Early late Miocene extinction of the Listriodontinae
Very different opinions are to be found in the literature on the temporal distribution of Li. pentapotamiae. According to Pilgrim (1926), this species occurred in the Chinji, Nagri, and Dhok Pathan Formations, but all the material he described or figured came from the Chinji Zone. Matthew (1929) cast doubt on the presence of Listriodon in the Middle Siwaliks. However, Colbert (1935) indicated the same distribution as Pilgrim and listed seven items from the lower part of the Middle Siwaliks at Nathor and Phadial. Hussain et al. (1979) and Barry et al. (1982) mentioned the co-occurrence of Hipparion and Listriodon in Daud Khel and locality Y454 in the Tatrot-Andar Kas section. Pickford (1988, Fig. 6) indicated in his range chart the occurrence of Listriodon in the very lowermost part of the Nagri Formation, overlapping with Hipparion. Raza et al. (2002) estimated the temporal overlap of Listriodon and Hipparion to be less than 0.5 My.
Various papers mention more recent and dated records of Listriodon. Flynn et al. (1995) and Barry et al. (2002) indicated the last record of Li. pentapotamiae as 9.6 Ma and 10.3 Ma, recalculated as 10.379 Ma and 10.478 Ma, respectively. The date 10.3 Ma has been repeated by various later authors, including Antoine et al. (2013), who also mentioned Listriodontini or Listriodon sp. with a last record in the approximate interval from 11 to 10 and perhaps even 8.4 Ma. Barry et al. (2002) mentioned a Schizochoerus gandakasensis with a time range of 10.1–8.7 Ma. In reality, this is a species of Yunnanochoerus, a sublophodont palaeochoerid which survived long after Listriodon (Van der Made 2010; Pickford 2017). The molars of all these genera are lophodont or sublophodont and very similar if worn. The Listriodon sp. of Antoine et al. (2013) has not been described, so we cannot discuss it, but given the coincidence in temporal distribution, we do not rule out the possibility that it might represent a Yunnanochoerus. Alternatively, it could represent Li. dukkar. Khan et al. (2012) described fossils from Sethi Nagri and assigned them to Li. pentapotamiae. This locality corresponds to loc. Y311 (Barry et al. 1982, 2002) with an approximate age of 10.72 Ma (range 10.108–10.035 Ma). In the Indian Siwaliks, the last record of Listriodon pentapotamiae is from Nurpur and dates to ~ 9.8 Ma (Patnaik 2013). This is based on an upper canine described by Vasihat et al. (1983) from Nurpur. The curvature of the canine forms a large part of the circumference of a circle as is common in Li. pentapotamiae and possibly Li. dukkar, while in the other Suidae of this time, the curvature forms a lesser part of a circle. This fossil comes from a section that has been magnetostratigraphically dated by Rao (1993) and that was reinterpreted by Sangode and Kumar (2003). This is the youngest dated fossil of the Listriodontinae in the Indian Subcontinent.
Even though we have the date of − 9.8 Ma for the last appearance of Listriodon in the Indian Subcontinent, we do not know the morphologically relevant features of the youngest samples. Nearly, all material described comes from the Chinji Formation or age equivalent strata (Pilgrim 1926; Colbert 1935; Pickford 1988; Van der Made 1996a). An exception is the paper by Khan et al. (2012), which describes Late Miocene material from Sethi Nagri, the type locality of the Nagri Formation, and assigned it to Li. pentapotamiae. The P4 figured by Khan et al. (2012) has a morphology that is normal for Li. pentapotamiae, with a hypoconid near the axis of the tooth and a very modest hypoendocristid. Although there are no upper premolars, nor P2-3 among the specimens, it seems that this material does not represent Li. dukkar, for which we expect a short talonid and reduced hypoconid in the P4.
The age of Pasuda is early Late Miocene and was roughly estimated to be between 11 and 10 Ma (Bhandari et al. 2015). Because Li. dukkar evolved from Li. pentapotamiae and because it is unlikely that two nearly identical species lived in a relatively small area, it is the last documented listriodont species known from the Indian Subcontinent, and given the differences in morphology, it should be clearly younger than Sethi Nagri (10.72 Ma). The locality Pasuda and the extinction of the Indian Listriodon lineage should also be younger and could be close to the last record of the genus at 9.8 Ma as mentioned by Patnaik (2013). Whereas the fossil record in the northern part of the Indian Subcontinent is particularly rich, in the southern part, it is poor. Listriodon dukkar could have retreated to the South and have survived there longer, but we do not have any fossil record, which could confirm or deny this.
In Europe, the last listriodont was Listriodon splendens. It was a very common species and is known from over 100 localities of the units MN6 to MN9. In MN9, it is known from Aveiras de Baixo, Azambujeira (both Portugal), Santiga, Can Ponsic I, Can Llobateres 1 (Spain), Doué-la-Fontaine (France) (Van der Made 1996a), and in addition, it is cited from the Vallesian of Can Flaqué, Relea, subsuelo de Sabadell, Creu del Batlle, Can Amat, and St. Miquel (Golple-Posse 1972), but we did not study this material. Its last dated occurrence is at Can Llobateres 1 (Van der Made 1990a, b, 1996a), a locality dated to 9.78 Ma based on magnetostratigraphy and close to the end of MN9, which is estimated as 9.7 Ma (Casanovas Vilar et al. 2014). No locality with Li. splendens has been assigned to a younger MN unit or has been dated to less than 9.78 Ma.
Asia is a huge continent. In western Asia, like in Europe, the last occurrence of Listriodon splendens is in MN9. It is known from Esme-Akçaköy and Çorak Yerler (Becker-Platen et al. 1975a, b), as well as from the Lower Sinap Formation (Ozansoy 1965). The fossils from Çorak Yerler come from two horizons, but have been studied as a whole, and based on different taxa, different correlations have been proposed, to the Kayadibi and Garkin Faunengruppen (Becker-Platen et al. 1975b). Sen et al. (1998) stated “In summary, there are obviously two different faunas at Çorak Yerler, one possibly of late Astaracian or early Vallesian and the other of late Vallesian or early Turolian.” Geraads (2013) suggested mixing of the collections. Kaya et al. (2016) mentioned two fossiliferous levels and, based on a magnetostratigraphic section with two reversals, favored an age between 8.13 and 7.55 Ma for the upper level, but did not obtain results for the lower level and did not mention Listriodon. The faunal list includes Listriodon (Becker-Platen 1975b), but Hippopotamodon major has also been reported from the locality (Fortelius et al. 1996). If we accept the presence of two fossiliferous levels of different ages (as suggested by Sen et al. 1998), it is possible that the lower level, with Listriodon, correlates to MN9 and the upper one, with H. major, to MN10 or 11.
