Evolution Icon Evolution
Paleontology Icon Paleontology

Fossil Friday: Fossil Elephant Shrews and the Abrupt Origin of Macroscelidea

Günter Bechly
Photo: Namasengi mockeae, mandible, Eocene, Namibia, from fig. 11 in Senut & Pickford 2021, fair use.

This Fossil Friday we look into the origins of the placental mammal order Macroscelidea, which comprises the very strange but uber-cute elephant shrews or sengis. These small insectivorous mammals are speedy runners endemic to Africa, where they occur in 20 living species (Heritage et al. 2020). With their long mobile snouts they look a bit like the fictitious snouter animals described by the German zoologist Gerolf Steiner in his satirical book Bau und Leben der Rhinogradentia (Stümpke 1967), which has a kind of cult status among German-speaking animal lovers. Elephant shrews are sometimes considered to be living fossils (Novacek 1984) and their origin is believed to go back 57.5 million years in the Paleocene with a subsequent divergence of the crown group in the early Oligocene about 33 million years ago (Heritage et al. 2020: fig. 7) and its main diversification in the Miocene correlated with aridification events in Africa (Douady et al. 2003). Such a deep division of the main groups of extant Macroscelidea is also indicated by another recent molecular study (Krásová et al. 2021). So much for the theory, but let’s check what the actual fossil record, mostly consisting of jaw fragments and isolated teeth, really says. As noted by Stevens et al. (2021), “unravelling the origin and affinities of macroscelideans (also termed sengis, or elephant shrews) has long presented a palaeontological puzzle.” According to the technical literature the fossil record of the order Macroscelidea and its assumed fossil relatives spans from the Early Eocene to the Pleistocene (Patterson 1965Butler 1995Holroyd 2009Asher & Seifert 2010Naish 2013), and was always restricted to the African continent (for a possible exception see below). The earliest fossils of Macroscelidea are said to be 45.6 million years old (Hartenberger 1986Holroyd 2010Seiffert 2010a).

Considerable Scientific Debate

Of course, the question of the earliest fossil record of Macroscelidea hinges crucially on the question of which fossil taxa qualify as Macroscelidea in the first place. Unfortunately, this latter point is a matter of considerable scientific debate with very different suggestions for the composition, subdivision, and classification of Macroscelidea. We here follow the most recent work by Senut & Pickford (2021), but include in our discussion some more questionable fossil taxa considered by Holroyd (2010) and Hooker & Russell (2012) (see the Appendix for a classification of all genera with age ranges of their fossil record).

The oldest unequivocal Macroscelidea are the fossil taxa Eorhynchocyon rupestris (Rhynchocyonidae), Namasengi mockeae (Macroscelididae: Namasenginae), Promyohyrax namibiensis (Myohyracidae), Afrohypselodontus minus and A. grandis (Afrohypselodontidae), all described by Senut & Pickford (2021) from the tufas of the Eocliff locality in Namibia. They demonstrate the simultaneous abrupt appearance of all four families of elephant shrews in the Middle Eocene (Bartonian/Priabonian, 40.4-37.2 mya), which is not exactly what Darwinism would predict. A somewhat older Middle Eocene record could be a single tooth (specimen SN 10’08 ) of an unnamed species from the Sperrgebiet in Namibia described by Pickford et al. (2008: 470), which is of probable Lutetian age (47.8-41.2 mya) and may belong to an early ancestor of Myohyracidae.

Previously, the oldest crown group macroscelideans were the rhychocyonids Oligorhynchocyon songwensis from the Late Oligocene (25.2 mya) Nsungwe Formation in Tanzania (Stevens et al. 2021) and Miorhynchocyon meswae from the Early Miocene (22-21 mya) of Kenya (Butler 1984Holroyd 2010Heritage et al. 2020).

