z z 2 z ee

TITUTION NOILOLILSNI~ NVINOSHLIWS S3IY¥VUGIT LIBRARIES SMITHSONIAN !

5 - ca ani

—= o : = wo — i)

= a \ 5 a 5 ka

: 2 Ne: 3 > 5

= zw A YAS bo] — w

- = \ \S - std a =

” Ne ” 2)

z m ~ AS = m =z a,

<= 7) < n = a8

yuvusit ~t! B RARI ES SMITHSONIAN INSTITUTION NOILALILSNI _NVINOSHLINS, S34 1uVva wi

a = = = < = fy,

3 z a : 3 : Ua

< oO uw D B 2 w hip ay

= AN. > = > = >

” =. 7) 2 77) =z

NOILNLILSNI_ NVINOSHLINS LIBRARIES SMITHSONIAN

” 2 % wn > ”

as mz) Ba w a un

lead = oe = ae Sa

si e Fe c < oon

ow S cc = role =

m ro) : a re) a re}

= =z a 2 ae z

iINVYHNYSIT LIBRARIES SMITHSONIAN INSTITUTION NOILONLILSN!I NVINOSHLIWS

2 5 = = ZO ae z

o = oO = o an 4

x 5 tit GF > 5 2 SN 5

; GM : : ENY =

2 a U, ras - = =

m 2 jl m Z2 m Loe :

TITUTION NOILNLILSNI_ NVINOSHLINS S3IYWYSIT_LIBRARIES. SMITHSONIAN INSTITUTE

: o” = n Zz ae ” = x.

Se ee = = =< G = = Xx

. = z = ZWNS 5S =z S

: at re) = 59 CR = 6 :

2 2 g ; ? ® \.. 8 2 4

2 = z iE SS =

2 2 S = NOs =

_LIBRARIES SMITHSONIAN _ INSTITUTION NOILNLILSNI_ NVINOSHLIWS s3aiuva |

a ws Z iw 2 &. U;

g aay a6 f Ms, 0G PES pa -_

: Eff 3 : : :

re) Fs Ye °o = ro) =

2 J = a z =)

TITUTION NOILNLILSNI NVINOSHLINS S3I1¥vVuYdgIT LIBRARIES

z ee a Se

~_ ise] = ow — (‘fp wo

5 2 5 2 5 ly, Sa

= ee * E Lee

e 2 = 2 = OS Ki 2

” m m n°” m

= “ = n < 7) 1

iuvudgd rut B RARI ES SMITHSONIAN INSTITUTION NOILNLILSNI _NVINOSHLIWS S3 1uVvEg ,

, 7, < ¥ = z ‘ = < = yg

Wp = o z x = t z 4

be J 3 iG 8 Nw a id a

If = ™. = ae Pa ‘AW r Oo <x oO

= AS = E YQ Ss z = =,

Bre ie 5: :

EITUIOR NOLEN LUSNIT NVINGSHLNG 33 lyvuag I7_LIBRARI ES SMITHSONIAN _INSTITUTI

= o a a = a PLD,

_ 4 = 4 = ~ ::

< CS . < c < _ =

e par) : ce par) a , +4

S ro) A OS ro} Be Oo .

ng 2 _ 2 a) = ie

1yvugi7z LIBRARIES SMITHSONIAN EN sat Sere NOILNLILSNI NYINOSHLINS S31 uVUE

rm a i za iG

= Oo = Sealy ro} = Oe re)

iz) o — 47a “gs SN = ASX of S =

;

a ax

= LIBRARIES SMITHSONIAN” INSTITUTION NOILALII

NVINOSHLINS S31luvugit

~ 3 fe)

and 2a =a

c z e = SEO z

ao » = ow =_ (3 ow =

2 \X® 2 Zz & $Y > 2

1} > MS fe > = 7 i> =

a WS - se ra YF fies =

: = WY 2 = ee ae Q

/ w fart = (27) = 77) =

ARIES, SMITHSONIAN INSTITUTION NOILOJILSNI_ NVINOSHLINS, Saluvud 11 LIBRAR

= z 5 = < = Vi a

4 Pape = = = z= .

= s NAG ig 5 Uefia SX

2 E Lv" 2 iS Z, Ee ON

i #2 See ee : : ees

ALILSNI. NVINOSHLINS S3IYVYGIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLI

ae 7) = ” S w”

a = o = G zr

: & : < 3 :

s : sd = or =: re)

- 2 "S 3 S ant = ar

. 5 wd

RARIES SMITHSONIAN, INSTITUTION NOLLALILSNI NVINOSHLINS Sa1uvudi7_LIBRAR

z a & z iB

5 : = = : :

ES = E = - z

a = = = . =

hee Z p ae a

JNVINOSHLINS, S31YVY E17 LIBRARIES, SMITHSONIAN _INSTITUTION,, NOILALI

Sf. Ne.

”n wD» ” WX - @ am ° SA D ¥f.

= 2 EWG 2 EWS 2 Z|

a z a . 2B = Fa

RARIES SMITHSONIAN INSTITUTION NOILNLILSNI NVINOSHLINS Saluvugi7_l!BRAF

w” > “” S ee n” =

te) SS

= ae - . 72 4 pl = e ASS = a

a . aa 4 Hd -_ a -~ cS oS a ‘ SA a par

aif Z a < 2 WAS < c

« Gry = ce ear SS oe 4

a Y rat = = - om 5

Be vii a = < WANT * NOLLAL

INLILSNI_NVINOSHLIWS ~°3 tyuvyuag Pte LIBRARI ES_ SMITHSONIAN, INSTITUTION = !

= . (@) =— Oo A = (e)

ow ‘ = o = (fp wo eS

cad ys 5 a 5 , My, ah 5

> VOW E > = F jp E

Ps] 3 YA ra Bs] cay Vf C4. wa ke

= AS SS wo par De * = a

m “aS z m z= m =z

4) ; = w = 77) =

IRARIES SMITHSONIAN INSTITUTION NOILNLILSNI_NVINOSHLINS SaluvadI7_LIBRAI

t 2 *

= Pa X + = =a ws (7 [@) ;

NY Ae! rT ~W IANS Oo = SOG FN

sy z E XS = = = Ss,

} as So = 2 = ee ne

AMLULSNI_NVINOSHLINS S3IYVYAIT_LIBRARIES SMITHSONIAN _INSTITUTION Coe

z = 4 FA ul

“n = Ww Z =

c = = YG," =

eo + < YY fg Ii A <

c c “2° GS fe

5 = 3S = * Y fo 5 m

2 a 2 = z =

BRARIES SMITHSONIAN” INSTITUTION NOILNLILSNIT NVINOSHLINS Saluvadi7 LIBRA

<1HSON, 5 < auSON, S = NS = v 1150

Sao og Ce So oem a Os 8 CSD © Gan

ANNALS OF THE _ ANNALE VAN DIE

SOUTH AFRICAN MUSEUM SUID-AFRIKAANSE MUSEUM

VOLUME 81 BAND 81

ANNALS OF THE SOUTH AFRICAN MUSEUM

ANNALE VAN DIE SUID-AFRIKAANSE MUSEUM

VOLUME 81 BAND

THE TRUSTEES OF THE DIE TRUSTEES VAN DIE

SOUTH AFRICAN MUSEUM SUID-AFRIKAANSE MUSEUM

CAPE TOWN KAAPSTAD

1980

@ SET, PRINTED AND BOUND IN THE REPUBLIC OF SOUTH AFRICA BY

THE RUSTICA PRESS (PTY.) LTD., WYNBERG, CAPE

336

LIST OF CONTENTS

Page

CHALONER, W. G., Forey, P. L., GARDINER, B. G., Hitt, A. J. & YOunNG, V. T.