The age of the Listriodon record from the Lower Sinap Formation can be estimated. In Ozansoy’s (1965) scheme, the Upper, Middle, and Lower Sinap Formation consist of beds 1–19, 20–43, and 44–47, respectively. A small fauna with Listriodon comes from bed 46, about 58 m below bed 25, the lowermost fossiliferous level of the Middle Sinap Formation. This fauna is generally considered to be late Middle Miocene or MN8 (e.g., Steininger et al. 1996). More recent work led to the recognition of a dense sequence of over 30 fossil localities in, or correlated to the Sinap Tepe (= Hill) composite magnetostratigraphic section (Kappelman et al. 2003). The entry of the hipparions is in locality 4 at a height of about 45–50 m and localities 88, 64, 104, and 65 are at heights of about 25–45 m in the profile. These four localities are correlated to chron C5n.2n–4n, which implies ages between 11.056 and − 10.7 Ma (Kappelman et al. 2003). This means that they are early Late Miocene (MN9) in age and not Middle Miocene (MN8). These localities do not have Hipparion. However, like bed 46, they have short faunal lists and probably lack Hipparion for that reason. The lowermost fossiliferous locality is loc. 65 with an interpolated age of 10.899 Ma (recalculated as 11.004 Ma). Probably, Ozansoy’s (1965) bed 46 was not much older than locality 65. Ozansoy’s Middle Sinap fossiliferous levels (beds 20, 23, 25) were identified as OZ01, OZ02, and S01 and are at a height of 125–130 m of Kappelman’s et al.’s (2003) section. Ozansoy’s bed 46 would have been 58 m lower than bed 25, which is at a height of about 67 m at the level of locality 93 in Kappelman’s et al.’s (2003) section, with an interpolated age of 10.488 Ma (recalculated as 10.576 Ma). It seems reasonable to assume an age for the Listriodon from Sinap between 11.004 and 10.576 Ma.
On the other side of Asia, various species of Listriodon were named on the basis of Chinese fossils, but these are indistinguishable from L. splendens (Van der Made 1996a). Deng et al. (2013) indicated the presence of Listriodon mongoliensis and two species of Hipparion in the lowermost Upper Miocene of the Guonigou Formation in the Linxia Basin.
The last African listriodont is Lopholistriodon kidogosana from Member D of the Ngorora Formation (Pickford and Wilkinson 1975). This member was reported to date to 9.82–9.68 Ma (Bishop and Pickford 1975), but more recent work suggested older ages for the Ngorora Fm (Tauxe et al. 1985; Deino et al. 1990) and Pickford (2001) situated locality 2/11, with the last Lopholistriodon kidogosana, between 11.84 and 11.54 Ma. Tsubamoto et al. (2017) described a tooth fragment of a listriodont with an estimated age of 10 Ma and assumed that Li. splendens or Li. pentapotamiae dispersed to Africa. This is only a small fragment with a morphology and size that also fit Lo. akatidogus (= Lo. bartulensis) and such an identification would not need to be explained by an intercontinental dispersal. In contrast to the two species of Listriodon, these two species of Lopholistriodon are known from fewer localities. In addition, it should be noted that the African upper Middle and lower Upper Miocene fossil record is less abundant than that from the preceding and following times and as a consequence, the time of extinction is less precisely known.
The last appearances of the Listriodontinae are in the early Late Miocene: Listriodon splendens in Europe and western Asia after 9.78 Ma, Listriodon dukkar in the Indian Subcontinent at or shortly after 9.8 Ma, and a possible Lopholistriodon akatidobus in Africa around 10 Ma. The ages of the decrease in abundance leading to extinction, the death of the very last individual, and the youngest recorded fossil of a species are not the same, but the difference is only relevant if we know it to be large compared to the precision we can obtain in dating. Given the temporal range of about 10 My of the Listriodontinae, the extinction of the last three lineages adapted to different climates and environments is remarkably close in time.
Suoidea are specialized in rooting and during seasons when their favourite food is not available; they have access to a source of food for which they do not compete with other large mammals. They may also eat soil that is rich in organic material. Suoidea complement their diet with food of animal origin: animals they catch (e.g., rodents, reptiles, and birds), carrion, animals living in the soil (e.g., earthworms, larvae), and on occasions, this may make up 88% of their diet (Wilson and Mittermeier 2011). The specialized dentition of the Listriodontinae suggests that they did not ingest important quantities of food of animal origin. Skull morphology suggests that the Listriodontinae were never particularly strongly adapted to rooting and incisor morphology and wear indicate that they progressively abandoned the rooting habit (Leinders 1977a; Van der Made 1996a). This has several implications. One is that their diet was not complemented by animals living in the soil and they depended on leaves and fruit for their proteins. The other is that they are expected to have been limited to environments where nutritious leaves (and fruit) were available year-round.
Many Suoidea are very selective when feeding. The snout of Babyrousa is narrow and it is reported to select precisely different plant parts and in the case of captive individuals even to shell peanuts and peel bananas (Leus 1994). By contrast, the Listriodontinae have very wide snouts and were probably not very adept in precisely selecting the best plant parts. This is primarily due to their very wide incisors, which have been interpreted as an adaptation to bulk feeding on forbs but not on grasses (Van der Made 1996a, 2003b).
Because of the dental morphology and macrowear, Li. splendens has been interpreted to be a specialized folivore (Leinders 1977a, b) and this is confirmed by microwear analysis (Hunter and Fortelius 1994). Stable isotopes are compatible with folivorous diet, different from that of other Suidae (Quade et al. 1995; Domingo et al. 2012). Except for Phacochoerus and Hylochoerus, most species of wild pigs and peccaries have a preference for fruit and shift to roots and tubers when fruit is not available and domestic pigs have been reported to have a preference for sweet fruits (Leus 1994). Fruits tend to be rich in carbohydrates (sugar and starch), but tend to have less than 0.5% crude protein of the wet weight. Tapirs have lophodont cheek teeth, like Listriodon, and their diets consist of 63–86.5% of leaves, 8.1–37% of fruit, and the rest of twigs, bark, etc. (Wilson and Mittermeier 2011). We are not aware of anyone having addressed the possibility that Listriodontinae ingested important amounts of fruit.