The Pliocene genus Mylomygale was a rodent-like herbivore (Naish 2013) and is classified in a separate subfamily Mylomygalinae (Camp et al. 1953Patterson 1965Holroyd 2010). However, “Corbet & Hanks (1968) pointed out that it shows a considerable resemblance to Macroscelides, and it may be a specialized member of the Macroscelidinae” (Butler 1995). Until recently, Mylomygale was only known from Early Pleistocene cave sediments in South Africa, contemporaneous with the close-by locality of the famous Taung Child. Now, this genus has also been found in Middle Pliocene paleokarst of South Africa (Senut et al. 2022).

Another group that is very close to modern macroscelideans but rather resembled hyraxes is the extinct family Myohyracidae (Patterson 1965Van Valen 1967Butler 19841995, McKenna & Bell 1997, Naish 2013), which initially indeed was misidentified as fossil hyraxes. It is mainly known from the Miocene, but the earliest representatives are Myohyrax and Promyohyrax from the Late Eocene of Namibia (Senut & Pickford 2021). The above mentioned unnamed specimen from the Lutetian of Namibia might be a basal Myohyracidae, since Pickford et al. (2008) found that it has “a morphology that could represent the ancestral morphotype for Myohyrax and Protypotheroides.

Finally, there is the new macroscelidean family Afrohypselodontidae erected by Senut & Pickford (2021) from the Late Eocene of Namibia, which is unique with its ever-growing rootless cheek teeth.

Two other extinct taxa that were excluded from unequivocal Macroscelidea by Senut & Pickford (2021) but are quite likely indeed stem macroscelideans are the Metoldobotidae and Herodotiidae. The later could extend the fossil record to 55.8 million years ago, which would perfectly align with the brief Paleogene window of time, when most of the placental orders abruptly appear in the fossil record.

The genus Metoldobotes from the early Oligocene of Fayum in Egypt was often considered the oldest and largest unequivocal Macroscelidae (Patterson 1965Simons et al. 1991Asher & Seiffert 2010Stevens et al. 2021). Strangely, it would also be “one of the most derived macroscelideans” (Simons et al. 1991). Tabuce et al. (2001) suggested that Metoldobotes is the sister group to Myrohyracidae and living elephant shrews, while Seiffert (2007) found strong bootstrap support for a sister group relationship with crown group Macroscelidea. However, in the most recent study this position has been questioned by Senut & Pickford (2021), who considered Metoldobotidae to be only doubtfully related to Macroscelidea, mainly because of its poorly preserved and scanty fossil record. As an interesting side note, Naish (2013) speculated that Metoldobotes may have been the only venomous sengi because of a conspicuous groove in the lower third incisor, but also cautioned that many mammals have such grooves without being venomous.

The extinct family Herodotiidae is more primitive and older, with some similarities to the Paleogene “condylarthrans” (Butler 1995; see below), but still has often been included as subfamily Herodotiinae in a more widely understood family Macroscelidae (Simons et al. 1991Butler 1995Tabuce et al. 20012012Benoit et al. 2013Tabuce 2018). The studies by Seiffert (2007) and Asher & Seiffert (2010) “reviewed arguments for and against including herodotiines (Nementchatherium, Chambius and Herodotius) in the Macroscelidea.” According to Seiffert (2007) “the Eocene herodotiines Chambius and Herodotius are generally considered to be primitive macroscelideans” (e.g., Hartenberger 1986Tabuce et al. 2007Seiffert 2010b), with a fossil record that extends to the Early or early Middle Eocene Chambi Massif in Tunisia, where Chambius was first discovered. However, in Seiffert’s own analysis (Seiffert 2003, 2007) the genera Herodotius and Chambius never clustered with Macroscelidea but with aardvarks in some of the trees, and in others with pseudoungulates, paenungulates, or hyraxes. Senut (2008) concluded that “the Palaeogene Macroscelididae from Northern Africa classified in the Herodotiinae (Simons et al., 1991), might not be ancestral to any of the Miocene to Recent Macroscelididae.” However, this result could be based on undersampling of data as a larger updated data set (Seiffert 2010b: fig. 14) again placed herodotiines with Macroscelidea. Benoit et al. (2013) studied the evolution of the middle and inner ear of elephant shrews and in their cladistic study resolved Chambius as basal-most representative, but they ignored the study of Seiffert (2007). I think the case for a macroscelidean relationship is reasonably well supported, which would extend the fossil record of Macroscelidea to about 55.8 million years ago.