Devonian fish and plants from the Bokkeveld Series of South Africa. (Published

February 1980.) aa Ba ae aS ae ie a ras ned 27h

GRINE, F. E. see TOLLMAN, S. M.

Haun, B. D. see TOLLMAN, S. M.

HENDEY, Q. B.

Agriotherium (Mammalia, Ursidae) from Langebaanweg, South Africa, and

relationships of the genus. (Published February 1980.) .. ae as Me 1

KENNEDY, W. J. see KLINGER, H. C.

KLEIN, R. G.

Environmental and ecological implications of large mammals from Upper

Pleistocene and Holocene sites in southern Africa. (Published June 1980.) .. 223

KLINGER, H. C. & KENNEDY, W. J.

The Umzamba Formation at its type section, Umzamba Estuary (Pondoland,

Transkei), the ammonite content and Se cee distribution.

(Published August 1980.) .. Ais at a ; Ko — ae 207,

Louw, E.

The South African Museum’s Meiring Naude Cruises. Part 10. Station data 1977,

1978, 1979. (Published April 1980.) oF ; a Si

TOLLMAN, S. M., GRINE, F. E. & HAHN, B. D.

Ontogeny and sexual dimorphism in Aulacephalodon eae Anomodontia).

(Published February 1980.) 56 ae = : ai ae ae tel

VAN DEN HEEVER, J. A.

On the validity of the Therocephalian oe asc ae paisa

(Published February 1980.) ; , 111

;

: *

; 7

1 \

a 4

; ;

a }

. ro * A

) aia

;

a 2

7 a %

J *

a ‘ : Sass er, oe

y - os é 4 . ‘ om ; + a

ey ere ee wal Piya ghee a ala NS

; ’ 7 ) b Pear 4

i : Ses i oo. ‘

i it r ¥

a ay

2 , : :

‘

’ r

a » \ rf

, zs

i 1

Dar

: , is

‘ ae oe

Uy a”

; ,

‘

mae 5

NEW GENERIC NAMES PROPOSED IN THIS VOLUME

Page

Barrydalaspis Chaloner, Forey, Gardiner, Hill & Young, 1980 ts ae a el 29

‘ 'v

; , “ 7 . i

‘ ab,

‘ ‘

4 * ,

‘

“

5 ? 7

: t

a ox ft he

a q en

. ru *

Re 4

z *

; y 4

roa? te yy On Me ns F i

Ri ‘ ‘ F

‘ ; y i =

iy j

‘ |

f BS

f +t

pes Ne

é ‘

I»

‘

OLUME 81 PART 1 FEBRUARY 1980 ISSN 0303-2515

ANNA

OF THE SOUTH AFRICAN |

MUSEUM |

CAPE TOWN

INSTRUCTIONS TO AUTHORS

|. MATERIAL should be original and not published elsewhere, in whole or in part.

2. LAYOUT should be as follows:

(a) Centred masthead to consist of

Title: informative but concise, without abbreviations and not including the names of new genera or species

Author's(s’) name(s)

Address(es) of author(s) (institution where work was carried out) _

Number of illustrations (figures, enumerated maps and tables, in this order)

(b) Abstract of not more than 200 words, intelligible to the reader without reference to the text

{c) Table of contents giving hierarchy of headings and subheadings

(d) Introduction : : :

(e) Subject-matier of the paper, divided into sections to correspond with those given in table of contents

(f) Summary, if paper is lengthy

(g) Acknowledgements

(h) References

(i) Abbreviations, where these are numerous

3. MANUSCRIPT, to be submitted in triplicate, should be typewritten and neat, double spaced

with 2,5 cm margins all round. First lines of paragraphs should be indented. Tables and a list of

legends for illustrations should be typed separately, their positions indicated in the text. All

pages should be numbered consecutively.

Major headings of the paper are centred capitals; first subheadings are shouldered small

capitals; second subheadings are shouldered italics; third subheadings are indented, shouldered

italics. Further subdivisions should be avoided, as also enumeration (never roman numerals)

of headings and abbreviations.

Footnotes should be avoided unless they are short and essential.

Only generic and specific names should be underlined to indicate italics; all other marking

up should be left to editor and publisher.

4. ILLUSTRATIONS should be reducible to a size not exceeding 12 « 18 cm (19 cm including

legend); the reduction or enlargement required should be indicated; originals larger than

35 x 47 cm should not be submitted; photographs should be rectangular in shape and final

size. A metric scale should appear with all illustrations, otherwise magnification or reduction

should be given in the legend; if the latter, then the final reduction or enlargement should be

taken into consideration.

All illustrations, whether line drawings or photographs, should be termed figures (plates

are-not printed; half-tones will appear in their proper place in the text) and numbered in a

single series. Items of composite figures should be designated by capital letters; lettering of

figures is not set in type and should be in lower-case letters.

The number of the figure should be lightly marked in pencil on the back of each illustration.

5. REFERENCES cited in text and synonymies should all be included in the list at the end of

the paper, using the Harvard System (ibid., idem, loc. cit., op. cit. are not acceptable):

(a) Author’s name and year of publication given in text, e.g.:

“Smith (1969) describes...’

‘Smith (1969: 36, fig. 16) describes...’

“As described (Smith 1969a, 19696; Jones 1971)’

‘As described (Haughton & Broom 1927)...’

“As described (Haughton et a/. 1927)...’

Note: no comma separating name and year

Dagination indicated by colon, not p.

names of joint authors connected by ampersand

- et al. in text for more than two joint authors, but names of all authors given in list of references.

(b) Full references at the end of the paper, arranged alphabetically by names, chronologically

within each name, with suffixes a, b, etc. to the year for more than one paper by the same

author in that year, e.g. Smith (1969a, 19695) and not Smith (1969, 1969a).

For books give title in italics, edition, volume number, place of publication, publisher.

For journal article give title of article, title of journal in italics (abbreviated according to the World list o,

scientific periodicals. 4th ed. London: Butterworths, 1963), series in parentheses, volume number, part

number (only if independently paged) in parentheses, pagination (first and last pages of article).

Examples (note capitalization and punctuation)

BuLLouGn, W. S. 1960. Practical invertebrate anatomy. 2nd ed, London: Macmillan.

FIscHER, P.—H. 1948. Données sur la résistance et de le vitalité des mollusques. J. Conch., Paris 88: 100-140.

FiscHER, P.-H., DuvAL, M. & Rarry, A. 1933. Etudes sur les échanges respiratoires des littorines. Archs

Zool. exp. gén. 74: 627-634.

Konn, A. J. 1960a. Ecological notes on Conus (Mollusca: Gastropoda) in the Trincomalee region of Ceylon.

Ann. Mag. nat. Hist. (13) 2: 309-320.