Listriodon splendens lived in environments with abundant fruit. Fuss et al. (2018) documented caries in the hominid Dryopithecus from the locality St. Stefan in Austria (about 12.5 Ma), caused by the ingestion of fruit. The environment was forest, dominated by Quercus, Fagus, Castanea, Carya, Pterocarya, Engelhardia, and Juglans. Nine fossil plant species, mostly from the understory of the forest produced ripe sugar-rich fruits during the months June–December: Prunus sp. 1 and 2, Vitis sp., Elaeagnus sp., Morus cf. nigra, Arbutus sp., Castanea sp., Carya sp., and Toddalia sp. Listriodon splendens is one of the most abundant species from this locality and may have fed on the same fruits.
Suoidea may ingest large quantities of protein-rich nuts and acorns. Acorns allow Sus scrofa to gain weight to get through the winter. Their bunodont dentition is well suited for such a diet. Whether Listriodontinae ingested important quantities of nuts is unknown, but their lophodont dentition seems less suited for such a diet. The dental adaptations of Li. pentapotamiae and the last Lopholistriodon species are similar to those of Li. splendens and we assume that this was also the case with their diets.
Herbivores depend on a symbiosis with microorganisms which digest cellulose for them and thus also make cell contents available. Cellulose fermentation takes place in fermentation chambers either in the foregut or hindgut, and the position has further implications (Janis, 1976). Ruminants are the most specialized foregut fermenters. They are more efficient in digesting cellulose than hindgut fermenters, and take more time to digest food, and their digestive tract and contents comprise a larger proportion of the total bodyweight. Whereas food intake in ruminants is limited by rumen fill, hindgut fermenters may increase intake, decrease digestion time and efficiency, and, as a result, increase total nutrients obtained. Sugar is fermented along with cellulose in the rumen, and as a result, ruminants have lower blood sugar levels, while hindgut fermenters absorb the sugar before it reaches the place where fermentation takes place. The way nitrogen becomes available to the herbivore is different and seems to be more efficient in foregut fermenters. Ruminants have the most complex forestomachs, but also camels, hippopotamuses, peccaries, and the babirusas are foregut fermenters, while Potamochoerus, Phacochoerus, and Sus are hindgut fermenters (Langer 1986; Leus et al. 1999; Clauss et al. 2008). Peccaries and babirusas benefit from the formation of microbial proteins by fermentation in the forestomach. Those microbial proteins are subsequently digested in the stomach, resulting in higher protein digestibility and lower protein requirements. The hindgut fermenting Suidae have a more efficient access to sugars in fruit and leaves. Either foregut fermentation evolved several times in the Artiodactyla or it is the ancestral state and some Suidae became hindgut fermenters. Listriodontinae, which split off early from the other Suidae, may have been foregut fermenters, like the babirusas and peccaries. This may have had consequences for the protein requirements of the Listriodontinae.
The energy and protein requirements of several species of ungulates have been studied in detail. Borges et al. (2017) and Nogueira-Filho et al. (2014) gave the minimum daily protein intakes in relation to body weight: Pecari tayacu 514 mg N/kg0.75, Tayassu pecari 336.5 mg N/kg 0.75, Cervus elaphus 680 mg N/kg0.75, Odocoileus virginianus 710 mg N/kg0.75, and Philantomba monticola 643 mg N/kg0.75. The requirements are also expressed as a minimum % of crude protein of the diet (dry matter), which is 5.4% for Pecari tayacu, 4.5% for Tayassu pecari, 4–7.1% for Babyrousa babyrussa, 6.98% for Boselaphus, and 8.27% for Antilope cervicapra, but 12% for domestic pigs and even 15% for lactating sows (Das et al., 2012; Wilson and Mittermeier 2011; Nogueira-Filho et al. 2014; Borges et al. 2017). Though we cannot measure the protein requirements of the Listriodontinae, these may have been more similar to those of the peccaries and babirusas and even ruminants than of Sus scrofa. While the Listriodontinae were probably not omnivorous, did not root, nor eat nuts, their diet may have consisted of protein-rich leaves and possibly protein-poor, but sweet, fruit. Fruit has relatively low protein contents and fruit bats (Pteropodiade) overingest energy to meet their protein requirements (Thomas 1984). Obesity is a problem in overfed zoo babirusas to an extent that it may interfere with reproduction (Leus 1994), so overfeeding on low protein food may not have been an option for the Suoidea.
The extremely large male canines of Li. splendens, Bunolistriodon latidens, and B. meidamon have been interpreted to be used in display in inter-male interactions, suggesting a preference for habitats with good visibility at a distance and thus relatively open landscapes (Van der Made 1996a, 2003b). These canines reached heights (“lengths”) of about 30 cm, and had large diameters and large radii of curvature. While the height of these teeth can be measured only in relatively few complete specimens, the diameters and radii of curvature can be measured in many broken specimens. Although cheek teeth of Li. pentapotamiae are only a little smaller than those of Li. splendens, the canines of the males are much smaller (Fig. 7). Most Listriodontinae, including Lopholistriodon and Li pentapotamiae, retained modest canines, suggesting that these species may have lived in closed habitats, feeding on herbaceous vegetation in the under story or fallen fruit, while Li. splendens may have lived near the interface of closed and open environments or in a mosaic landscape.
Body weight is an important parameter in ecology: it affects strength, speed, the relationship with predators, energy and protein requirements, food selection, and through intestine length and the passage time through the intestines, and it also affects digestion. Various ways to estimate body weight have been proposed, based on the length x width of the first lower molar (Legendre 1986), the length of the second upper molar (Fortelius et al. 1996), or the length x width of the astragalus (Martinez and Sudre 1995). The body weights of Li. dukkar, Lo. akatidogus, and Lo. akatikubas are estimated on the basis of the M2 as: 92 (n = 1), 77 (n = 1), and 23, 31, and 35 kg (minimum, mean, maximum; n = 4). The body weight (minimum, mean, maximum) of Li. pentapotamiae is estimated as 82, 122, and 182 kg (n = 11; M1), 61, 87, and 124 kg (n = 41; M2), and 32, 57, and 73 kg (n = 4; astragalus). For Li. splendens this is: 102, 148, and 281 kg (n = 58; M1), 70, 108, and 177 (n = 92; M2), and 88, 111, and 131 kg (n = 19; astragalus).
Legendre (1986) used cenograms to study ecology. As he used them, these are diagrams with the natural logarithm of the average body weight of a species on the vertical axis and the species ordered from large to small on the horizontal axis. He noted that there is a gap between the large and small mammals and that intermediate sizes are absent. This gap is very large in deserts, smaller in savannah and woodland, and absent in rain forest. This is logical in view of the anti predator behaviour: sprint and hide in small mammals living in or near closed environments, outrun the predators in the intermediate sized mammals living in more open landscapes, and defence in the large mammals. The short-legged Suidae are not the type of animals that outrun a predator, and especially small species like Lo. akatikubas would not venture far in an open environment.