Uncertain and Doubtful

All remaining candidates as earlier stem macroscelideans are much more uncertain and doubtful. They mostly belong to a group of enigmatic fossil mammals from the Paleocene and Eocene that was previously classified within the highly polyphyletic ungulate-like order “Condylarthra” in the family Hyopsodontidae, which seems to be a likewise polyphyletic waste basket taxon (Tabuce et al. 20012006Zack et al. 2005aHalliday et al. 2015). The study by Halliday et al. (2015: fig. 1) mentioned a close affinity of Hyopsodontidae and Macroscelides as “current consensus,” but found that the “case of the macroscelidean relationships … depends on the assumption that apheliscid ‘condylarths’ fall within Hyopsodontidae,” which they failed to establish in their own analysis.

Zack et al. (2005b) removed the apheliscine and louisinine genera from Hyopsodontidae and placed them in a separate family, Apheliscidae. This family Apheliscidae comprises several genera and species including Haplaletes andakupensis and Litomylus orthronepius, which could be the oldest stem macroscelideans by far with an estimated stratigraphic range of 66.043-63.3 mya. A relationship of Apheliscidae with Macroscelidea was supported by several studies (Simons et al. 1991Zack et al. 2005aPenkrot et al. 2008), which also postulated a possible Holarctic origin of this subgroup of Afrotherian mammals. This would be strange indeed, but also on anatomical grounds such a relationship with Macroscelidea still has to be considered as highly controversial, since Halliday et al. (2015) instead recovered Apheliscidae as a basal member of the Laurasiatheria, thus not even as afrotherian mammals.

A Big Question Mark

As just mentioned, Zack et al. (2005b) had transferred Louisininae from Hyoposodontidae to the reinstated family Apheliscidae. Louisinines also include the species Cingulodon magioncaldai and Monshyus praevius from the Hainin Formation in Belgium, which yielded the oldest Cenozoic mammalian fauna of Europe with its estimated age range of 66.043-61.7 mya (Sudre & Russell 1982De Bast & Smith 2017). Several authors considered a possible relationship of louisinines with Macroscelidea (e.g., Hartenberger 1986Simons et al. 1991Tabuce et al. 20012007, Holroyd & Mussell 2005, Zack et al. 2005a; see Benoit et al. 2013). This was indeed strongly suggested by Hooker & Russell (2012), who elevated them to a separate family, Louisinidae, in the stem group of Macroscelidea. Tabuce (2018) tentatively and with a big question mark concurred with this result, even though the study had significant flaws that were highlighted by De Bast & Smith (2017) who commented:

Hooker & Russell (2012) followed this hypothesis in the last revision of the family; however, they relied on a cladogram that lacked any extant member of Macroscelidea and in which most basal dichotomies are found only in the majority-rule consensus but not in the strict consensus tree. Here we place the family in Eutheria without further precision because we believe that the characters shared by Louisinidae and Hyopsodontidae with Macroscelidea are convergences due to the development of saltatorial capacities.

Two other hyopsodontid “condylarths,” i.e. Teilhardimys (= Microhyus) and Paschatherium,were recovered as very basal stem group representatives of Macroscelidea by Zach et al. (2005a) and Tabuce et al. (2006)Zach et al. (2005b) classified these two genera in Louisininae within Apheliscidae. In the cladogram of Tabuce et al. (2007) the North American apheliscines are closer to Macroscelidea, but the European louisinines “Microhyus” and Paschatherium closer to Paenungulata. On the other hand, Cooper et al. (2014) found that “the consistent recovery of the European louisinids Paschatherium and Teilhardimys as stem perissodactyls challenges the view that they are afrotherian macroscelideans.” The seminal work by Halliday et al. (2015) on Paleocene mammal affinities found neither Teilhardimys nor Apheliscidae to be closely related with Macroscelidea and indeed did not even support any of the mammalian supergroups (Xenarthra, Afrotheria, Euarchontoglires, and Laurasiatheria) as monophyletic unless constrained with molecular data. It must be really emphasized again: the very same fossils were in some studies resolved as close relatives of the order Macroscelidea and Paenungulata within the supergroup of Afrotheria, and in other studies recovered as close relatives of the unrelated order Perissodactyla (odd-toed ungulates) in the different supergroup Laurasiatheria. Have you ever heard any physicists hotly debating if a certain particle is a fermion or a boson? All such substantial problems were likely the reason that Senut & Pickford (2021) in the most recent work on macroscelidean evolution considered Louisinidae to be doubtfully related to Macroscelidea.

