An Early Pleistocene hippopotamus from Westbury Cave, Somerset, England: support for a previously unrecognized temperate interval in the British Quaternary record

Although fossil assemblages from the late Early Pleistocene are very rare in Britain, the site of Westbury Cave in Somerset, England, has the potential to address this gap. The mammal fossils recovered previously from the Siliceous Member in Westbury Cave, though few in number, have hinted at an age for the deposits that is as yet unparalleled in Britain. Here, we describe the first bona fide occurrence of Hippopotamus in the British Early Pleistocene, discovered during recent reinvestigation of the Siliceous Member. The hippo fossil indicates a refined biochronological age of ca. 1.5–1.07 Ma for the Siliceous Member and a palaeoclimate that was warm and humid, which accords well with previous palaeoenvironmental inferences. A synthesis of late Early Pleistocene hippo occurrences suggests that the Siliceous Member hippo may have been part of an early colonization of north‐west Europe by these megaherbivores, possibly during MIS (Marine Oxygen Isotope Stage) 31. Alternatively, it evidences a currently cryptic northward migration during an even earlier temperate phase. In either case, the Siliceous Member is likely to represent a warm period that has not been recognized previously in the British Quaternary record.


Introduction
The late Early Pleistocene is a significant interval in the Quaternary period. It marks a major climatic shift in the periodicity of glacial/ interglacial cycles (the Mid-Pleistocene Transition, MPT; Head and Gibbard, 2015;Berends et al., 2021) and the fossil record from this time bears witness to major faunal turnovers in terrestrial ecosystems, especially among mammals (e.g. the Villafranchian-Galerian transition in European land mammal ages; Rook and Martínez-Navarro, 2010;Madurell-Malapeira et al., 2014;Bellucci et al., 2015). Moreover, the earliest known hominin occurrences in the Mediterranean basin and north of the Alps are of late Early Pleistocene age Arzarello et al., 2012;Toro-Moyano et al., 2013;Ashton et al., 2014).
Britain formed the north-western-most fringes of continental Europe during the late Early Pleistocene (Funnell, 1995) and, as such, it is a geographically important region for examining responses of the terrestrial biota to climate changes on the edge of Europe. Consequently, the almost complete lack of fossilbearing sites of this age from Britain presents a notable problem. A well-known gap occurs in the British fossil record from ca. 1.8 Ma to ca. 0.8 Ma (Gibbard et al., 1991;Preece and Parfitt, 2012), with only one fossil assemblage from Happisburgh 3 argued to date to the younger part of this interval ; but see Westaway, 2011). Very few sites known in Britain have potential to address this hiatus, but Westbury Cave in the Mendip Hills of Somerset, south-west England, has recently been identified as a candidate (Adams et al., 2019).
Westbury Cave is an exceptional site in the British Quaternary record. It is best known for its early Middle Pleistocene cave breccias (the Calcareous Member), which have yielded a spectacularly rich vertebrate assemblage (Bishop, 1982;. However, the stratigraphical and palaeontological importance of the underlying water-lain sediments (the Siliceous Member) has only recently been acknowledged (Adams, 2017;Adams et al., 2019). Reinvestigation of these sediments has provided strong support for their Early Pleistocene age, revealed the complexities of their internal stratigraphy and resulted in the development of a new depositional model (Adams et al., 2019). Here, we present the first description of Hippopotamus from recent excavations through the Siliceous Member. The fossil constitutes the earliest bona fide record of Hippopotamus in the UK and provides critical new biochronological and palaeoenvironmental information for the Early Pleistocene mammal assemblage in Westbury Cave. In combination with the few fossils reported previously from the Siliceous Member (Bishop, 1982;Gentry, 1999), the new record of Hippopotamus provides strong support for a late Early Pleistocene age for the deposits. In interpreting the significance of the find, we review the Pleistocene record of hippos in Britain and produce a new synthesis of dated hippo occurrences in the late Early Pleistocene of western Europe.
Broadmead Quarry, National Grid ref. ST 5081 5036; Fig. 1A). The stratigraphical context of the fossil is well constrained to the depositional unit F1U1 of Section 1c (S1c; Fig. 1B-D) described by Adams et al. (2019). This unit is laterally continuous with units F2U2 in S1c and F5U3 in S1a, which have also yielded an important assemblage of other large and small fossil mammals currently under study (see Adams, 2017). Together, these units were interpreted as distal talus cone deposits (part of Facies Association B in Adams et al., 2019).
There has been variable usage of species concepts for Pleistocene Hippopotamus in Europe. Both H. antiquus and H. major have been used since the 19th century to describe large Early and early Middle Pleistocene forms, with the later erection of H. incognitus for smaller Middle and Late Pleistocene hippos supposedly distinct from the extant hippopotamus, H. amphibius (Faure, 1984;Mazza, 1995). Although H. antiquus has since been identified as having priority over the species name H. major (Hooijer, 1942;Mazza, 1995), common usage of the latter made this synonymy contentious (Faure, 1985). Another species, H. tiberinus, was also proposed for some European specimens (Mazza, 1991), which has similarities to the African Early Pleistocene H. gorgops (van der Made et al., 2017a,b). In addition, there has been debate about whether Early and early Middle Pleistocene hippos exhibit wide variation within just one valid species, the extant H. amphibius, or whether they represent a process of phyletic gradualism that resulted in multiple species (Kahlke, 1997(Kahlke, , 2001. Accordingly, some European Pleistocene specimens have been assigned to subspecies of the extant hippopotamus, e.g. H. amphibius antiquus and H. amphibius incognitus (Kahlke, 1997). Synonymy is suspected among these species/subspecies (Mazza, 1995;Petronio, 1995), but few formal acts of systematic synonymization have occurred and those that have (e.g. Faure, 1985) are often disregarded in favour of other schemes. A broadly accepted model has now emerged with two species of Hippopotamus in the Pleistocene of Europe (e.g. Petronio, 1995;Kahlke et al., 2011;Martínez-Navarro et al., 2015;Pandolfi and Petronio, 2015;Konidaris et al., 2018;Kierdorf and Kahlke, 2020)  from ca. MIS 13 into the Late Pleistocene (Pandolfi and Petronio, 2015;Petronio et al., 2019;Martino and Pandolfi, in press). We follow this model, contrasting views notwithstanding (e.g. van der Made et al., 2017a,b), pending further systematic and phylogenetic studies.