Listriodont extinction in the European context
The context of the extinction of the last Listriodon in Europe is better known than for other regions of the world. Its last record is in Can Llobateres 1 (9.78 Ma) and coincides there with the last records of other species of Suoidea. An important turnover in the Suoidea in Spain and elsewhere in Europe, resulting in a major decrease in species richness, was documented (Van der Made 1988, 1990a, b, 1991, 1997c). This appeared to be part of a faunal event, which also involved the extinction of Hominoidea and was called the mid-Vallesian Crisis (Agustí and Moyà Solà 1990; Moyà Solà and Agustí 1990), later shortened to Vallesian Crisis. This event was also noted in the Bovidae: existing lineages went extinct and new arrivals included Tragoportax and Protoryx (Moyà Solà 1983; Moyà Solà and Agustí, 1990; Alcalá and Montoya, 1990).
The causes and extent of the Vallesian Crisis have been much debated in the literature. The Vallesian Crisis was believed to have been caused by a change towards more open and dry landscapes or to increased seasonality, resulting in an increase of deciduous vegetation and seasonal variations in the availability of fruit (Van Dam 1997; Suc et al. 1999; Agustí et al. 2003). It has also been correlated to Mi7, an oxygen isotope event around 9.7 Ma, reflecting decreased global temperatures (Agustí et al. 2013). The Vallesian Crisis has been considered to be continent-wide (e.g., Fortelius et al. 1996a) or only local (e.g., Madern et al. 2018). Hominoidea went extinct in western and central Europe, but are known from younger localities in Greece, Bulgaria, Turkey, and Georgia (Koufos and De Bonis 2006; Spassov et al. 2012; Kaya et al. 2016; Agusti et al. 2020), which has its parallel in longer survival of palaeotropical plants in SE Europe and the Caucasus (Kovar-Eder et al. 1996). The existence of the Vallesian Crisis has been doubted, based on the micromammal record (Casanovas Vilar et al. 2014).
The fossil record of the Suoidea supports the existence of the Vallesian Crisis (Fig. 8). Listriodon splendens, Propotamochoerus palaeochoerus, and Parachleuastochoerus steinheimensis were common species with temporal ranges lasting 5–2 My and their last record is at Can Llobateres. None of these species has been identified from any locality younger than MN9 (or 9.7 Ma) or dated younger than Can Llobateres 1 (9.78 Ma). Parachleuastochoerus crusafonti and Schizoporcus are less common, but both have their last record at La Tarumba (Van der Made 1990a, b, 1997c). La Tarumba is placed in MN10, is dated to 9.56 Ma (Casanovas Vilar et al. 2016), and is only very little younger than Can Llobateres 1. Conohyus is not very abundant and Albanohyus is rare, but both have their last record in MN9 (11.2–9.7 Ma), though no precise ages are available for these localities (Van der Made 1996b; Van der Made and Morales 2003). Nowhere in Europe or Anatolia, these Suoidea survived, nor do any of them co-occurr with H. major, a very common species from MN10 onwards. Within the genus Hippopotamodon (= Microstonyx), H. antiquus was replaced by H. major (Van der Made 1990a, b), which happened either late in MN9 or early in MN10. With up to eight contemporaneous species in Europe and up to five species in one locality before and one or two contemporaneous species after the Vallesian Crisis, European suoid species richness never recovered. While variations in sampling density may affect local sequences, this pattern is observed all over Europe. None of the mentioned species survived anywhere in Europe after MN9. Everywhere in Europe species richness decreased.
Agustí et al. (2003) described floral change coincident with the Vallesian Crisis. Previous to the crisis and still in Can Llobateres 1, there was a humid subtropical flora, while flora from Terrassa—Talud Sud Autopista (9.02 and 9.23 Ma) consists of 36 taxa, of which about 45% are deciduous trees, 15% Mediterranean elements, 33% evergreen trees, and 7% warm evergreen elements, suggesting about 1000 mm mean annual precipitation, mean annual temperatures between 16 and 19ºC, and seasonal availability of fruit. Listriodon may have lived in other habitats in Europe, and, for instance, is known from localities in the interior of Spain, which today and probably also in the Miocene were drier. However, its last records are in Can Llobateres and other localities in the same area, or with an expected similar environment, and the documented increase in seasonality, restricting the staple food of Listriodon, may have been decisive.
As far as the Suoidea are concerned, the Vallesian Crisis is a reality and Listriodon splendens became extinct as part of this crisis. If the study of other taxonomical groups cannot confirm the existence of the crisis or of environmental change, this can be because these groups were not affected in the same way. Though generally the age of the crisis is considered 9.7 Ma, the extinction of Listriodon and other Suoidea occurred after 9.78 Ma, that of still other Suoidea after 9.56 Ma, and the first dates for the replacing suid and new vegetation date to 9.4 Ma. Therefore, if there were two events, they occurred in the period from 9.78 to 9.4 Ma, and if it was only one, it was between 9.56 and 9.4 Ma. From the foregoing it appears that environmental change and increase in seasonality restricted the availability of the food of L. splendens, which may well have been the cause of its extinction in Europe.
Listriodont extinction in the West Asian context
The Anatolian fossil record is less complete than that of the whole of Europe and relevant material of the Suoidea has not yet been published, but the available data fit perfectly the pattern observed in Europe. Figure 9 shows the various species of Suoidea from Anatolia, which in Europe were involved in the Vallesian Crisis and shows a selection of relevant sites with their dates or with an approximate position according to biostratigraphy. The “Faunengruppen” (faunal associations), indicated in the figure, were defined by Becker-Platen et al. (1975a) on the basis of Turkish localities. These Faunengruppen were never widely used because of the MN units which were introduced at the same time (Mein 1975). As argued above, the locality Çorak Yerler has two different levels, one with Listriodon and another one with Hippopotamodon major. The Alçitepe Mb of the Eceabat Formation on the Gallipoli (Gelibolu) peninsula has several units: the Nebisuyu, Sigindere, and Degirmendere units, correlated to the Aragonian, Vallesian, and Turolian, respectively (Van der Made and Tuna 1999). The locality Nuri Yamut is in the Sigindere unit.