Konn, A. J. 1960b, Spawning behaviour, egg masses and larval development in Conus from the Indian Ocean.

Bull. Bingham oceanogr. Coll. 17 (4): 1-51.

Ture, J. 1910. Mollusca: B. Polyplacophora, Gastropoda marina, Bivalvia. Jn: SCHULTZE, L. Zoologische

und anthropologische Ergebnisse einer Forschungsreise im westlichen und zentralen Siid-Afrika 4: 269-270.

Jena: Fischer. Denkschr. med.-naturw. Ges, Jena 16: 269-270.

(continued inside back cover)

ANNALS OF THE SOUTH AFRICAN MUSEUM

ANNALE VAN DIE SUID-AFRIKAANSE MUSEUM

Volume 81 Band

February 1980 Februarie

Part 1 Deel

AGRIOTHERIUM (MAMMALIA, URSIDAE) FROM

LANGEBAANWEG, SOUTH AFRICA, AND

RELATIONSHIPS OF THE GENUS

Q. B. HENDEY

Cape Town Kaapstad

The ANNALS OF THE SOUTH AFRICAN MUSEUM

are issued in parts at irregular intervals as material

becomes available

Obtainable from the South African Museum, P.O. Box 61, Cape Town 8000

Die ANNALE VAN DIE SUID-AFRIKAANSE MUSEUM

word uitgegee in dele op ongereelde tye na gelang van die

beskikbaarheid van stof

Verkrygbaar van die Suid-Afrikaanse Museum, Posbus 61, Kaapstad 8000

OUT OF PRINT/UIT DRUK

1, 2(1-3, 5-8), 3(1-2, 4-5, 8, t.-p.i.), 5(1-3, 5, 7-9),

6(1, t.—p.i.), 711-4), 8, 9(1-2, 7), 10(1-3),

11(1-2, 5, 7, t.—p.i.), 15(4-5), 24(2), 27, 31(1-3), 32(5), 33

Kopieregnavrae aan die Suid-Afrikaanse Museum

ISBN 0 908407 84 X

Printed in South Africa by In Suid-Afrika gedruk deur

The Rustica Press, Pty., Ltd., Die Rustica-pers, Edms., Bpk.,

Court Road, Wynberg, Cape Courtweg, Wynberg, Kaap

AGRIOTHERIUM (MAMMALIA, URSIDAE) FROM LANGEBAANWEG,

SOUTH AFRICA, AND RELATIONSHIPS OF THE GENUS

Q. B. HENDEY

South African Museum, Cape Town

(With 42 figures and 21 tables)

LMS. accepted 11 September 1979]

ABSTRACT

Agriotherium africanum from the latest Miocene/early Pliocene Varswater Formation at

Langebaanweg, and other relevant material indicate that Agriotherium was descended from

late Miocene Indarctos. Later Ursidae are divided into the subfamilies Ursavinae (Ursavus

spp), Agriotheriinae (tribes Agriotheriini and Ailuropodini), and Ursinae (Ursini and Tre-

marctini). ‘Ursavus’ depereti and Ailuropoda melanoleuca constitute the Ailuropodini, and

Indarctos and Agriotherium the Agriotheriini. The latter consist of a primary European lineage

from which branches in Africa, Asia and North America arose. Either all species of Indarctos,

or the later ones only, should perhaps be referred to Agriotherium.

CONTENTS

PAGE

Introduction . : ; 1

The Langebaanweg Agriotherium ; 4

Material . : ; ; : : 4

The skull ‘ ; ‘ 9

The postcranial skeleton : : 29

Discussion . ¥ ‘ ; ‘ 51

Palaeoecology : ; : 53

Relationships of griotherium : ; 72

Nomenclature PUM Gees Soll fei as 98

Summary : 4 3 . 104

Acknowledgements saan . mr 106

References . ; ? : ; . 106

INTRODUCTION

Agriotherium africanum from the latest Miocene/early Pliocene Varswater

Formation exposed in a phosphate mine (‘E’ Quarry) at Langebaanweg, Cape

Province (Fig. 1), was the first member of its genus to be recorded in Africa,

and the first bear known from sub-Saharan Africa (Hendey 1972: Wolff et al.

1973). Although the first Agriotherium specimens were found in Europe at

least as long ago as 1809 (Stehlin 1907), and others have since been collected

at scattered localities through much of the Old World and North America, this

genus has remained relatively poorly known, being represented for the most

part by fragmentary material belonging to few individuals.

Initially this also applied to the Langebaanweg representative of the genus,

but later more material came to light and it is remarkable that ‘Agriotherium

africanum, the most recently described and most remote record of the genus,

is apparently also the best represented’ (Hendey 1977: 112). It is in the latter

Ann. S. Afr. Mus. 81 (1), 1980: 1-109, 42 figs, 21 tables.

2 ANNALS OF THE SOUTH AFRICAN MUSEUM

LOWER COURSE OF RIVER

DURING LATE MIOCENE/

EARLY PLIOCENE

ATLANTIC —

OCEAN

20 «kilometres

Fig. 1. The south-western Cape Province.

respect that A. africanum is particularly noteworthy. It provides the best indi-

cation yet that Agriotherium, like other bears, was characterized by appreciable

variation in its dentition and in size, the latter evidently being due to marked

sexual dimorphism.

A. africanum specimens are recorded from two stratigraphic horizons in

‘E’ Quarry, namely beds 3aS and 3aN of the Pelletal Phosphorite Member

(Hendey 1976). The bed 3aS sample, which includes the holotype, comprises

only a small part of the total assemblage. Most of the bed 3aN sample was

collected in a restricted area (the fossil accumulation at Locality 5—Dingle

et al. 1979, fig. 2). This material is from a river channel lag deposit, which was

laid down in the lee of a phosphate rock outcrop, while the remainder of the

bed 3aN sample was from lateral extensions of the lag deposit. Since this

material was probably accumulated during a very restricted period, the bed 3aN

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 3

A. africanum sample may represent remnants of a single population in a temporal

as well as a geographical sense. There is no reason to believe that the character

diversity observed in this sample can be ascribed to anything but normal intra-

specific variation.

On the other hand, there are differences between bed 3aS and bed 3aN

specimens which apparently reflect evolutionary changes, suggesting that the

time interval between deposition of these beds was of sufficient duration for

such changes to have occurred. The differences are, however, insufficient to

warrant formal nomenclatural recognition. A. africanum thus joins the growing

number of species which indicate that deposition of the Varswater Formation

took place over an appreciable period in time (see Hendey 1978; Gentry

1980). Langebaanweg is the only recorded locality where Agriotherium is

known to occur in more than one stratigraphic horizon.

A. africanum is also of interest since it is one of several species now known

from sub-Saharan Africa which indicate that towards the end of the Miocene

this region had closer faunal links with Eurasia (Hendey 1978). The present

study of Agriotherium has suggested that faunal interchange at this time was

between Africa and Europe, presumably by way of an Iberian/north-west

African connection, rather than, or in addition to, one between Africa and

Asia via the Middle East. Apart from A. africanum, the Langebaanweg car-

nivores, Plesiogulo monspessulanus and Dinofelis diastemata, also point to such

a connection, as have recent studies on other mammals (e.g. Forstén 1978).