Description
The tooth is slightly oblong in occlusal view (length-to-width ratio of 1.12) and is characterized by four cusps, each with a distinctive trefoil-shaped wear pattern. This pattern is caused by the development of strong styles to the anterior (mesial) and posterior (distal) of each cusp. The four cusps are the paracone (anterior labial), protocone (anterior lingual), metacone (posterior labial) and metaconule [posterior lingual; also called the hypocone by some authors, but Boisserie et al. (2010) argue a true hypocone is not present in hippopotamids]. The cusps are moderately worn, with the anterior pair slightly more so than the posterior. The protocone and, particularly, the paracone are inclined anteriorly. The transverse valleys between the anterior and posterior cusp pairs have outlets with widely tapering sides. This valley is broader between the paracone and metacone than it is between the protocone and metaconule. A strong cingulum is evident, particularly along the anterior margins of the paracone and protocone and the posterior margins of the metacone and metaconule. An entostylar pillar is only very weakly developed at the outlet of the lingual transverse valley. Although the tooth crown is near complete, a posterior part of the paracone and the ectostylar region are missing. Thus, it is not possible to assess the presence or strength of an ectostylar pillar at the labial transverse valley outlet. There are four roots, which are variably preserved but most complete beneath the metacone.

Remarks
The relative dimensions of occlusal length and width indicate an upper molar; lower molars tend to be narrower in occlusal view. The absolute dimensions suggest a position as a first rather than second upper molar. Comparisons with molar measurements of Hippopotamus antiquus from across western Europe by Faure (1985) and Mazza (1995)  In addition, Mazza (1995) suggests that M 2 s are often more trapezoidal in occlusal view, with anterior widths greater than posterior widths (5.2 mm greater on average; Fig. 3B,C); this is also true of M 3 s (6.1 mm greater on average; Mazza, 1995). By contrast, M 1 s tend to have anterior and posterior widths that are more equal. The difference in anterior and posterior width of the Siliceous Member specimen is only 1.51 mm, providing further evidence for a position as a first, rather than second, upper molar.
A position as a third molar can be ruled out on anatomical grounds. In H. antiquus, the style posterior to the metaconule (posterior hypostyle in Mazza, 1995) is often weakly developed on the M 3 , resulting in a comma-rather than trefoilshaped wear pattern on this cusp. The M 3 of H. antiquus also typically has a well-developed tubercle along the posterior cingulum (Mazza, 1995;distocone in Boisserie et al., 2010), which acts as a supplementary cusp and augments the occlusal surface. A similar feature is found on the M 3 , with the talonid bearing a prominent posterior tubercle (Mazza, 1995;hypoconulid in Boisserie et al., 2010). These characters are not found on the M 1 /M 2 , nor on the Siliceous Member specimen.
The description above accords well with features typical of H. antiquus in western Europe, as set out in comprehensive detail by Mazza (1995). However, it is difficult to distinguish between H. antiquus and H. amphibius based on isolated molars alone. More dependable differences are found in the cranial and postcranial skeletons (Caloi et al., 1980;Mazza, 1995;Pandolfi and Petronio, 2015). Mazza (1995) tentatively pointed out a number of differences in M 1 morphology between species, including subtly contrasting occlusal proportions, differences in the shapes of the transverse valley outlets, and the variable presence and strength of ecto-and entostylar pillars. Unfortunately, all of these are variable within each species and do not constitute demonstrable taxon-specific rules. Based on similarity in morphology and dimensions to specimens of H. antiquus, we assign the Siliceous Member specimen to H. cf. antiquus. It is relevant to note that, at present, H. antiquus is the only widely accepted species of hippopotamus in western Europe during the Early Pleistocene (see above). Therefore, given the age of the Siliceous Member, assignment to this species also provides the best fit to current taxonomic and biogeographical models.