Several of the sites in Fig. 9 overlie or underlie dated tuffs (Becker-Platen et al. 1975a, 1975b, 1977). A composite faunal list of the two localities Yeni Eskihisar 1 and 2 was used to define the Faunengruppe of this name. However, Yeni Eskihisar 2, mainly a micromammal locality, is overlain by two tuffs, dated 13.2 ± 0.35 Ma and 11.1 ± 0.2 Ma, which again are overlain by Yeni Eskihisar 1, which is mainly a large mammal locality. The latter locality has Hippopotamodon antiquus (Fortelius et al. 1996) and together with the date, this points to the locality being MN9 and Vallesian, while the other locality is MN7 + 8 and Aragonian. The localities Bayirköy and Eski Bayirköy overlie a tuff dated to 10.2 ± 0.15 Ma and apparently also one dated 9.25 ± 0.2 Ma (Becker-Platen et al. 1977, Fig. 9), but are correlated to the Kayadibi and Kinik Faunengruppe, respectively (Becker-Platen et al. 1975b). Ozansoy’s Middle Sinap fossiliferous beds 20, 23, and 25 correspond to OZ01, OZ02, and S01 of Kappelman et al. (2003), and have interpolated ages of 9.683–9.590 Ma (recalculated as 9.766–9.673 Ma).
As can be seen in Fig. 9, the Suoidea show the same pattern as in Europe. Hippopotamodon antiquus occurs in MN9, while localities of MN10 and younger have H. major, like in Europe. Even though H. major is known from many sites (e.g., Pickford and Ertürk 1979; Fortelius et al. 1996; Van der Made et al. 2013), none of these sites is dated and known to be older than Igbek, nor any is placed in an MN unit older than MN10. Bayirköy and Eski Bayirköy have Hippopotamodon, but we did not study the one from Bayirköy. Five of the eight species that went extinct or were replaced during the Vallesian Crisis in Europe did the same here, including Li. splendens. This happened between Sinap “bed 20” (9.673 Ma) and Igbek (9.118 Ma). Beyond Anatolia, the western Asian fossil record is not very well known, but the Hippopotamodon from Marageh Kopran I (Iran; about 9 Ma; Fortelius et al. 1996; Bernor et al. 1996), as well as in Udabno (Georgia; MN10) and Eldar (Azerbaidjan; MN10/11) are compatible with the event having occurred in a larger part of western Asia. Even though dating is less precise, the turnover may have happened at the same time. In Europe, the Vallesian Crisis occurred between 9.78/9.56 and 9.369 Ma and the faunal turnover in Anatolia occurred at about the same time between 9.659 and 9.118 Ma.
This faunal event in Anatolia seems to coincide with a change in the flora. In Turkey, a series of “Pollenbilder” or sporomorph associations (“zones”) have been defined and are correlated to the Faunengruppen and MN units (Benda et al. 1975; Benda and Meulenkamp 1990). These provide information on the environment in which Listriodon splendens lived and went extinct. The Eskihisar Pollenbild includes many tropical and few subtropical elements and the following Yeni-Eskihisar association is transitional. Listriodon splendens is present in localities correlated to those Pollenbilder, but not in sites correlated to the next one. The following Kizilhisar association has but few subtropical elements, but more conifers, species related to oaks, and sedges and grasses, indicative of steppe like to semi-arid conditions (Benda and Meulenkamp 1990). The Yeni Eskihisar Pollenbild is defined on a sample from the level of the Yeni Eskihisar 1 micromammal locality, but is said to be typical of the upper part of the Sekköy Formation (which also includes Yeni Eskihisar 2, < 11.1 Ma) (Becker-Platen et al. 1977). The Yeni Eskihisar Pollenbild is known only from a few sites with fossil mammals, but has been correlated to mammal sites. The transition to the Kizilhisar floral associations occurs between the Yeni Eskihisar 1 and Kayadibi mammal associations, and possibly within the Esme Akçaköy Faunengruppe (Benda et al. 1975). The locality Küçük Çekmece is of the Kayadibi faunal association, which has H. major and the Kizilhisar pollen association. Benda and Meulenkamp (1990) indicated that Kastellios Hill 1 in Crete has the Kizilhisar pollen association. This locality is placed in MN10 and correlated to chron C4Ar.2r (Steininger et al. 1996). This places the transition between the two pollen zones, which marks a decrease in temperature and a change to a landscape with more grasses between 11.1 ± 0.2 and 9.426–9.647 Ma. The whole Esme Akçaköy faunal association is between these two dates and precedes the faunal change, which also marks the MN9–10 transition. Benda’s Pollenbilder are not used anymore, and their use in correlation for the Lower and Middle Miocene has been criticised. However, these Pollenbilder have been found in or lithostratigraphically correlated to sites with fossil mammals of which the age is known and the environmental interpretation for these sites or times should remain valid. More recent work confirms in a broad sense a transition of more humid environments that included subtropical forests to steppe (Yavuz-Isik and Toprak 2010; Bouchal et al. 2017; Steinthorsdottir et al. 2021).
The available data show or suggest that: (1) the extinction of Listriodon in western Asia formed part of a turnover event, which involved the same species of Suoidea as the European Vallesian Crisis; (2) the events in both areas are coincident; and (3) also in western Asia, these faunal events may have coincided with a major floral change.
Listriodont extinction in the context of the Indian Subcontinent
Whereas suoid species richness in Europe declined during the Vallesian Crisis, it remained high in the Indian Subcontinent till well into the Pliocene (Van der Made 1991). Barry et al. (2002) found a generally low level of faunal turnover in the Siwaliks in the interval of 10.7–5.7 Ma, with peaks at 10.3, 7.8, and 7.3–7.0 Ma. The event at 10.3 Ma involved also non-suid taxa, but coincides with the last appearance of Li. pentapotamiae and Conohyus sindiensis. The last appearance of ?Hippopotamodon “Y450 unnamed species” and the first appearance of Hippopotamodon sivalense and Propotamochoerus hysudricus was indicated to be at 10.2 Ma. The temporal ranges of these and other Suoidea as given in the literature are indicated with thick lines in Fig. 9.
As we have seen already, there are records of Listriodon in the Indian Subcontinent, which are considerably younger than the 10.3 Ma, indicated by Barry et al. (2002), but this is not the only one. Conohyus sindiense (= Retroporcus) was cited from a magnetostratigraphic section at Tinau Khola (Munthe et al. 1983). This record is from a long interval with normal polarisation, correlated to C5n.2n, and, based on Hilgen et al. (2012), its interpolated age is 9.984 Ma.
Von Meyer (1866) described fossils from three different fossiliferous levels at Kushalgar and named Sanitherium schlagintweiti on the basis of material from a level with Hipparion. Later authors stated that the specimen came from Chinji-equivalent strata (Pilgrim 1926; Colbert 1935). The description of new material from Tapar from a level with Hipparion shows that the species did indeed survive till after the arrival of that equid (Bhandari et al. 2015).