Until the recent discovery of an agriotheriine at Sahabi in Libya (Boaz

et al. 1979), the Langebaanweg Agriotherium was the only African record of

the group, and its location at the southern continental extremity was evidence

that it had been widespread on this continent.

On a more mundane level, A. africanum has an appeal in that it is still

South Africa’s only known bear, living or fossil.

The material described below is housed in the South African Museum,

and catalogue numbers are prefixed SAM-PQ., which identifies the institution

and department concerned. This lettering is omitted from the text, and the

locality prefix (L) and serial numbers of specimens only are given. Modern

comparative material in this museum is distinguished by the prefix ZM.

Other institutional abbreviations used in the text are as follows:

BMNH-—British Museum (Natural History), London

GSI —Geological Survey of India, Calcutta

NMB —Naturhistorisches Museum, Basel

NMW —Naturhistorisches Museum, Vienna

Originals and casts of Agriotherium and related taxa in, or from, the above

institutions were examined in the course of the present study. In addition,

skulls and skeletons of extant ursids in the South African Museum and the

British Museum (Natural History) were studied. Comparative data were other-

wise obtained from the publications cited.

4 ANNALS OF THE SOUTH AFRICAN MUSEUM

THE LANGEBAANWEG AGRIOTHERIUM

MATERIAL

The available Agriotherium africanum specimens are listed below according

to the horizon from which they were derived, and, where possible, according

to sites or areas within ‘E’ Quarry.

Pelletal Phosphorite Member, bed 3aS

L2045—Left maxillary fragment with P* (holotype)

L1868—Left I,, right P°, fragments of right P* and M?

L1844, L3141—Left I, and I,, probably of the same individual (Wolff et al.

(1973) were incorrect in identifying L1844 as a right I,)

L12637—Incomplete left M! (Hendey (1972) was incorrect in identifying this

tooth as M?)

L2154—Incomplete proximal left ulna

The above specimens were described or discussed by Hendey (1972) and

Wolff et al. (1973).

L12561—Right M, (see Hendey 1972)

L12033, L41270—Right I*’s

L40031—Left hemimandible fragment and distal left humerus

L40030— Right metatarsal II and 2nd phalanx, probably of same individual as

L40031

L40002—Left femur and patella, and a thoracic vertebra, of one individual

L3433—Left humerus lacking proximal parts

L3994— Shaft of left humerus

L40040—Distal right humerus

L12383—Proximal right ulna

L40003—Left radius lacking distal parts

L41702—Left scapholunar

L41295— Right cuneiform

L12503—Right pisiform

L20998, L25862— Right metacarpals III

L41575— Right metacarpal IV

L40043, L40128— Proximal right and left femora

L40028, L40029— Proximal right tibiae

L41108—Proximal right metatarsal V

L10411—Proximal Ist phalanx

L42667—2nd phalanx

The above material represents at least three individuals, but since the

specimens were collected over a wide area and from different levels within

bed 3aS, the actual number is certainly much higher. Unless otherwise stated,

each catalogue number could represent a different individual, which makes a

total of 28.

It is not certain that those specimens in this series with numbers above

L40000 are all from bed 3aS (see p. 62).

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 5)

Pelletal Phosphorite Member, bed 3aN—excavation LBW-E 1975/1 (site—

TCWW Prom)

L33160—Incomplete left manus, comprising scapholunar, unciform, magnum,

trapezoid and metacarpals II to V, the latter lacking distal ends and some

of the carpals incomplete

L33341, L34188—Incomplete right and left innominates

L30205—Left metatarsal III

L33557— Proximal left metatarsal V

The above material represents at least 2 individuals.

Pelletal Phosphorite Member, bed 3aN—excavation LBW-E 1976/1 (site—RP)

L33824—Crown of left C.

L33825—Left I,

L13826—Proximal right radius

L33828— Right metacarpal IV

L33830—Distal metapodial fragment

The above material represents at least one individual.

Pelletal Phosphorite Member, bed 3aN—excavation LBW-E 1976/2 (site—IWRP

or Locality 5 of Dingle et al. 1979)

L45062—Incomplete skull (partly restored) and mandible (restored), lacking

left I, right I°, left and right P', right lower incisors, and left P,

Associated postcranial bones, including: parts of at least 7 vertebrae

(some numbered L49048 and L49115); fragment of distal right humerus;

right ulna lacking distal parts; right scapholunar, cuneiform, pisiform,

unciform, magnum and trapezoid; proximal right metacarpal III; proximal

left humerus; left ulna with distal parts detached and fragmented; left

radius lacking part of shaft; right tibia and proximal fibula; right astragalus,

calcaneum, navicular and cuboid; right metatarsals II, IV and V; left

astragalus; left metatarsal III and proximal metatarsal II; 11 sesamoids,

five Ist phalanges, four 2nd phalanges and four 3rd phalanges

L45137—Right I?, C lacking root, and P* to M2; left C and P* to M2; some skull

fragments

L45114—Right hemimandible lacking ascending ramus, incisors, P,; and M3

The above specimens were briefly discussed by Hendey (1977).

146605, L48564, L48577—Left and right maxillary fragments with M?’s and

part of left M1

L46573—Crown of right I?

L47758/9—Right and left I°, probably of the same individual

L48851— Fragment of left I°

L46074—Right P*

L47698—Left M?

1L46563—Left M., probably of same individual as L45114

L48742— Fragment of mandibular condyle

6 ANNALS OF THE SOUTH AFRICAN MUSEUM

L47449, L47701, L47830—Incomplete cervical vertebrae

L45063—Incomplete left forelimb, comprising humerus (partly restored), with

proximal end detached; ulna and radius (restored); scapholunar, cuneiform,

pisiform, unciform, magnum, trapezoid and trapezium; metacarpal, V,

and fragments of metacarpals I, III and IV; three sesamoids; three Ist

phalanges

L46602— Proximal left humerus

L48741, L48747—Proximal left radii

L47699 /700— Proximal left and right radii of one individual

L46076—Right ulna lacking distal end

L46134—Left scapholunar, right pisiform and metapodial fragment, probably

of one individual

L46132, L48021—Left and right scapholunars

L47074— Right magnum

L45448—Left metacarpal I and distal metapodial fragment, probably of one

individual

L48432—Right metacarpal IV

L49889— Distal left femur

L47533, L47910—Left and right astragali

L47387— Proximal right metatarsal I

L48572 /3— Right metatarsal V and left metatarsal II, probably of one individual

L46133, L46216, L48766—Metapodial fragments

L47358, L48533, L48730, L49888—1st phalanges

L48213, L48230—2nd phalanges

The above material represents at least five individuals.

Pelletal Phosphorite Member, bed 3aN—surface finds in the vicinity of LBW-E

1976 |2

L47242—Left premaxilla and right M? of one individual.

L50636—Right I?

L50981—Left M,

L42537—Proximal left ulna and distal femur, lacking epiphyses, of one

individual

L50635—Left scapholunar

L50638— Patella

L50637—Distal metapodial

L41468— Ist phalanx

The above material represents at least two individuals.

Pelletal Phosphorite Member, bed 3aN—dump 10 sample from deposits immedi-

ately north of LBW-E 1976/2

L55012—Left and right P*’s and right M? of one individual

L55015, L55016—Right I*’s

L55014—Left I*

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA if.