Taphonomy and deposition of the Siliceous Member
Previous workers noted that the Siliceous Member faunal assemblage was dominated by rolled and heavily abraded teeth, which indicated considerable water transport (Bishop, 1982;Andrews and Ghaleb, 1999). This 'derived' assemblage was argued to be of little use for palaeoecological reconstructions, since the distance and time required to abrade the fossils so heavily would result in a probable conflation of taxa spanning a long time interval (Andrews and Ghaleb, 1999;Andrews and Stringer, 1999). While this is true, the provenance of finds from earlier investigations is unknown in relation to stratigraphy within the Siliceous Member, which was grouped as a single basal unit by Bishop (1982) and . As a result, the depositional context of those fossils is poorly constrained. The hippo molar described here differs quite markedly in its preservation to the very rounded large mammal (mostly bovid) molars observed by Andrews and Ghaleb (1999). Despite some limited breakage and minor rounding of broken roots (Fig. 2C,D), the hippo antiquus molar from the Siliceous Member, Westbury Cave, compared with first and second molars of H. antiquus (= H. major) from across western Europe (Faure, 1985;Mazza, 1995). molar is well preserved and clearly travelled a much shorter distance before entering Westbury Cave. A solution to this taphonomic contrariety was advanced by the sedimentological study of Adams et al. (2019). Several distinct gravel facies of variable preservation potential were identified within the Siliceous Member. The coarse-grained Facies Associations C and D were interpreted as stream or flood deposits, and the lack of non-durable limestone clasts within the gravels suggested considerable bedload abrasion and attrition (Adams et al., 2019). By contrast, the preservation of limestone clasts within Facies Association B gravels attested to the shorter transport distance of this material. Together with bed geometry and particle size data, this evidence was used to interpret Facies Association B as a set of distal talus cone deposits (Adams et al., 2019). While rounded fossils could be conceivably found in all of these facies associations, wellpreserved remains are only likely in the latter. The completeness of the hippo molar and its provenance in Facies Association B support this depositional model. Importantly, the recovery of well-preserved fossils from talus cone sediments highlights the fact that the Siliceous Member mammal assemblage is not highly derived in its entirety. The hippo molar is likely to be contemporaneous with talus cone deposition and similar finds will be of great value in refining Siliceous Member biochronology and palaeoecology.

Biochronology and palaeoenvironment of the Siliceous Member
Despite the suggestion that the Siliceous Member contains a derived faunal assemblage (Bishop, 1982;Andrews and Stringer, 1999), possibly spanning multiple time intervals and climatic episodes during the Early Pleistocene, there are no biochronological or palaeoenvironmental inconsistencies among the previously reported taxa to support this.
Several of these taxa are too poorly resolved taxonomically (Lynx sp. and Dama sp./Cervidae sp. indet.) to provide useful biochronological information (Bishop, 1982;Gentry, 1999). Other Siliceous Member mammals were identified to species [Castor fiber -Eurasian beaver, Stephanorhinus etruscus (= Dicerorhinus etruscus) -Etruscan rhinoceros, and Pachycrocuta brevirostris (= Hyaena brevirostris)giant short-faced hyaena; Bishop, 1982] but have long ranges that span the entire Early Pleistocene (Barisone et al., 2006;Pandolfi et al., 2017;Marciszak et al., 2021). The most biochronologically useful taxa from past work (Fig. 4C,D;Bishop, 1982;Gentry, 1999) are the bovid Leptobos, an archaic water vole, Mimomys savini, and Allophaiomysthe ancestor of modern grassland voles (Microtus spp.). Of these taxa, the identification of Leptobos is most tentative. Bishop (1982) assigned the Siliceous Member bovids to Bovinae cf. Bison schoetensacki, whereas Gentry (1999) favoured assignment to Leptobos. The fragmentary and rolled nature of the teeth, as well as the difficulty in distinguishing between dental remains of Leptobos and primitive Bison (e.g. B. degiulii, B. menneri, B. schoetensacki), has left the status of the bovids open to question (Mead et al., 2014;Cherin et al., 2019;Sorbelli et al., 2021). This taxonomic uncertainty has important implications, because Leptobos is limited to the Villafranchian land mammal age in Europe (ca. 3.5-1.0 Ma; Mead et al., 2014), which would preclude a post-Jaramillo age (Fig. 4D). In contrast, although rare early occurrences of Bison are known in the Late Villafranchian (Masini et al., 2013), this genus occurs mostly at younger sites, from the transitional Epivillafranchian age (ca. 1.2-0.9 Ma; Bellucci et al., 2015) and extending into the Middle Pleistocene (Breda et al., 2010;Petronio et al., 2019). Despite these uncertainties, it is clear from the small mammals that the Siliceous Member represents a period of the Early Pleistocene very rarely expressed in the British record.
Allophaiomys is a critical taxon in European small mammal biochronology during the mid-to late Early Pleistocene (e.g. Maul and Markova, 2007;Cuenca-Bescós et al., 2010), but its only reported British occurrence is in the Siliceous Member (Bishop, 1982). As identified by Gibbard et al. (1991), through comparisons with the more complete Dutch stratigraphy, a large gap exists in the British fossil record between the 'Pastonian' and 'Cromerian' stages (ca. 1.8-0.8 Ma). An important exception occurs at Happisburgh 3, where an assemblage correlated to MIS 25 or 21 was discovered (0.99-0.77 Ma; Parfitt et al., 2010; but see Westaway, 2011). However, this site lacks Allophaiomys and instead has more advanced Microtus species (M. arvalis and M. oeconomus). Allophaiomys, like Leptobos, is rare after the Jaramillo subchron in central Europe (Maul and Markova, 2007), hinting that the most likely age for the Siliceous Member is within the interval 1.77-1.07 Ma. As a result, the Siliceous Member is likely to represent an unknown period of British faunal history.