In their mentioned revision of Louisinidae, Hooker & Russell (2012) suggested in their maximum parsimony cladogram (see above for flaws in this analysis) that the genus Adunator and the families Adapisoricidae and Amphilemuridae successively represent the most basal branches of the macroscelidean lineage. Apparently, this was not accepted by any other authors (compare Tabuce et al. 2018), but still made it into the Paleobiology Database and the German Wikipedia. In spite of some similarities with hyopsodontid condylarths, these taxa are generally considered to be related to erinaceomorph insectivorans (hedgehogs) of the order Eulipotyphla (Russel 1964, Van Valen 1967Novacek et al. 1985Halliday et al. 2015). Senut & Pickford (2021) did not even bother to list them among their “families of doubtful or uncertain macroscelidean affinities.”

Last but not least, we must briefly discuss the leptictidians, an enigmatic group of bipedal mammals from the Paleogene. Asher & Seiffert (2010) placed the genus Leptictis from the Late Eocene and Oligocene of North America as sister group to Macroscelidea in their tree of Afrotherian phylogeny and biostratigraphic distribution. This phylogenetic position of leptictids was also supported in the cladograms of Zack et al. (2005a: fig. 3c)Hooker & Russell (2012), and O’Leary et al. (2013). The genus Leptictis and its relatives are generally recognized as a distinct order Leptictida, which even includes the Late Cretaceous genus Gypsnonictops. This would of course be an extremely interesting and very unusual dating for a subgroup of Afrotheria and alleged close relative of living elephant shrews. However, according to other experts (e.g. Wible et al. 2009), including the most recent study (Halliday et al. 2015), the Leptictida instead belong into the paraphyletic stem group of placental mammals, and therefore do not even represent real eutherians. Similarities of leptictids and macroscelideans therefore have to be considered as convergences, as already suggested by Butler (1995). Likewise, Meehan & Martin (2010) noted that “the morphology of leptictidans was highly convergent to that of extant macroscelideans, due to similar ecological specialisations to insectivory, digging, and saltatory locomotion” (Halliday et al. 2015).

An Apparently Simple Question

As you likely have realized by now, it can get pretty complicated to answer the apparently simple question of what the earliest fossil record of a certain group is, or even to find a consensus about what the group is. Anyway, from the above we can at least conclude that the oldest fossil record of unequivocal Macroscelidea is about 55.8-40 mya and that of the more doubtful alleged stem group about 66-56 mya. While there are some aberrant rodent-like and hyrax-like fossil elephant shrews, we do not find any fossil evidence for a gradual building of the distinct macroscelidean body plan during the Paleogene and Neogene periods (formerly called Tertiary). Likewise their theoretically predicted stem group in the Upper Cretaceous is totally missing in the fossil record, since Leptictida turned out to be stem eutherians. Just as all the other placental mammal orders, elephant shrews appeared abruptly on the scene, unless “hyopsodontids” like Louisinidae should indeed turn out to be stem group macroscelideans, which is still far from certain (see above). But even then, there would remain a large morphological gap between the condylarth-like stem taxa and unequivocal macroscelideans.