L55013—Crown of right C

L55017— Four anterior premolars of more than one individual

L55029— Right metacarpal I

L55019— Distal left fibula

L55021— Right navicular

L55022/28—Seven metapodial fragments

L55020—Sesamoid

L55030/35—Six Ist phalanges

L55036—2nd phalanx

L55037/43—Seven 3rd phalanges

The above material represents at least two individuals.

Pelletal Phosphorite Member, bed 3aN—dump 9 sample from deposits immedi-

ately west of LBW-E 1976/2

L50445—Incomplete left hemimandible with C and P,

L50453— Fragment of left P*

L50458— Right I,

L50446— Right M,

L50457— Left magnum

L50454— Distal metapodial fragment

L51592—2nd phalanx

L50455/6—Two 3rd phalanges

The above material represents at least one Gadiadnall

Pelletal Phosphorite Member, bed 3aN—dump 8 sample from deposits immedi-

ately south of LBW-E 1976/2

L50003—Incomplete left hemimandible with P, and M,

L50004— Incomplete left hemimandible with P, and associated M, and M,

L50903— Fragments of right hemimandible with incomplete C

L50008—Incomplete right P*

L50005/6— Right and left M,’s

L50007— Right M,

Unnumbered teeth and tooth fragments, including right I’, I° and I,, a canine

and two anterior premolars

L50843 /54— Eleven vertebra fragments

L50857/8—Incomplete left and right scapulae

L50834/5/6/8/9—Humerii fragments of at least two individuals

L50806, L50816—Proximal left and right radii

L50807— Distal right radius fragment

L50777— Distal epiphysis of right radius

L50763—Left ulna lacking distal end

L50764— Left ulna fragment

L50840— Proximal ulna fragment

8 ANNALS OF THE SOUTH AFRICAN MUSEUM

L50805, L50808— Distal right ulnae

L50767— Right scapholunar

L50813, L50819—Fragments of right and left scapholunars, probably of one

individual

L50786, L50795—Left and right cuneiforms

L50772, L50774— Right unciforms

L50791, L50794— Right magnums

L50775—Left metacarpal I

L50783, L50788—Left and right metacarpals II

L50810—Proximal right metacarpal III

L50855/6/9/60—Femora fragments of at least two individuals

L50841, L50833— Proximal and distal left tibia, probably of one individual

L50842—Proximal right tibia

L50769— Distal epiphysis of right tibia

L50815, L50828— Distal left and right fibulae

L50765/6—Left astragali

L50770, L50789— Right astragali

L50768— Right calcaneum lacking tuber calcis epiphysis

L50773—Left navicular

L50778, L50790— Right and left entocuneiforms

L50771, L50787—Left and right metatarsals I

L50824— Proximal left metatarsal IV

L50812, L50829— Proximal right metatarsals V

L50809/11/14/17/18/20/22/23 /25/26/30—Eleven metapodial fragments

L50802—Sesamoid

L50776/80/81 /84/93 /97—Six 1st phalanges

L50821/7—Two incomplete Ist phalanges

L50785/96/804—Three 2nd phalanges

L50782/98/99 /800/801—Five 3rd phalanges

L50831—Incomplete 3rd phalanx

The above material represents at least four individuals.

Pelletal Phosphorite Member, bed 3aN—carbonaceous deposit south of dump 8

area

L41404— Fragmented and incomplete skull, with only the left premaxilla and

maxilla largely intact, and with right ?P?, and left C (damaged), P*, M1 and

M? (see Hendey 1977)

L40044— Mandible fragment with associated right radius and fragments of ulna

L43126— Fragments of at least six thoracic and lumbar vertebrae.

The above material represents at least two individuals.

The minimum number of individuals represented in the combined bed 3aN

samples is eleven, although the actual number may be far higher. These samples

are from a single horizon of river channel and associated deposits, which were

exposed over a linear distance of about 200m, with sample areas either contiguous

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 9

or not far removed from one another. Elements of individual skeletons are

likely to be represented in more than one sample unit.

THE SKULL

Apart from L45062, the only other described skull of Agriotherium is one

belonging to A. sivalense from the Siwalik Hills in India (Falconer & Cautley

1836; Lydekker 1884). Both skulls are incomplete, and although L45062 is also

slightly distorted in parts, it is perhaps the more informative of the two speci-

mens. They are similar in overall size, and, in so far as comparisons are possible,

they are also similar in morphology.

Another skull which is remarkably like that of the Langebaanweg Agrio-

therium is of an Indarctos from Florida which was recently described by Wolff

(1978). Although the Langebaanweg and Florida skulls undoubtedly do belong

to Agriotherium and Indarctos as these genera are presently conceived, the

description of skull characters in the latter (Wolff 1978: 2-4) could, with only

slight modification, serve as a description of the Langebaanweg specimen. The

significance of this will be discussed later (see p. 93).

Other described skulls of Indarctos, of which there are two from Samos

(Helbing 1932; Thenius 1949, 1959) and one from Spain (Crusafont & Kurtén

1976), are less like the Langebaanweg specimen because they represent species

which are more primitive than that from Florida.

A summary account of the skull characters of L45062 has been given

elsewhere (Hendey 1977), but they are dealt with in more detail here.

The skull of A. africanum (Fig. 2) differs in certain aspects from those of

all living bears. Its most striking characteristic is its massive size (Table 1),

although in some dimensions it is matched, or even surpassed, by skulls of

male Kodiak bears (Ursus arctos middendorffi), which are amongst the largest

of living bears (Hendey 1977, table 1). A. africanum is otherwise most readily

distinguished from living bears by its relatively short and broad snout, while

the braincase, which is surmounted by a very high saggital crest, appears small

by comparison. The nuchal crest is also very prominent and dorsally projects

well behind the occipital condyles. The zygomatic arches are very stout and

strongly arched. They resemble those of the giant panda, Ailuropoda melano-

leuca, more than any other living ursid, and, amongst extinct ursids in which

the zygomata are known, they are closest to those of the Florida Indarctos and

North American Arctodus simus (Kurtén 1967).

In all these respects the skull of A. africanum is reminiscent of that of the

lion, Panthera leo, although the latter is far smaller, and, of course, very different

in detail (Fig. 3). The resemblances between the skulls of Agriotherium and P.

leo are probably due to their sharing adaptations to a carnivorous way of life.

By curious contrast, the A. africanum skull also resembles that of the most

herbivorous of all bears, Ailuropoda, as well as that of Indarctos, another

supposed herbivore (Wolff 1978). This may be an indication that the actual

nature of the diet is less important than the requirement in certain Carnivora

Fig. 2.

ANNALS OF THE SOUTH AFRICAN MUSEUM

Dorsal, lateral and ventral views of the Langebaanweg Agriotherium skull, L45062.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 11

TABLE 1

Dimensions of Langebaanweg Agriotherium skull and mandible.