The new record of Hippopotamus cf. antiquus supports this hypothesis, because the species has not been reported in situ from any other British Early Pleistocene site. Furthermore, hippo remains are only widespread in Europe from ca. 1.5 Ma (Rook and Martínez-Navarro, 2010;Martínez-Navarro et al., 2015). As summarized by Martino and Pandolfi (in press), the oldest record of hippo in Europe is controversial and for a long time was thought to be from the Upper Valdarno Basin in Tuscany, Italy (ca. 1.8 Ma). This was until Napoleone et al. (2003) argued that the hippo specimens were likely to be younger than the main Upper Valdarno faunal assemblage, partly due to their poorly resolved provenance and their similarities to hippo fossils from younger Italian sites (e.g. Colle Curti). Conversely, Martino and Pandolfi (in press) argue that it is difficult to deny the provenance of hippo fossils from the Upper Valdarno and point to several hippo-bearing localities within the Montevarchi Synthem of the Upper Valdarno Basin (dated to ca. 1.9-1.7 Ma; Fidolini et al., 2013). Even older Italian occurrences (ca. 2.2 Ma) have been proposed from Coste San Giacomo and from the Chiusi Basin (Bellucci et al., 2012(Bellucci et al., , 2014Pandolfi and Petronio, 2015), but doubt about these has been expressed because they were ex situ finds without stratigraphical context (Martínez-Navarro et al., 2015;Pandolfi and Petronio, 2015;Marra et al., 2018;Martino and Pandolfi, in press). Marra et al. (2018) suggest that the hippo fossils from Coste San Giacomo could have their provenance in younger deposits overlying the main fossiliferous strata, and it is possible that the surface-collected assemblage from the Chiusi Basin contains fossils of multiple ages (Pandolfi and Petronio, 2015). However, proponents of an older age estimate point to similarly old (Middle Villafranchian) records of Hippopotamus from Greece (Reimann and Strauch, 2008). Such old records have also been postulated to represent early colonizations of the Italian peninsula, without further northward dispersal, followed by local extinction before a later, more successful wave of colonization from Africa (van der Made et al., 2017a). Although not impossible, it is unlikely that hippos migrated to Britain before 1.5 Ma based on available European occurrences (see below). In combination with the evidence above, the new hippo find supports a refined age of ca. 1.5-1.07 Ma for this part of the Westbury Cave sequence.
The mammal taxa currently known from the Siliceous Member also provide a coherent reconstruction of the palaeoenvironment in south-west England during the late Early Pleistocene. Bishop (1982) summarized the palaeoecology of the 'Bed 1' (Siliceous Member) assemblage as an open woodland fauna of temperate climate, where bovids were the dominant faunal element. The presence of beaver and archaic water vole also attested to the presence of permanent surface water sources in the area (Bishop, 1982). These conclusions are clearly supported by the new record of hippopotamus. Hippos are generally regarded as thermophilous, water-dependent taxa and indicative of humid conditions, mild winters and average annual temperatures several degrees higher than present (Candy et al., , 2010; but see below).
While the ecology of H. antiquus is considered largely similar to that of its extant relative, the anatomy and diet of the extinct form suggest even greater dependence on aquatic environments. Isotopic dietary reconstructions show that H. antiquus fed largely on aquatic plants rather than the terrestrial grasses favoured by extant hippos (Palmqvist et al., 2003(Palmqvist et al., , 2008. Palmqvist et al. (2008) also propose that the anatomy of H. antiquus was more poorly adapted than that of H. amphibius to terrestrial locomotion, with a much larger body mass and shortened metapodials, while several other features were better adapted to an aquatic lifestyle (e.g. more elevated orbits). Therefore, H. antiquus is probably an even more reliable indicator of humid conditions than the extant species. The probable habitat preferences of the rest of the Siliceous Member mammals are compatible with temperate conditions. They indicate a mixed environment of savanna-like grassland for grazers (bovids, ancestral grassland voles) and scavenging carnivores (hyaena), with nearby areas of open woodland or forest for browsers/mixed feeders (bovids, deer, rhinoceros), forest hunters (lynx) and species that favour wooded habitats (beaver). Hippos add the presence of a sizeable river or lake to this landscape.

The Pleistocene record of hippo in Britain
Hippopotamus is well known from several temperate periods in the British Pleistocene: the early Middle Pleistocene interglacials of the 'Cromerian Complex' (MIS 17-13) and the last interglacial (MIS 5e or the Ipswichian) in the Late Pleistocene.
Previously, the oldest records of Hippopotamus in Britain came from the East Anglian Crag Basin, where hippo remains have been recorded from numerous sites. These include East Runton, Cromer, Overstrand, Sidestrand, Trimingham, Mundesley, Bacton, Happisburgh, Norton Subcourse, Pakefield, Kessingland and Walton-on-the-Naze (Owen, 1846;Newton, 1882;Reynolds, 1922;Sutcliffe, 1959;Stuart, 1974Stuart, , 1982Stuart, , 1986Stuart and Lister, 2001;Lewis et al., 2004). One interesting exception is West Runton (Stuart, 1974), which lacks hippo despite a history of extensive fossil collecting. Although many hippo fossils are known from this region, very few have secure stratigraphical context. Most occurrences represent isolated bones or teeth, or fragments thereof, found loose on the beach, having eroded out of the cliffs or foreshore (Stuart, 1996).