Hippopotamodon major is present in Sethi Nagri (10.072 Ma), where it existed along with H. sivalense (Van der Made and Hussain 1989). This H. major may correspond to Dicoryphochoerus robustus and D. titanoides of Pilgrim (1926) and to ?Hippopotamodon “Y450 unnamed species” of Barry et al. (2002), because it is the only other Hippopotamodon mentioned by these authors. If this is correct, the curious situation exists that in the Indian Subcontinent, around 10.2 Ma, H. major is replaced by the larger H. sivalense, whereas in Europe and Anatolia, between 9.78 and 9.4 Ma, H. major replaced the large H. antiquus, which is very close to H. sivalense.
In Pickford’s (1988) scheme, Hyotherium pilgrimi occurred in the Chinji Formation and the very base of the Nagri Formation, and was replaced by Propotamochoerus hysudricus in the Nagri, Dhok Pathan, and Soan formations. Pickford (1988) named H. pilgrimi and listed in its synonymy the following species named by Pilgrim (1926): Propotamochoerus salinus, P. uliginosus, Dicoryphochoerus chisholmi, ?D. haydeni, and D. instabilis. All these species would have priority over H. pilgrimi, but this was not sufficiently discussed. The species P. salinus was described from Tapar as a new genus Kachchchoerus (Bhandari et al. 2015). Pickford’s (1988) generic diagnosis of Hyotherium states that the P4 is without sagittal cusplets. According to this criterion, the holotype of H. pilgrimi does not fit Hyotherium and it is more likely that Pilgrim (1926) was right and that it belongs to Propotamochoerus. Whatever the names applied, the primitive species of Propotamochoerus was replaced by P. hysudricus and this may have happened around 11 Ma (more conform to Pickford 1988) or around 10.3 Ma (more in line with Barry et al. 2002) or at a still other date, but this remains to be documented.
Lophochoerus nagrii was not mentioned by Barry et al. (2002), but according to Pickford (1988), it made a short appearance in Nagri times, but went extinct afterwards. Barry et al. (2002) gave the temporal distribution of Tetraconodon magnus as 10.0–9.3 Ma. In fact, there is a size increase in the lineage T. minor—T. intermedius—T. magnus (Van der Made 1999), but we lack precise data on the temporal distribution of these species. According to Barry et al. (2002), Schizochoerus gandakasensis appeared between 10.1 and 8.7 Ma. This genus is a junior homonym and its replacement name is Schizoporcus (Van der Made 2010), but this species belongs to Yunnanochoerus (Van der Made 1997a; Pickford 2017).
It appears, thus, that there is some evidence for a turnover in the Suoidea around 10.3–10.2 Ma according to the data of Barry et al. (2002), but that data from India and Nepal indicate that Listriodon and R. sindiense survived till around 9.8 Ma. The species involved are closely related to those in Europe and Anatolia, but not identical.
Barry et al. (2002) also gave data concerning a faunal turnover in the Tragulidae, with two species last occurring at 10.5–10.4 Ma and three species appearing at 10.4–10.3 Ma. It is not quite clear how this compares with the data presented by Barry (2014), showing that Tragulidae made up 30– > 60% of Tragulidae and Bovidae combined until 10 Ma, around 50% at 10 Ma and a decrease from about 20–5% from 9 to 6 Ma. This spectacular drop occurs broadly coincidently with the final extinction of Listriodon. Living Tragulidae are primarily frugivores, but may feed on leaves and occasionally eat carrion or may catch animals (Wilson and Mittermeier 2011). Many of the fossil tragulid species reached much larger sizes than the living species and may have been folivorous. Bovidae range from browsing folivores to grazing bulk feeders and their rise in abundance was probably related to a shift in vegetation.
Stable isotopes from soil carbonates and from teeth provide information on the environment and diet. However, this information is mainly restricted to differentiating C3 and C4 vegetation. After 7.3 Ma, C3 vegetation (probably trees and shrubs) is replaced by C4 vegetation (grasslands) in the Siwaliks of Pakistan (Cerling et al. 1993; Quade and Cerling 1995). Sanyal et al. (2004) studied soil carbonates near Haripur (India) and found a similar change around 6 Ma. Stable isotopes from tooth enamel of fossils from the Indian part of Siwaliks show Late Miocene dietary shifts among various mammallian taxa including primates (Patnaik 2015; Patnaik et al. 2014, 2019). Also it has been hypothesised that this shift in vegetation could have led to extinction of the hominoids in the Siwaliks (Patnaik et al. 2005). Stable isotope data suggest that Li. pentapotamiae lived in environments with C3 vegetation, which is in line with its dental morphology, and that it went extinct long before C4 grasses started to dominate the landscapes.
The Siwaliks of northern India and Nepal have an abundant record in leaf and fossil wood floras (Prasad 1971; Awasthi 1992; Prasad and Awasthi 1996; Prasad and Pandey 2008), but most of these fossil plant sites are situated in India (Awasthi 1992) and further to the east than the sites with Li. pentapotamiae. The fossil plants indicate a luxuriant humid forest environment, which in the Upper Siwaliks became drier, more open and with grasses. Of particular interest is a sedimentary sequence containing 52 horizons with plant macrofossils and 584 pollen samples at Surai Khola, tied to a magnetostratigraphic section, described by Corvinus and Rimal (2001). They interpreted tropical broad leaved evergreen rainforest and swampy environments from 13 Ma onwards, which is compatible with the sedimentology of the Chinji Formation in Pakistan (Zaleha 1997; Willis and Behrensmeyer 1995), a gradual shift towards moist semi-evergreen forest between some 10 Ma and 9 Ma, after about 7.5 Ma a shift to dry deciduous forest, and still later a shift to grasslands. They noted that the shift towards grasses in their section is much later than in Pakistan. The evergreen rain forest included trees like Dipterocarpus, which may be up to 50 m tall. Srivastava et al. (2018) used the plant fossils in the same section to estimate the principal climatic parameters for the period between 13 and 11 Ma as 1748–2869 mm mean annual precipitation and a mean annual temperature of 21.1–25.4 °C. For the period from 9.5 to 6.8 Ma, they estimated 2592–3151 mm mean annual precipitation (which is an increase) and a mean annual temperature of 26–-27 °C and estimated that seasonality in precipitation increased. This increase in precipitation contrasts with the smaller river channels in the Dhok Pathan Formation in Pakistan. Perhaps, one of the two is a local phenomenon. It should be taken into account that the Himalaya was already rising and that, at present, there is a narrow area of high precipitation along the mountain front, so fossil plant sites may not be representative of the environment more to the South and West, where most of the fossil mammals are found. Quade et al. (1995) analysed stable isotopes from soil carbonates at Surai Khola and found a similar pattern as in Pakistan.