SKULL L45062 141404

Basallength . : ete : ; ; ; ‘ : ; 4 c. 381,0 —

Condylobasal length . c. 420,0 —

Palate length (posterior alveolar ‘margin of Ps to posterior palatine

incisure) ; ak # Ay Sets : : : : : ; ‘ c. 165,0 —

Zygomatic width . : . i. : ; : , ¢ : c. 305,0 —

Rostral width (over _C’ C's). ‘ 3 ; 5 f ‘ ; ; c. 118,0 —

Width over M?’s_. ‘ : ite Re : ; F ; ; ; , 135,0 —

Interorbital width . : ¢ . : ; : : : : F c. 125,0 —

Width over postorbital processes , OPER. ona oP ROT Se Ss c. 150,0 —

Occiput width at base of mastoid processes eg see Woy eees cok icbic Bens c. 130,0 —

Condylar width veo, i Gt BES PRR aa ey ee c. 90,0 —

C-M? length at alveolar margin : : es : 5 iS Dek : 149.0 154,0

P=M* length at alveolar margin . -. . . . . +. . . 83,5 82,5

MANDIBLE 145062 145114 50003 150004

Length (C to condyle) . ; : : ; j 296,0 — = —

Height of ascending ramus . ; ; : , 142,0 = = =

Transverse diameter of condyle . : : : 71,0 = os ==

Depth below M, . é : : g : : 69,0 68,5 c. 74,0 —_

Breadth below M, . : 4 : d : : 26,9 26,8 24,7 —

Depth at diastema . 3 5 : 3 : 61,0 66,0 65,3 59,4

C—M,; length at alveolar margin ; : : F 174,0 c.190,0 c.190,0 —

P,-M; length at alveolar margin . : F : 104,0 =. 117,0 111,0 —

for unusually powerful jaw musculature, with consequent similar modification

of the masticatory apparatus.

There are also some similarities between the skulls of A. africanum and the

polar bear, Thalarctos maritimus, which is the most carnivorous of living

Ursinae. Like A. africanum, Thalarctos also has a relatively broad snout and a

relatively straight dorsal profile in lateral view. These, and other, resemblances

between Thalarctos and A. africanum will be discussed again later.

Although Thalarctos has a relatively broad snout, it is simply a modifica-

tion of the ‘long-faced’ ursine condition, and it is easily distinguished from

‘short-faced’ Tremarctinae, which in turn bear a greater resemblance to Agrio-

therium and Indarctos, and, amongst the latter, particularly the Florida specimen.

The relatively short and broad tremarctine snouts are, however, ursine-like

in having the posterior palatine incisure well posterior of the M?’s (see Kurtén

1966, 1967). By contrast, in A. africanum the posterior ends of the M?’s are in

line with this incisure. The situation in Jndarctos is similar (e.g. I. atticus from

Samos—Helbing 1932; Thenius 1959), except that in this instance the M?’s

project slightly more posteriorly because, unlike these teeth in Agriotherium,

they have a talon and are more anteroposteriorly elongated. Ailuropoda is

virtually identical to Indarctos in this respect. The situation of the posterior

palatine incisure relative to the M?’s sets Agriotherium, Indarctos and Ailuropoda

apart from all other later ursids.

It is worth noting here that Davis (1964: 50) believed that the lengthened

palate in Ursus relative to that in Ailuropoda ‘is an illusion created by the large

Fig. 3.

ANNALS OF THE SOUTH AFRICAN MUSEUM

Dorsal views of skulls. A. Thalarctos. B. Ailuropoda. C. The Langebaanweg

Agriotherium. D. Panthera leo. A-B after Gregory (1936, figs 13-14).

——

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 13

teeth of the latter’ and that relative ‘to the anterior end of the braincase, the

palate actually extends farther posteriorly in the panda.’ This is true, but Davis

makes no mention here of the great shortening of the posterior parts of the

Ailuropoda skull, which undoubtedly has an effect on the position of the palate

relative to the braincase.

The palate of L45062 is otherwise unremarkable, except that it lacks the

minor posterior palatine foramen which is present in all other later ursids.

In addition, it has a relatively small anterior median palatine foramen. In the

latter respect it resembles Indarctos and Ailuropoda, and differs from the Ursinae

and Tremarctinae. The minor posterior palatine foramen has apparently been

incorporated with the major one, probably as a result of shortening of the palate.

A vestige of the minor foramen is indicated on the posterior part of the major

foramen by a shelf of bone projecting from the lateral wall of the latter. Earlier

in the history of Agriotherium this shelf may well have extended to the medial

walls of the major foramen, thus forming a separate minor foramen posteriorly.

The infraorbital foramen is situated above the posterior part of P+, as in

Indarctos. This foramen is more posteriorly situated in Ursinae and some

Tremarctinae, but is more anteriorly situated in Ailuropoda. It is relatively

small compared with that in other Ursidae, and in those A. africanum specimens

in which it is preserved (L2045, L45062, L41404) it is in the form of a single

opening. In other ursids, such as the A. sivalense skull and some tremarctines,

there may be more than one opening. The latter condition is probably of no

great significance, since multiple infraorbital foramina have been observed in

other carnivores (e.g. Proteles cristatus).

There is a marked concavity of the maxilla above the reduced anterior

premolars, anteroventrally from the infraorbital foramen, which is caused by

the expansion of the maxilla over the massive root of the canine and over the

roots of the posterior cheek teeth. In anterior view the bulges over the canine

roots completely obscure the infraorbital foramina. Amongst the Ursinae a

similar tendency was observed in Thalarctos, although in this instance the

infraorbital foramina were only partly obscured. Wolff’s (1978: 2) description

of the snout of the Florida Jndarctos shows that it is essentially similar to

A. africanum. The latter is distinct only in having features such as the bulge

over the canine roots exaggerated, apparently because of the larger size of the

Langebaanweg species. Wolff unfortunately provided few measurements of the

Florida skull, and consequently most dimensions must be estimated from

illustrations.

The nasal aperture of L45062 does not recede as markedly towards the

nasals as it does in the Ursinae, and A. africanum resembles Indarctos and the

Tremarctinae in this respect. This feature is evidently due to relative shortening

of the snout.

Observations on the frontal region of L45062 are omitted, since there was

severe fragmentation of this part of the skull and the restoration is not neces-

sarily accurate. For example, the orbit appears smaller and the frontals more

14 ANNALS OF THE SOUTH AFRICAN MUSEUM

inflated than in the skull of A. sivalense, in which this region is well preserved

and therefore certainly accurate. It is nevertheless clear that the post-orbital

processes of A. africanum are relatively less prominent than those of Ursinae,

and the former is probably more like Jndarctos (I. atticus) and Ailuropoda in

this respect.

In his description of the zygomatic arches of Ailuropoda, Davis (1964: 47)

stated that in dorsal view they ‘form nearly a perfect circle, compared with the

triangular outline in Ursus and other carnivores’. In A. africanum, which has

exceedingly stout zygomata, their shape approaches that in Ailuropoda. The

glenoid fossa and postglenoid process are also large, and in the postglenoid

region a wide shelf of bone extends over the external auditory meatus, linking

the zygomatic arch with the nuchal crest. The situation in Jndarctos and Ailuro-

poda is similar, although in the latter the shelf of bone is much shorter, owing

to the anteroposterior compression of the basicranial region in Ailuropoda. The

shelf of bone is less well developed in the Ursinae.