Much of the coastal land in this area is made up of Early and early Middle Pleistocene sediments belonging to the Crag Group (Norwich Crag and Wroxham Crag Formations) and the Cromer Forest-bed Formation Rose, 2009). These deposits often occur in direct superposition and fossils are known from horizons of both ages, resulting in site collections of mixed ages (Lister, 1996(Lister, , 1998. Only hippo fossils from Pakefield/Kessingland (Stuart and Lister, 2001) and Norton Subcourse (Lewis et al., 2004) can be unambiguously linked to specific horizons within a site stratigraphy. Both sites date to the early Middle Pleistocene 'Cromerian Complex' (Fig. 4A; Preece and Parfitt, 2012), and the normally magnetized sediments from both sites suggest an age in the Brunhes magnetochron (<0.773 Ma; Channell et al., 2020). However, the precise age of the hippo-bearing strata within the Cromer Forest-bed Formation at Pakefield and Norton Subcourse is contentious and open to different interpretations, which have included MIS 15, MIS 17 and late MIS 19 (e.g. Parfitt et al., 2005;Lee et al., 2006;Westaway, 2011;Preece and Parfitt, 2012). MIS 17/15 may be more likely since much of the MIS 19 interglacial, including its peak (MIS 19c), occurred during a period of reversed magnetization in the very latest part of the Matuyama magnetochron (Channell et al., 2020) that is likely to be unrepresented in the 'Cromerian' deposits of East Anglia (Candy et al., 2015). Whatever their precise age within the early Middle Pleistocene, the Pakefield and Norton Subcourse hippo fossils were hitherto the oldest well-provenanced records of hippo in the British Pleistocene.
Despite the occurrence of mammal-bearing Early Pleistocene sediments in the Crag Basin, no hippo remains have been demonstrated to have their provenance in these deposits. In addition, no other British Early Pleistocene site beyond the Crag Basin has records of hippo, e.g. Dove Holes Cave in Derbyshire (Spencer and Melville, 1974) and the Dewlish bone fissure in Dorset (Fisher, 1905;Carreck, 1955). Therefore, the Siliceous Member hippo represents the first bona fide occurrence of the genus and family in the British Early Pleistocene.
Nevertheless, some Crag Basin hippo specimens are more likely to represent Early Pleistocene occurrences than others. A careful analysis of the elephant and deer remains by Lister (1996) showed that some sites have mostly or entirely early Middle Pleistocene species (West Runton, Trimingham, Pakefield, Kessingland), while others are dominated by Early Pleistocene species (East Runton) or have mixed assemblages (Overstrand, Sidestrand, Mundesley, Bacton, Happisburgh). Therefore, the occurrence of hippo at East Runton is particularly noteworthy. While Lister (1996) argued that the small number of early Middle Pleistocene species at East Runton may derive from easterly exposures of the West Runton Freshwater Bed, or may have been transported to the locality by tidal processes, the fact that no hippo fossil has ever been recorded from West Runton renders this quite unlikely for the East Runton hippo specimens. The dominance of Early Pleistocene large mammals at East Runton suggests that hippo specimens from this locality are the most likely to be Early Pleistocene in age out of all Crag Basin occurrences. Stuart (1986) suggested that most, if not all, Crag Basin hippo fossils were likely to be of early Middle Pleistocene 'Cromerian' age, because 40 of 87 Crag Basin hippo specimens from the NHMUK Savin Collection were recorded from cliff bases. He argued that, in general, 'Cromerian' deposits were found at the cliff bases in the area, while older deposits occurred on the foreshore. Nevertheless, 13 of 87 specimens were found on the foreshore, where Early Pleistocene sediments are exposed, and Early Pleistocene deposits are also found in the cliffs. The foreshore hippo specimens, particularly if from East Runton, could thus represent genuine Early Pleistocene occurrences, but this is currently not possible to demonstrate.
Although pollen analysis of sediment adherent to museum hippo specimens from Trimingham, Mundesley and Bacton was argued to support a 'Cromerian' age (Stuart, 1986), Stuart (1996) admitted that the later recognition of multiple interglacials within the 'Cromerian Complex' prevented direct correlation to a single temperate episode. In addition, the high frequency of Abies pollen was used to rule out a 'Pastonian' age, since Abies was unknown from the British Early Pleistocene at the time (Stuart, 1986). The Abies fossil record has been extended into the late Early Pleistocene in Britain recently (Farjon et al., 2020) and, as a result, an abundance of Abies pollen no longer precludes an Early Pleistocene age.
A remarkable exception to the isolated and fragmentary hippo fossils from the Crag Basin is a well-preserved mandible of H. antiquus (NWHCM 1845.30.1) that was recovered from 'freshwater clay-beds' near Cromer (Owen, 1846, p. 399;Fig. 4B). Unfortunately, its precise provenance and age are poorly known (Stuart, 1986). The most likely case is that the mandible is of 'Cromerian' early Middle Pleistocene age. However, if these beds equate to the clay-rich strata that overlie the Weybourne Crag near Cromer and yield Early Pleistocene large mammals, this specimen may be of mid-Early Pleistocene/'Pastonian' age (ca. 1.8-1.7 Ma; see Lister, 1998). Although late Early Pleistocene deposits with fossils are very rare between the 'Pastonian' and 'Cromerian' in the UK (Gibbard et al., 1991), they are likely to exist locally in the Crag Basin (Lister, 1998;Lee et al., 2018). One such example is the site of Happisburgh 3 Preece and Parfitt, 2012;Ashton et al., 2014;but see Westaway, 2011). The hippo mandible from Cromer, too, could be of intermediate age. It is important to note that no hippo remains have been recorded from the vertebrate assemblage at Happisburgh 3 . Palaeoclimate reconstructions by Parfitt et al. (2010) and Farjon et al. (2020) indicate that the site represents a cool period towards the end of an interglacial, with winter temperatures possibly too low for hippos to tolerate. Given that Hippopotamus was not widespread in Europe until ca. 1.5 Ma, a 'Pastonian' occurrence in Britain at ca. 1.8-1.7 Ma would be extraordinarily early. However unlikely, the recent suggestions that hippos were on the Italian peninsula ca. 2.2 Ma (Bellucci et al., 2012(Bellucci et al., , 2014 and ca. 1.9-1.7 Ma (Martino and Pandolfi, in press) could be used to hypothesize earlier northward migrations than are known. Few methods exist that can directly and reliably date Early Pleistocene fossils (e.g. Duval et al., 2012) and that could resolve these uncertainties. The problem is exacerbated when there is no stratigraphical context for the remains, but recent advances show promise (e.g. Dickinson et al., 2019). Continued development of these methods would be of great value for specimens of great antiquity but poor provenance, such as the hippo fossils from the Crag Basin.