The available geochemical, paleontological, and sedimentological information suggests that Li. pentapotamiae may have lived in humid environments with a luxuriant C3 vegetation, probably forest and that it may have fed on leaves of the plants of the understorey and perhaps also fruit. It survived into times with increasing seasonality. Its last records, around 9.8 Ma, date from a time when there was an increase in deciduous vegetation and an opening of the landscape, and coincided with the frugivorous Tragulidae becoming less abundant and Bovidae becoming more abundant. Listriodon went extinct well before landscapes became dominated by C4 grasses.
Listriodont extinction in the African context
In Europe and the Indian Subcontinent, suoid diversity was high with Cainochoerinae, Tetraconodontinae, Suinae, and Taucanaminae living along with the Listriodontinae, as testified by the co-occurrence of several of these in many localities (Van der Made, 1990a, b, 1991). By contrast, in Africa, diversity of the Listriodontinae was higher, while that of other Suoidea was much less (Van der Made, 1996a, b). A species of Lopholistriodon survived till about 10 Ma (Fig. 10).
Within the 10 My duration of the Listriodontinae in Africa, there were three major peaks in faunal turnover, around 17, 13.7, and 9.7 Ma (Van der Made 2014). The first two coincide with turnover in the Listriodontinae and sandwich the Mid-Miocene Climatic Optimum, which is when listriodonts reached their maximum geographic extension and diversity. The event around 9.7 Ma is close to the extinction of the Listriodontinae at ~ 10 Ma. It is noteworthy that the extinction of the last Listriodontinae and a major peak in faunal turnover in Africa are so close to the European Vallesian Crisis.
Initially, Bovidae formed a small proportion of the large mammals in Africa, but their diversity started to increase leading to a spectacular diversity of the Quaternary (Van der Made 2014, fig. 6). Other Suidae may have escaped direct competition, because of their omnivorous diet and rooting habits, but dental morphology suggests that Lopholistriodon was folivorous and more prone to suffering from the bovid radiation.
The eastern African palaeobotanic record is not very dense, but the best sequence of sites is from the Ngorora and Mpesida Formations in the Tugen Hills (Kenya; Jacobs et al. 2010), which also records the latest African Listriodontinae. The flora from Kabarsero (12.6 Ma) has 11 out of 25 taxa that grow today in wet forests or rain forests, including herbaceous plants and grasses that are found in the understory of wet forests (Jacobs and Kabuye 1987; Jacobs and Winkler 1992). Accordingly, the environment was interpreted as moist to wet forest. The flora from Waril (about 10 Ma) is dominated by a species of legume growing in seasonally dry environments (Jacobs et al. 1999). The last Lopholistriodon species must have lived in this environment and will have fed on this legume. The paleoflora from Kapturo (7.2–6.7 Ma) was interpreted as deciduous woodland with seasonality in the moisture (Jacobs and Deino 1996). The silicified wood from Rurmoch (> 6.37 Ma) with stem diameters of > 90 cm suggests trees over 50 m tall, and the association was interpreted as indicative of wet lowland rainforest, while stable isotopes from paleosols and tooth enamel indicate the presence of locally open habitats (Kingston et al. 2002).
These data suggest that Lopholistriodon lived in wet forest, may have fed on leguminous plants, and went extinct when the vegetation was on its way to become more seasonal and drier and when bovid diversity increased.
Possible reasons for the extinction of the Listriodontinae
The last record of the Listriodontinae in Europe is at 9.78 Ma, which is before the Vallesian Crisis, (9.78/9.56–9.396 Ma), in the Indian Subcontinent at ~ 9.8 Ma or somewhat later, and in Africa around 10 Ma. The extinctions in these areas seem to have occurred within a remarkably short period, suggesting a common cause. The three different lineages that went extinct lived at different latitudes in Europe and Anatolia (about 38–50° N), the Indian Subcontinent (about 21–34° N), and Africa (about 0°) and consequently in different climates and environments. If there is a common cause for these three extinctions in different continents, it has to be a global event or process, but most of the changes which coincided with the extinction of the Listriodontinae are local or gradual.
It has been suggested that the Vallesian Crisis was related to the Mi7 event (Agustí et al. 2013). Such events are global. Miller et al. (1991) named the Mi events for rapid increases in the values of the record of δ18O from the skeletons of benthic foraminifera in the Miocene. These events reflect sudden decreases in temperature and eustatic sea level and increases in global glacial ice volume. The timing of these events seems to be related to 400 ky, 100 ky, and 180 ky cyclic variations in precession, obliquity, and asymmetry of the obliquity cycle. Pagani et al. (1999b) situated the Mi6 event around 9.6 Ma and coincident with a peak value of pCO2, reaching the highest value (about 310 ppmv) in a period of over 15 My. The Mi7 event could date to 9.56, 9.40, 9.31, 9.22, 9.14, 9.04, or be still younger (Westerhold et al. 2005). It is possible that a temperature decrease combined with pronounced seasonality may have caused seasonal limitation of nutritive plant parts, causing the extinction of Suoidea and early European hominids, but it is more difficult to see how this may have affected the Listriodontinae living at the Equator.
Variations in atmospheric CO2 concentration are global and have a great impact on the vegetation and on herbivores, as well as humans. The projected increase in CO2 levels by 2050 is considered to put at risk 148.4 million people, who depend mainly on rice, wheat, barley, and potato for their protein intake (Medek et al. 2017). The protein content of these foods is expected to drop by 6–14%. Two billion people suffer from dietary iron and zinc deficiencies, which cause millions of death, and the problem is expected to increase with rising pCO2 (Myers et al. 2014). A rise in pCO2 will enhance productivity and will increase carbohydrate and sugar content in leaves, grains, and fruit (e.g., Högy and Fangmeier 2008; Dong et al. 2018; Yang et al. 2020), and it also leads to a decrease in zinc and iron, and protein content (Ehleringer et al., 2020). Low protein, zinc and iron intake affects growth and may lead to a series of pathologies. There is evidence that elevated pCO2 may lead to reduced growth rates in cattle (Ehleringer et al., 2020). Meeting the protein requirements of captive ruminants is a concern (e.g., Priebe and Brown 1987; Crissey 2005). There is reason to believe that changes in pCO2 may have affected the nutrition of Miocene ungulates.