Amongst fossil specimens the zygomata which most closely resemble those

of L45062 belong to the Florida /ndarctos, and once again Wolff’s (1978: 2)

description also applies to the Langebaanweg skull. The latter differs in

apparently having zygomata of larger size, and in being more strongly arched

and thus more Ailuropoda-like.

The sagittal crest of L45062 is extremely well developed, reaching a height

of at least 50 mm, and, as in the Florida Indarctos (Wolff 1978: 2, 3), it has

‘a very conspicuous cleft between the parietals’. It is, however, slightly deeper

(up to 7 mm) and possibly longer (at least 70 mm) than that of the Florida skull.

A similar cleft is found in the sagittal crest of Ailuropoda (Davis 1964).

According to Wolff (1978: 3), ‘several large, rather irregularly sized and

positioned nutrient foramina appear on either side of the parietals just above

the temporal shelves near the posterior of the skull in several agriotheriine

specimens’, including the skull described by him. There are two such foramina

in L41404, one on either side of the sagittal crest, immediately adjacent to it

and close to its posterior limit. The same applies to L45062, but since this

region of the skull is incomplete, it is possible that there were more than two

such foramina.

The large size of the zygomata and enormous sagittal crest, together with

various rugosities for muscle attachments similar to those of the Florida Jndarc-

tos (Wolff 1978), indicate that the masticatory musculature of A. africanum

was exceptionally powerful, and in keeping with the massive canines and

posterior cheek teeth of this species.

The basicranial regions of L45062, L41404, and L45137 are unfortunately

poorly preserved and incomplete, although sufficient remains to show that they

are ursid-like. Amongst the living Ursidae there are similarities to both Ursinae

and Ailuropoda, while of described fossil specimens they closely resemble, and

in several respects are indistinguishable from, the Samos J. atticus (Thenius

1949, 1959) and the Florida Indarctos (Wolff 1978).

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 1S

As in Ailuropoda and Indarctos, the foramen rotundum and orbital fissure

form a single opening in both L45062 and L41404. There is a well-developed

horizontal division separating them a short distance posterior to the common

opening as in the Florida Jndarctos. In this respect Ailuropoda differs in that

there is at most ‘a paper-thin partition separating them’ (Davis 1964: 49).

A. africanum is also like Ailuropoda and Indarctos, and different from most,

if not all other Ursidae, in lacking an alisphenoid canal. In addition, A. afri-

canum, Indarctos, and Ailuropoda are similar in that the medial edge of the

glenoid fossa is closer to the foramen rotundum than is the case in the Ursinae,

evidently because of the greater transverse length of this fossa in these three

taxa.

The same applies in the case of the foramen ovale, which is preserved in

L45062, L41404, and L45137. This foramen opens opposite the posterior wall

of the glenoid fossa in these specimens, and they are similar to the Samos

I. atticus, and probably also the Florida Indarctos, in this respect. In the Ursinae

the foramen ovale is more posteriorly situated, while in Ailuropoda it is further

forward.

The postglenoid foramen in A. africanum is situated between the external

auditory meatus and the medial edge of the postglenoid process, but is closer

to the latter than is the case with living ursids. Amongst the latter, Ailuropoda

is the most distinct, since the postglenoid foramen is ‘more laterally situated

than in Ursus’ (Davis 1964: 52). The situation in the Florida Jndarctos and

I. atticus is similar to that in A. africanum, although the specimen described by

Thenius (1949) differs in having a double opening.

The external auditory meatus is incomplete in L45062, but it is evidently

similar in position and orientation to that in the Florida /ndarctos.

The anterior (squamosal) part of the mastoid process of L41404 is complete.

It had not yet fused to the posterior (periotic) portion of this process, nor to

the bone which caps this process. In size and orientation it is apparently similar

to the mastoid process of the Florida Jndarctos, and appears to have been more

laterally directed than that of later ursids. Both the mastoid and paroccipital

processes of L45062 are lost. The relative position of the bases of these processes

differs from that in later ursids in that the base of the mastoid process is only

slightly more laterally situated than that of the paroccipital process. The

mastoid process of L45062 apparently differed from that of L41404 in being

orientated ventrally.

Part of the lateral walls of the stylomastoid foramina, and the posterior

margins of the posterior lacerate foramina, are preserved in L45062. Their

positions relative to one another are as in the Ursinae and Jndarctos. The same

applies to the hypoglossal foramina and the anterior lacerate foramina.

Little of the bulla of L45062 remains. The posterior parts appear to have

been more inflated than in Ursus arctos, and in this respect the bulla may have

been more Jndarctos-like. In overall size the bullae of A. africanum and a

European U. arctos (ZM39056) were apparently similar, which is surprising in

16 ANNALS OF THE SOUTH AFRICAN MUSEUM

view of the very much larger size of the A. africanum skull. In this respect

A, africanum is also Indarctos-like. Ailuropoda is remarkable in that ‘externally

there is no indication of a bulla’ (Davis 1964: 319, 320). .

This is but one of the peculiarities of the Ailuropoda skull, which has been

‘profoundly modified by the demands of mastication’ (Davis 1964: 46). In less

specialized ancestral forms the basicranial and other regions of the skull may

well have resembled their counterparts in Indarctos and Agriotherium more

closely than those in the Ursinae. The absence of the alisphenoid canal in

Indarctos, Agriotherium, and Ailuropoda is probably the single most important

basicranial character which distinguishes them from the Ursinae. This canal

is present in the Canidae and all other Ursidae (Wolff 1978), including the

Amphicyoninae (Ginsburg 1977; Hunt 1977) and, presumably, the

Hemicyoninae.

The occipital region of A. africanum is known from L45062 (ventral parts)

and L41404 (dorsal part and ventral part adjacent to, and including, the squa-

mosal part of the mastoid process). In posterior view it is relatively narrow and

steeply arched, rather like that of U. arctos and Indarctos. In L41404, which

represents a young adult with some sutures unfused, there is an indication of a

constriction above the mastoid process. In L45062, an older individual, the

nuchal crest is strongly developed, and instead of passing ventrally directly

on to the lateral side of the mastoid process, as in most living ursids, it is linked

anteriorly with the shelf of bone which projects posteriorly from the zygomatic

arch. The base of the mastoid process in L45062 is actually recessed beneath

this shelf of bone. In most of the available ursine comparative specimens, the

nuchal crest merges ventrally into a ridge of bone on the mastoid process,

which is more laterally situated than the shelf extending posteriorly from the

zygomatic arch. Only in Thalarctos does the arrangement approach that in

A. africanum. It is not clear from illustrations of J. atticus (Thenius 1959) and

the Florida Indarctos (Wolff 1978) which arrangement characterizes this genus.

The mandible is massive, its size being in keeping with that of the skull

(Figs 4-5). It is similar in shape to that of J. atticus and U. arctos, and is dis-

tinguished principally by the presence of a premasseteric fossa. This fossa is

deep in older individuals (e.g. L45062), but much less pronounced in immature

adults (e.g. L45114) (Fig. 33). The only living bear with a premasseteric fossa

is the South American spectacled bear, Tremarctos ornatus, while it is otherwise

known amongst Ursidae in extinct Tremarctinae, Hemicyoninae and other

species of Agriotherium. In tremarctines the masseteric and premasseteric fossae

are separated by a prominent ridge of bone, and the latter fossa is deep, exten-

sive and more or less circular in shape, with its limits clearly defined (e.g. see

Kurtén 1966, plates 8-9). In hemicyonines the ridge separating the two fossae

is not as prominent, the premasseteric fossa is less extensive in a vertical sense,

but more elongated anteroposteriorly, with the anterior end gradually merging

with the buccal surface of the mandible beneath M, (e.g. see Frick 1926, figs 2-3,

12).