Beyond East Anglia, Preece and Parfitt (2012, their Parfitt (1999) and Roberts and Parfitt (1999). However, this appears to be in error, as neither of these publications record hippo at the site. Other sites correlated to MIS 13 in the UK include Sidestrand, Ostend, Brooksby, Waverley Wood, High Lodge, Happisburgh 1 and the Calcareous Member of Westbury Cave (Preece and Parfitt, 2012;Candy et al., 2015). Of these, hippo is only known from Sidestrand (Preece et al., 2009;Candy et al., 2015). Although this occurrence is dubious because it was not recorded in situ and the specimens could have come from Early or early Middle Pleistocene deposits (Preece et al., 2009), the large mammals from Sidestrand are mostly early Middle Pleistocene forms (72% of elephants and 78% of deer; Lister, 1996).
During the late Middle Pleistocene (MIS 11-7), Hippopotamus is thought to be largely absent from Britain (Schreve, 2001). Only during the last interglacial (Ipswichian; MIS 5e) in the Late Pleistocene are hippo remains widespread and known from over 35 sites across the country (Sutcliffe, 1959;Stuart, 1986;Currant and Jacobi, 2001). Although hippo has been described in Gloucester from what was thought to be an MIS 7/6 context, an MIS 5 age for this occurrence is considered more likely (Schreve, 2009). Despite the supposed absence of hippo during the late Middle Pleistocene, Schreve (1997) also drew attention to a hippo incisor fragment in the NHMUK collection (NHMUK PV OR 21653) from the MIS 9 interglacial site at Grays Thurrock, Essex (Schreve, 2001). This specimen was acquired at the same time as the rest of the Grays assemblage and is of a similar preservation type (Schreve, 1997), which led to the suggestion that the hippo occurrence could be genuine and that hippo presence during MIS 9 should not be dismissed out of hand (Schreve, 1997).

Palaeoclimates during hippo occupation of the British Isles
Given the stratigraphical distribution outlined above, it is possible to make some observations about the climatic conditions under which Hippopotamus existed when they comprised part of the British fauna. Sites where Hippopotamus remains occur in situ with robust palaeoclimatic reconstructions are restricted to a number of MIS 5e sites, including Trafalgar Square, which is discussed here as it contains the richest and most detailed palaeoclimatic record (see Candy et al., 2016 for a review), and Pakefield (Parfitt et al., 2005;Candy et al., 2010). Furthermore, it appears probable, but not undeniable, that the Hippopotamus remains from Sidestrand are associated with the early Middle Pleistocene 'Cromerian' deposits at this site. Consequently, it is possible to discuss the climate and environments of Hippopotamus migration into southern Britain during MIS 5e (Trafalgar Square), MIS 13 (Sidestrand) and an earlier 'Cromerian' interglacial (Pakefield, MIS 17/15). Significantly, these sites yield the warmest interglacial temperature reconstructions of any British interglacial deposits. Coleopteran assemblages at all three of these sites contain southern and eastern European species indicating summer temperatures several degrees higher than the present day. This suggestion is supported by the presence, again at all three sites, of Trapa natans (the water chestnut), which is often reported as requiring mean July temperatures of >20°C to successfully germinate (Candy et al., 2010(Candy et al., , 2015. All three sites contain a rich assemblage of thermophilous flora and fauna (see discussion in Candy et al., 2010Candy et al., , 2015Candy et al., , 2016 and, therefore, produce the most robust record of enhancedin this case taken to mean greater than Holocene (see Candy et al., 2010Candy et al., , 2016)interglacial warmth of any interglacial deposits of the British Middle and Late Pleistocene. It is during these periods of enhanced warmth that Hippopotamus remains are found Britain.
Given the current biogeographical distribution of Hippopotamus, its presence in Britain during these particular interglacials would appear logical. However, it has to be stressed that human agency is the most likely primary control on the modern distribution of this taxon (Stoffel et al., 2015;Lewison and Pluháček, 2017), meaning that there is no way of reliably quantifying its fundamental environmental range or climatic tolerances. Certainly, for much of Quaternary history Hippopotamus appears to have been a pan-Mediterranean, as well as African, genus, extending into central and northern Europe as well as the Arabian peninsula during particularly wet or warm intervals (e.g. van der Made et al., 2017b;Stewart et al., 2019). In their review of thermophilous flora and fauna within the British Isles, Candy et al. (2010) suggested that the availability of permanent, open waterbodies was likely to be the main control on Hippopotamus distribution. As a result, the presence of hippo fossils required the occurrence of humid conditions with mild winters (to prevent seasonal freezing over of rivers and lakes). None of these environmental tolerances requires the occurrence of exceptionally warm interglacials; nonetheless, this taxon is currently only known from such intervals in Britain. Consequently, it is reasonable to assume that the warm interval indicated by the Hippopotamus remains in the Siliceous Member at Westbury Cave represents the occurrence of a humid interglacial characterized by mild winters and enhanced (greater than Holocene) warmth.