There are various ways to reconstruct the pCO2 of the past (e.g., Pagani et al. 1999a; Pearson and Palmer 2000; Demicco et al. 2003; Sosdian et al. 2018) and the results may differ according to the method applied. During the past 200 My, atmospheric CO2 concentrations of over 1000 were common and peaks of over 2000 ppmv (parts per million volume) occurred (Pearson and Palmer 2000), but the Miocene pCO2 was much lower. Most methods give relatively low values for the Miocene (below 600 ppmv), but higher values would fit better existing climate models (Steinthorsdottir et al., 2021). In addition, some of the records lack detail; others are of short periods or have long hiatuses. Van de Wal et al. (2011) reconstructed detailed records of pCO2 and temperature for the past 20 My. Temperatures and pCO2 were high between about 17 and 13.6 Ma, a period that has been called the Mid-Miocene Climate Optimum (Figs. 8, 10). Around 13.6 Ma, there was a rapid decline in temperatures and pCO2 and then a more gradual decline until about 8 Ma, punctuated by several moments of acceleration. Notably, at 9.539 Ma, a record low value of pCO2 was reached. Super et al. (2018) used the alkenone proxy and the resulting curve (Fig. 8) differs from that of Van de Wal et al. (2011) in generally lower values, but both curves resemble each other in general shape with high values for the Mid-Miocene Climatic Optimum and lower values later. There is a peak low value at 9.4 Ma, but the record is not very dense with the next older data point at 9.7 Ma. Sosdian et al. (2018) used boron isotopes to reconstruct pCO2, gave various variants based on different assumptions and methods, but all resulted in higher estimated values. The general shapes of these pCO2 curves are similar to the curve by Van de Wal et al. (2011) in a marked decrease in the pCO2 after the Mid-Miocene Climatic Optimum and in having a peak low value at 9.874 Ma, though the density of the record is low, with the next data points at 10.604 and 9.244 Ma. The parts with denser data show important fluctuations, but the lesser densely sampled times cannot show such fluctuations.
The onset of the Middle Miocene Climatic Optimum coincided with faunal turnover events in Africa and Europe (Van der Made, 2014). It also marks the onset of the radiation of the Listriodontinae in Africa and Europe (Figs. 8, 10). Later during this period, Lopholistriodon in Africa and Listriodon in the Indian Subcontinent evolved perfect lophodonty, became specialized folivores, and Listriodon acquired a Eurasian distribution from Portugal to China. The end of the Mid-Miocene Climatic Optimum, around 13.6 Ma, coincides with the major peak in faunal turnover in Africa between 16 and 10 Ma (Van der Made, 2014) and coincides with the end of several listriodont lineages (Figs. 8, 10). Apparently Listriodontinae did well during this period of high temperatures and high pCO2 as they reached their greatest geographic expansion and diversity (Fig. 10). The decrease in pCO2 from 13.6 Ma onwards is mirrored by the rise in Bovidae in Africa (Van der Made 2014, fig. 6) and elsewhere, and in the decrease in browsers noted by Janis et al. (2000, 2004). The final extinction of the Listriodontinae around 9.8–9.4 Ma and a spectacular drop in the abundance of the Tragulidae in the Indian Subcontinent (Barry 2014), where they were most diverse, coincide with one of the moments when the decrease in pCO2 accelerated (Figs. 8, 10). The pCO2 according to Van de Wal et al. (2011) shows several peaks with each time lower values from about 9.75 Ma onward and a sustained decrease in values from about 9.62 onward, reaching a record low at about 9.58 and continued to decrease till 9.539 Ma and peaked several times about 100, 200, and 300 ka later. The temporal duration of these events covers or overlaps the last records of the last listriodont lineages. The curves by Super et al. (2018) and Sosdian et al. (2018) show peak low values at 9.7 and 9.874, which are close, given the lesser temporal precision due to their less dense records.
The increase in sugar and reduction of protein, Zn, and Fe contents of fruit and vegetables is well studied in relation to the possible rise in pCO2 in the near future, starting from the present levels around 400 ppm. Little or nothing is known about the effect of the drop in pCO2, from over 450 ppm during the Middle Miocene Climatic Optimum till about 350 ppm around 9.8–9.4 Ma, on sugar, starch, protein, Zn, and Fe contents of leaves and fruit. The fact that during this period, the high diversity of browsers, including Tragulidae, declined markedly, suggests that a decrease of sugar and starch in fruit and leaves could have been prejudicial to them and that this was not compensated by the higher content in other nutrients.
Janis (1989) classified living and fossil Artiodactyla in three categories, those with: (1) little or no fermentation (Suidae, Tayassuidae, Entelodontidae and Palaeogene families); (2) some foregut fermentation (for example Tragulidae, Anthracotheriidae); and (3) full foregut fermentation plus rumination (including ruminants and camels). Category 2 is somewhat adapted to a fibrous diet, but category 3 consists of “cell wall specialists”, still better adapted to the digestion of cellulose. She noted that, during the Late Eocene and Oligocene, primitive foregut fermenters (type 2) diversified at the cost of hindgut fermenters and that from the Middle Miocene onward, ruminating artiodactyls (type 3) replaced less specialized foregut fermenters and the Listriodontinae. At present, Tayassuidae and Babyroussa are considered to be foregut fermenters, while Phacochoerus and Sus (both Suinae) and Potamochoerus are considered to be hindgut fermenters (Langer 1986; Leus et al. 1999; Clauss et al. 2008).
The living Suinae are hindgut fermenters, and the available data suggest that they have higher protein requirements than the peccaries and babirusas, which are foregut fermenters. Propotamochoerus from the Chinji Formation is the first representative of this subfamily and was probably a hindgut fermenter. This subfamily radiated and progressively replaced other Suoidea, most of which are likely to have been foregut fermenters (Palaeochoeridae, Listriodontinae, Cainochoerinae, perhaps Hyotheriinae). At the time when pCO2 decrease suggests an increase in protein content, Suinae, which were less efficient, replaced other Suoidea, which were more efficient in the digestion of proteins. At the same time, there was a decrease in sugar content and the hindgut fermenting Suinae handle sugars more efficiently. Either access to protein may have been a limiting factor or more efficient sugar digestion may have been a plus, or both may have been decisive in the spread of hindgut fermenting Suoidea. Listriodontinae may have been more competitive in a high pCO2 world, because they were more protein efficient, while they could afford to be less sugar efficient.
Though many data are still lacking, it seems possible that the decreasing pCO2 led to changes in the nutrient composition of the vegetation, favoring some herbivores and omnivores over others. The extinction of the Listriodontinae in the different continents may have occurred when a critical pCO2 threshold was met.