, the Bapang Formations, or of mixed origin. Taxa analysed by Janssen et al. (2016) include: cervids (n = 11), bovids (n = 7), suids (n = 10), and hominins (n = 6). A plot of individual values and the ranges of the δ¹³C and δ¹⁸O values of all samples from Sangiran area is presented in Figure 4.15A.

The δ¹³C values of all herbivores from Sangiran (between -2.7 and 1.7‰ for elephantoids, between -6.9 and 3.1‰ for bovids, and between -9.9 and 1.8‰ for cervids) are indicative for predominantly C4 grazers, while only four individuals are mixed feeders. Additionally, an outlier cervid individual has a δ¹³C value (-9.9‰) that fits a C3 browser in a more open vegetation preference. The δ¹³C values of omnivores are more broadly distributed in all dietary types (range between -12.5 and -5.7‰ for hominins, and between -14.3 and -7.2‰ for suids). Although for hominins, the majority of samples except for one fit in the expected range for a C₄-dominated diet. Statistically, the δ¹³C values of Elephantoidea in general are significantly higher than compared to both omnivore taxa (Kruskal-Wallis test p = 5.65 E-5, see Table 4.6B for Benferroni corrected p values of taxa pairings), while other pairings are statistically indistinguishable. The δ¹⁸O values of E. hysudrindicus, bovids and cervids are generally higher than for S. trigonocephalus, suids and hominins. However, only the δ¹⁸O values of E. hysudrindicus show statistical significant difference compared to S.

trigonocephalus, suids and hominins (Kruskal-Wallis test p = 0.0001, see Table 4.6B for Benferroni corrected p values of taxa pairings).

4.4.3.2.2 Trinil

The analysed specimens from Trinil all originate from the Haupt Knochenschicht layer.

Taxa analysed by Janssen et al. (2016) include: bovids (n = 25), cervids (n = 4), suids (n

= 3), hominin (n = 1) and felid (n = 1). Individual data and range plots for Trinil are presented in Fig. 4.15B. Analysed herbivore individuals from the Trinil H.K. (including bovids, cervids and S. trigonocephalus) yielded individual δ¹³C values that range between -2.7 and 3.9‰, indicating that they consumed predominantly C4 vegetation.

Most of these individuals plot in the expected range of predominant grazers, and only four individuals of S. trigonocephalus and bovids fit in the range of mixed feeders.

Nevertheless, δ¹³C values of these mixed feeder individuals are closer to the pure C₄ boundary, suggesting that their diet relied more on C4 grasses. A single individual of B.

palaeokerabau represents an outlier (δ¹³C value of -8.4‰) indicates a C₃ feeding preference in open woodland vegetation. The δ¹³C values of the carnivore and omnivores (between -5.6 and -2.0‰), are consistent with the range expected for predominant C₄ plant consumption and/or predating on grazer herbivores, except for the

δ¹³C value of a single suid individual (-12.8‰) that indicates a C3-dominated diet in an open canopy environment. Statistically, only the δ¹³C values of suids are significantly different from all other taxa (Kruskal-Wallis test p = 0.0281, see Table 4.6C for Benferroni corrected p values of taxa pairings), while the δ¹³C values of S.

trigonocephalus, bovids and cervids are statistically indistinguishable.

The δ¹⁸O values of all analysed taxa are moderate to high (between -7.5 and -0.5‰).

The bovids and cervids have significant higher δ¹⁸O values than S. trigonocephalus (Kruskal-Wallis test p = 0.0002, Benferroni corrected p for bovids-S. trigonocephalus = 0.0151 and for cervids-S. trigonocephalus = 0.0288) and suids (see Table 4.6C for Benferroni corrected p values of taxa pairings). As for felids and the single hominin, the sample numbers for these taxa are insufficient for statistical analysis. However, their individual δ¹⁸O values are notably lower than for S. trigonocephalus, bovids and cervids.

4.4.3.2.3 Serial analysis of an Elephas hysudrindicus molar from Sunggu

A total of 29 serial samples were taken along the enamel of a single, unworn lamella of the third molar (M³) present in the Sunggu elephant mandible. The δ¹³C values of the serial samples range between -13.4‰ and -12.0‰ (Figure 4.16), suggesting that the diet of this individual was constantly dominated by C3 plants in a densely vegetated environment. The narrow difference between the minimum and maximum values indicates non-significant shifts of food resources over the growth period covered by the samples. While the δ¹³C values display no clear seasonal pattern, the δ¹⁸O values provide stronger seasonal variation. The δ¹⁸O values of this individual range between - 6.6‰ and -3.9‰, indicating that this animal experienced a considerable seasonal fluctuation in surface water isotopic composition. The serial δ¹⁸O values of the molar show a series of low δ¹⁸O values between sub-sample 1 and 8, indicating wetter conditions during the earlier stage of molar development. A sharp increase occurs at the sub-sample 9, and remains high up to sub-sample 19, suggesting a drier period. The high δ¹⁸O values between sample number 9 and 19 followed by a decrease in the remaining samples, suggest a change to wetter conditions. Sudden decreases occur at sub-samples 20, 21 and 23, indicating the possibility of an unstable seasonal transition

between dry to wet conditions. Nine dry peaks are identified from the δ¹³C values (sample number: 2, 6, 9, 13, 19, 22, 24, 27 and 29), which tend to correspond with eight peaks identified in the δ¹⁸O values (sample number: 2, 6, 9, 13, 17-19, 22, 24 and 29).

Figure 4.16 Serial δ¹³C and δ¹⁸O values of the Elephas hysudrindicus molar from the Sunggu locality. Sub-sample numbers start at the tip of the crown (No. 1, earliest enamel formation) to the base of the crown (No. 29, last formed). Red dashed lines indicate corresponded dry peaks between δ¹³C and δ¹⁸O values.