Just like its internal classification, the systematic affinities of the order Macroscelidea used to be a big mess, with very different candidates discussed as putative closest relatives, such as Archonta (treeshrews, colugos, and primates), Glires (rabbits, hares, and rodents), Eulipotyphla (insectivores), Tubulidentata (aardvarks), and Afrosoricida, as well as extinct Condylarthra (Patterson 1965Simons et al. 1991Butler 1995Tabuce et al. 2001Asher & Seiffert 2010Naish 2013Stevens et al. 2021). Rathbun (2005) therefore noted that “few mammals have had a more colourful history of misunderstood ancestry than the elephant-shrews.” Only after the advent of phylogenomics have they been recognized as members of the Afrotheria clade (Nishihara et al. 2005, Seiffert 2003, 2007Asher & Seiffert 2010). Next Fossil Friday we will look into another member of the Afrotheria, i.e., the order Afrosoricida that includes golden moles, otter shrews, and the iconic tenrecs of Madagascar. Stay tuned for more evidence on abrupt appearances in the mammalian fossil record.


Macroscelidea Butler, 1956 (40.4-0 mya)
            Rhynchocyonidae Gill, 1872 (40.4-0 mya)

                        Brevirhynchocyon Senut & Georgalis, 2014 (20-18 mya)
                                    (= Brachyrhynchocyon Senut, 2008)

                        Eorhynchocyon Senut & Pickford, 2021 (40.4-37.2 mya)

                        Hypsorhynchocyon Senut, 2008 (20-18 mya)

                        Miorhynchocyon Butler, 1984 (22.4-20.0 mya)

                        Oligorhynchocyon Stevens et al., 2021 (25.2 mya)

                        Rhynchocyon (20.4-0 mya)

            Macroscelididae (= Macroscelidae) Bonaparte, 1838 (40.4-0 mya)

                        Macroscelidinae Bonaparte, 1838 (5.332-0 mya)

                                    Macroscelides (5.332-0 mya)

                        Elephantulinae (23.03-0 mya)

                                    Elephantulus Thomas & Schwann, 1906 (5-0 mya)

                                    Galegeeska Heritage & Rayaleh, 2020 (0 mya)

                                    Hiwegicyon Butler, 1984 (14.2 mya)

                                    Miosengi Grossman & Holroyd, 2009 (20.43-15.97 mya)

                                    Nasilio Thomas & Schwann, 1906 (0 mya)

                                    Palaeothentoides Stromer, 1932 (23-2.5 mya)
                                    Petrodromus Peters, 1846 (0 mya)
                                    Petrosaltator Rathbun & Dumbacher, 2016 (0 mya)

                                    Pronasilio Butler, 1984 (13.8-13.7 mya)

                        Mylomygalinae Camp et al. 1953? or Patterson, 1965 (4-2.3 mya)

                                    Mylomygale Broom, 1948 (4-2.3 mya)

                        Namasenginae Senut & Pickford, 2021 (40.4-37.2 mya mya)

                                    Namasengi Senut & Pickford, 2021 (40.4-37.2 mya mya)

            Myohyracidae Andrews, 1914 (40.4-11.608 mya)

                        Myohyrax Andrews, 1914 (20.43-15.97 mya)

                        Promyohyrax Senut & Pickford, 2021 (40.4-37.2 mya mya)

                        Protypotheroides Stromer, 1922 (20-18 mya)

                        Rukwasengi Stevens et al., 2021 (25.2 mya)

                        Unnamed SN 10’08 Pickford et al., 2008 (47.8-41.2 mya)

            Afrohypselodontidae Senut & Pickford, 2021 (40.4-37.2 mya mya)

                        Afrohypselodontus Senut & Pickford, 2021 (40.4-37.2 mya mya)

Questionable putative stem Macroscelidea (66.043-26.3 mya)

            Adunator Russell, 1964 (63.3-55.8 mya)

            Litocherus Gingerich, 1983 (61.7-56.8 mya)

            Adapisoricidae Schlosser, 1887 (58.7-55.8 mya)

                        Adapisorex Lemoine, 1883 (58.7-55.8 mya)

            Amphilemuridae Hill, 1953 (55.8-26.3 mya)

                        Amphilemurinae Hill, 1953 (55.8-26.3 mya)

                                    Alsaticopithecus Hürzeler, 1947 (47.8-38 mya)

                                    Amphilemur Heller, 1935 (48.6-40.4 mya)

                                    Ankylodon Patterson & McGrew, 1937 (33.9-26.3 mya)