It is here worth emphasizing the observations of Candy et al. (2010) and Candy and McClymont (2013) that the evidence for enhanced warmth in multiple British early Middle Pleistocene sites, and in some North Atlantic marine cores, is in strong contrast to many long palaeoclimate records that are routinely used to discuss Quaternary climate history. In records such as LR04 ( Fig. 4A; Lisiecki and Raymo, 2005) and EPICA Dome C (Jouzel et al., 2007), interglacials MIS 19-13 (the interval within which both Sidestrand and Pakefield are positioned) appear significantly cooler than the Holocene and most late Middle and Late Pleistocene interglacials. It is therefore crucial to use climate records from the region within which fossils occur to inform palaeoenvironmental interpretations, rather than climate data that are distant from the study site. This is particularly important when discussing the Early Pleistocene of Britain, as very few detailed palaeoclimatic reconstructions are available for this region at this time. It is consequently tempting to rely on long records, such as LR04, to identify periods of enhanced warmth that may permit the migration of Hippopotamus into Britain. However, such an approach is highly problematic because two of the key sites that yield both Hippopotamus fossils and evidence for enhanced warmth appear relatively 'cool' in several long climate records.

The late Early Pleistocene hippos of western Europe
The temporal (Fig. 4E) and spatial (Fig. 5) Fig. 4). The broad patterns identified are likely to be representative, but could be tested in future work.
As already mentioned, some early occurrences of hippo at ca. 2.2 Ma may represent the first colonization of Europe (Bellucci et al., 2012(Bellucci et al., , 2014, but this is not universally accepted (Martínez-Navarro et al., 2015;Marra et al., 2018;Martino and Pandolfi, in press). Between ca. 1.9 and 1.5 Ma, hippo presence has been suggested from Italian sites in the Upper Valdarno, Mugello and Ellera basins (Martino and Pandolfi, in press), but these include controversial, unstratified and poorly dated records, and hippo has not been recorded in western Europe beyond the Italian peninsula at this time. More widespread occurrences are known after ca. 1.5 Ma ( Fig. 4E; Rook and Martínez-Navarro, 2010;Martínez-Navarro et al., 2015). Between ca. 1.5 and 1.3 Ma, hippo is known from the Iberian peninsula (four sites: Venta Micena, Incarcal-I, Barranco León and Fuente Nueva-3), with a more northerly occurrence at Sainzelles in the Massif Central of France and a possible occurrence at Saticula, Italy (Fig. 5), but age uncertainties are wide at this site. Hippo does not then occur at numerous sites until ca. 1.1-1.0 Ma, around the time of the Jaramillo subchron (a period of 'normal' polarity of the Earth's magnetic field; Fig. 4E). It is during this period that hippo spreads north of the Alps/Massif Central and into north-west Europe (Germany and more northern parts of France). Hippo fossils from Courterolles in France could mark an earlier presence north of the Massif Central, but wide age estimates for the site (ca. 1.3-0.9 Ma; Brochet et al., 1983) also include the Jaramillo interval. The Early Pleistocene hippo from Westbury Cave is one of the most northern/north-western occurrences known so far in Europe, only rivalled by specimens from the Netherlands, where hippo remains are associated with Early Pleistocene assemblages dredged from the bottom of the North Sea (van Kolfschoten, 2001;Mol et al., 2003). These Dutch records have poor age constraints, but ranges also overlap with the ca. 1-Ma/Jaramillo occurrences in north-west Europe (Fig. 4E). Several occurrences are known after the Jaramillo subchron, but these appear largely restricted to southern ranges near the Mediterranean coast in Spain, France and Italy, with the exception of the Dutch records if younger ages within their uncertainties are upheld.
Although the fossil record is notoriously incomplete and provides only a partial picture of animal palaeobiogeography, it can be used to construct testable hypotheses, upon which future data can be brought to bear. Available occurrences suggest that late Early Pleistocene hippos were widespread north of the Alps/Massif Central around the time of the Jaramillo subchron. The best available palaeomagnetic and biochronological information, summarized above, suggests an age of ca. 1.5-1.07 Ma for the Siliceous Member. Given these data, it is plausible that the Siliceous Member hippo was broadly coincident with other northern occurrences and represents part of a colonization of north-west Europe that took place towards the younger end of the Siliceous Member biochronological age estimate.
A younger age within the biochronological inference is also supported by the form of the Allophaiomys molar from the Siliceous Member (Bishop, 1982;Fig. 4C). The confluence between the anterior dentine fields (the fourth and fifth triangles of the occlusal surface and the anterior cap) is consistent with the 'allophaiomyid' molar morphotype and, although the posterior lobe is incomplete, it is possible to provide an estimate of ca. 48 for the A/L ratio (length of the anteroconid complex relative to maximum molar length; see van der Meulen and Zagwijn, 1974). An 'allophaiomyid' morphology and a high A/L ratio (>44.5) are features shared with 'advanced' Allophaiomys (sensu Maul and Markova, 2007), which occurs in central Europe ca. 1.2-1.0 Ma.