                                    Echinolestes Maitre et al., 2006 (37 mya)

                                    Gesneropithex Hürzeler, 1946 (55.8-33.9 mya)

                                    Macrocranion Weitzel, 1946 (55.8-37.2 mya)

                                    Neomatronella Russell et al., 1975 (55.8-48.6 mya)

                                    Patriolestes Walsh, 1998 (46.2-40.4 mya)

                                    Pholidocercus Koenigswald & Storch, 1983 (47 mya)

                                    Scenopagus McKenna & Simpson, 1959 (55.8-40.4 mya)

                                    Zionodon Dunn & Rasmussen, 2009 (46.2-40.4 mya)

                        Hylomysoidinae Tong & Wang, 2006 (55.8-48.6 mya)

                                    Hylomysoides Tong & Wang, 2006 (55.8-48.6 mya)

                                    Qilulestes Tong & Wang, 2006 (55.8-48.6 mya)

                        Sespedectinae Novacek, 1985 (50.3-37.2 mya)

                                    Colpocherus Beard & Dawson, 2009 (55.8-50.3 mya)

                                    Crypholestes Novacek, 1980 (46.2-40.4 mya)

                                    Proterixoides Stock, 1935 (40.4-37.2 mya)

                                    Sespedectes Stock, 1935 (40.4-37.2 mya)

            Apheliscidae Matthew, 1915 (66.043-48.6 mya)

                        Aletodon Gingerich, 1977 (56.8-55.8 mya)

                        Apheliscus Cope, 1875 (56.8-50.3 mya)

                                    (= Parapheliscus Van Valen, 1967)

                        Dorralestes Gingerich, 1983 (61.7-56.8 mya)

                        Gingerichia Zack et al. 2005 (61.7-56.8 mya)

                        Haplaletes Simpson, 1935 (66.043-56.8 mya)

                        Haplomylus Matthew, 1915 (56.8-48.6 mya)

                        Litomylus Simpson, 1935 (66.043-56.8 mya)

                        Phenacodaptes Jepsen, 1930 (56.8-55.8 mya)

                        Utemylus Gingerich, 1983 (61.7-56.8 mya)

            Herodotiidae Simons et al., 1991 (55.8-28.4 mya)

                        Chambius Hartenberger, 1986 (55.8-40.4 mya)

                        Eotmantsoius Tabuce et al., 2012 (48.6-33.9 mya)

                        Herodotius Simons et al., 1991 (33.9-28.4 mya)

                        Nementchatherium Tabuce et al., 2001 (40.4-33.9 mya)

            Louisinidae Sudre & Russell, 1982 (66.043-55.8 mya)

                        Berrulestes Hooker & Russell, 2012 (58.7-55.8 mya)

                        Cingulodon Be Bast & Smith, 2017 (66.043-61.7 mya)

                        Dipavali Van Valen, 1978 (58.7-55.8 mya)

                        Gigarton Hooker & Russell, 2012 (58.7-55.8 mya)

                        (Khamsaconus Sudre et al., 1993) (55.8-48.6 mya)

                                    (turned out to be a Tethytheria)

                        Louisina Russell, 1964 (58.7-55.8 mya)

                        Monshyus Sudre & Russell, 1982 (66.043-61.7 mya)

                        (Paschatherium Russell, 1964) (55.8-48.6mya)

                                    (belongs to Perissodactyla according to Cooper et al. 2014)

                        Prolouisina Hooker & Russell, 2012 (61.7-58.7 mya)

                        (Teilhardimys Kretzoi & Kretzoi, 2000) (58.7-48.6 mya)

                                    (= Microhyus Teilhard de Chardin, 1927)

                                    (belongs to Perissodactyla according to Cooper et al. 2014)

                        Thryptodon Hooker & Russell, 2012 (58.7-55.8 mya)

                        Walbeckodon Hooker & Russell, 2012 (61.7-55.8 mya)

            Metoldobotidae Simons et al., 1991 (33.9-28.4 mya)

                        Metoldobotes Schlosser, 1910 (33.9-28.4 mya)