At present, most Jaramillo hippo occurrences in Europe cannot be constrained to a particular interglacial within the subchron due to the broad age uncertainties at most sites. However, the renowned site of Untermassfeld in Germany has been linked to the MIS 31 interglacial (ca. 1.08-1.06 Ma; Channell et al., 2020) through a combination of biochronology, magnetostratigraphy, lithostratigraphy and radiometric dating (see review by Kahlke, 2006;Gerdes et al., 2020) and is rich in Hippopotamus fossils, with over 800 individual hippo specimens (Kierdorf and Kahlke, 2020). Proxy records around the world indicate that MIS 31 was warmer than the current interglacial (Oliveira et al., 2017). This is confirmed at Untermassfeld by the discovery of several thermophilous vertebrates (including climate-sensitive reptiles, e.g. sand boa, snake-eyed lizard and a disputed occurrence of pond turtle; Böhme, 2020) as well as local isotopic palaeotemperature reconstructions higher than present (Stephan et al., 2001;Kahlke, 2006). MIS 31 was clearly warm and humid enough to support hippo populations in north-west Europe and it is likely, but admittedly not yet demonstrable, that conditions were similarly favourable in Britain. In addition, similarities between the German and British records of Hippopotamus in the Middle and Late Pleistocene have been highlighted previously, with preferential migration into Britain potentially occurring via the Rhine River valley (Schreve et al., 2007;Schreve, 2012;Fig. 5).
Given that hippos were present in Germany, France and possibly also the Netherlands during the Jaramillo subchron, it is not at all inconceivable that hippos migrated to the British Isles at a similar time. MIS 31 is not the only interglacial during the Jaramillo subchron (e.g. MIS 29, ca. 1.03-1.01 Ma), but it is the only one that includes periods of both reversed and normal magnetization. MIS 31 straddles the lower Jaramillo boundary, beginning during a period of reversed magnetization around 1.09-1.08 Ma and terminating at ca. 1.06 Ma within the normal subchron (Oliveira et al., 2017;Channell et al., 2020). It is also the only Jaramillo interglacial with well-constrained, site-specific evidence in nearby Europe for enhanced (greater than Holocene) warmth; only under these palaeoclimatic conditions are hippos currently known in Britain during the Pleistocene (see above). A post-Jaramillo age for the Siliceous Member is unlikely due to the occurrence of Allophaiomys (and possibly also Leptobos), and an age within the Jaramillo subchron itself is precluded by the reversed polarity of the sediments, so an assignment of the Siliceous Member to early MIS 31 (of reversed magnetization) appears to be consistent with Jaramillo hippo presence in north-west Europe and is a leading hypothesis that warrants further investigation.
While the early part of MIS 31 provides a good fit to available biogeographical models, biochronological and palaeomagnetic constraints, and palaeoclimate data, it is certainly not the only possibility. The Siliceous Member occurrence could also mark a currently cryptic migration of Hippopotamus into north-west Europe during an earlier temperate and humid phase in the late Early Pleistocene.

Conclusions
The hippopotamus molar described from the Siliceous Member represents the earliest occurrence of the genus and family in the British Pleistocene. Similarities in morphology and size to specimens from across western Europe suggest that the molar belongs to Hippopotamus cf. antiquus. The preservation of the fossil is consistent with its provenance in distal talus cone deposits and demonstrates that the Early Pleistocene faunal assemblage in Westbury Cave is not entirely highly derived, as previously believed. In combination with other palaeomagnetic and biochronological dating constraints, the Siliceous Member hippo supports a depositional age between ca. 1.5 and 1.07 Ma, a period unknown in the British fossil record before now. As a key thermophilous taxon, the presence of hippo bolsters the palaeoenvironmental inferences made from previously reported mammal taxa. It is likely that the Siliceous Member records an interglacial period, when temperatures were a few degrees higher than present, winters were mild and palaeoclimates were humid. A synthesis of dated hippo occurrences across western Europe suggests that the Siliceous Member hippo may have been part of a widespread colonization of north-west Europe that occurred from ca. 1.1 Ma, possibly during MIS 31. The Early Pleistocene sediments in Westbury Cave could record the early part of this interglacial, or could record an earlier temperate phase that would imply currently cryptic migrations of hippo into north-west Europe during the Early Pleistocene. In either case, the Siliceous Member is likely to represent a warm period in the Early Pleistocene that has not been recognized previously in the British Quaternary record.
Acknowledgements. The Palaeontological Association is gratefully acknowledged for funding, through the Callomon Award (grant ref. PA-CA201501), which enabled fieldwork in 2016 and led to the discovery of the hippo fossil. Alford Technologies Ltd and Natural England kindly granted access to land and permission to work at the site. Pierre Schreve is thanked for assistance in the field and with fossil conservation. Thanks also to David Waterhouse for providing specimen information on the Cromer hippo mandible, currently on permanent display at Cromer Museum, and to Amal Khreisheh for curatorial assistance at TTNCM. N.F.A. was supported by an NERC doctoral studentship awarded through the Central England NERC Training Alliance (CENTA; grant ref. NE/L002493/1) and by the University of Leicester during this project. This paper is dedicated to Nigel Taylor and the late Sidney Alford for their enthusiastic support of our recent work on the cave deposits in Westbury Quarry.
Conflict of interest statement-The authors declare no conflicts of interest.

Data Availability Statement
All data that support the findings are available in the main text or in referenced literature. The fossil specimen has been accessioned into a recognized public collection, where it is available for study.