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 17

Ee ee Kk Mt A A

Fig. 4. Buccal and dorsal views of Langebaanweg Agriotherium hemimandible, L45062.

The premasseteric fossa in Agriotherium is closer to the hemicyonine type,

and may even be virtually indistinguishable from it (e.g. the A. insigne specimen

figured by Viret (1939, fig. 6)). There are, however, Agriotherium specimens in

which this fossa does not extend as far anteriorly, terminating instead beneath

M, (e.g. the A. schneideri specimen figured by Frick (1926, fig. 36)). There is

at least one specimen assigned to Agriotherium in which this fossa is absent

(i.e. the A. palaeindicum specimen, GSI-D8, discussed by Pilgrim (1932)).

The latter is, however, one of the problematical intermediates between Jndarctos

and Agriotherium, which will be discussed in a later section of this report.

The premasseteric fossae in A. africanum specimens are generally similar

18 ANNALS OF THE SOUTH AFRICAN MUSEUM

ov ce eae

Fig. 5. Buccal and dorsal views of Langebaanweg Agriotherium hemimandible, L45114.

to that in the A. schneideri specimen mentioned above. In the four hemi-

mandibles belonging to three individuals in which the entire premasseteric fossa

is preserved, the anterior termination is beneath M,. These specimens apparently

differ from hemicyonines and other Agriotherium in having a less distinct ridge

of bone separating the masseteric and premasseteric fossae.

The mandible of A. africanum is otherwise distinguished only by the

presence of a distinct ‘chin’ in the symphyseal region, which contrasts with the

receding jaw-line in other ursids. The ‘chin’ is formed by an anteroventral

expansion of the symphysis, which enlarges the area of the symphysis and

presumably strengthened the connection between the two halves of the mandible.

With the exception of the nondescript and relatively unimportant [, P,

P? and P3, all the teeth of A. africanum are known from at least one complete

specimen still in position in a jaw. An incomplete I’ is represented in L45062,

while P! and P? are tentatively identified on the basis of isolated specimens.

Only P, and P, have not been identified, although they could be represented

amongst the six unidentified anterior premolars available. The best represented

of the teeth are the larger posterior cheek teeth (P{-M3), which are, fortunately,

the most informative in the dentition of Agriotherium (Figs 2, 6-7; Tables 2-3).

These teeth are represented by between six and ten specimens belonging to

between four and seven individuals.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

‘VOVIVT

‘uowsely JNOUS JO SMAIA [eIIUDA pue [Loong “| ‘ZIOSST “uewseIy AXe]PIXCU Wntayjolispy SaMueeqasueyT JO MOIA [eIUSA “W ‘9 ‘BLT

A I om

20 ANNALS OF THE SOUTH AFRICAN MUSEUM

nl eee

PUTA ALL

6 117 118 119 210 211 ;

112 1

Fig. 7. Ventral view of Langebaanweg Agriotherium maxillary fragment, L45137.

3 1

5 1

4 1

Apart from their large size, the incisors of A. africanum are unremarkable

and are little different from those of living Ursinae. Dimensions of some incisors

were given elsewhere (Hendey 1977, tables 2, 5). The I? is distinct in having two

lingual cusps projecting from the V-shaped cingulum posterior to the principal

(spatulate) cusp. The lateral accessory cusp is the smaller and the more anteriorly

situated, being little more than a small projection from the cingulum. The other

accessory cusp is much larger, covering much of the lower part of the lingual

surface, although it is still much smaller than the principal cusp (Fig. 8). In

older individuals (e.g. L45062) the three cusps develop horizontal wear facets

in a single plane. The [’ is similar to I*, but smaller, while the I°, which is the

largest of the incisors, is morphologically similar to its counterpart in Ursinae.

The lower incisors are similar to those of Ursus arctos and other ursines,

and are distinguished only by their larger size.

I sil +

mint

Fig. 8. Medial, posterior and lateral views of Lange-

baanweg Agriotherium I*, unnumbered Dump 8 specimen.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

6 6

rg ste

-6 7 —6S7

08st =: LOE

GG 0G

6EZ 6ST

LLC ae Oe

SiGe COS

G8. 20 1e

CGC. sar be

160 <Sil€

Sic Ge

8°LZ "2 6°67

qd qe

‘2

way = aU! yorl=7

DOSEN Oem neice auc cc enasice ees mite alae

-0'6z -9'8% -I'8z | O'%% -Z'IE 90% 07 ee ok OnUEN

SSeS SS ee ee a ee ee eee

Se Bee ae ee ee ee ee

ee eee ee ee ee Se " LLssy

BS iS Ct Gee a ieee Soe a eee

Ene the eS SSS ie ee

Ge nue Saee = OS Se ee

qd qe I q I UT NOZIYOH, wdsNoON

Ghbiege as

*y}00} Joddn wnisayjousp Somueeqosury Jo suoisuswiqd

Z Alavi,

ANNALS OF THE SOUTH AFRICAN MUSEUM

ioe |

sre 1% OP FOI patSe

BEG eS Clas LE

O97 = SEE

SEG RL

NGG ES ASITS

SUV) “<8 1c

Elie mar eses| nome 00 iy ei eer | — hil 0c 6.0e

Sal Ug

See i Se

i Ver

oo eve | see ree per | sor ov

in Se Bes

geo site | Sater 0c ier (sisi ree

out rt| 9'ez9 o'ez | ee Lot coe | ol vz"

gor zor] ov soz | ofc coc cise | chr BIZ

LOI SOI

q I

"We

S‘€E

€8 ZI = Ge 2

8 S‘ZI S6I 8°67

xe)

Re)

ee |

ao)

*4199} IOMO] Wniayjolusp SomueeqosuryT JO suOIsusUIG

€ A1avL

Nee paq

Nee poq

Nee paq

Nee poq

Nee paq

Nee peq

Nee poq

Nee paq

See peq

NOZINOH]

* sosuvy

sur

1860S

* L000ST

* 9000S

* $000ST

’ Orr0ST

* SrrosT

* 7000S

* €000S7

* plisrl

* €9S9bT

* C90SVT

~ T9STIT

wasWaAN

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 23

The canines are similarly remarkable only for their large size. Otherwise

they differ from the canines of Ursinae only in being less elongated

anteroposteriorly.

All the known anterior premolars (P} to P3) are small, low-crowned and

single-rooted teeth. The P, is slightly elongated anteroposteriorly, and the

crown is divided longitudinally by a crest in the enamel. This is evidently the

remnants of the anterior and posterior keels which have merged into a single

feature as a result of reduction of the principal cusp, and its disappearance as

an identifiable element in this tooth. The enamel on the buccal side of the crest

is smooth, while that on the lingual side is slightly rugose. The P® is similar,

but is more circular in outline, while the crest is convex buccally, with vestiges

of the principal cusp still evident. The tentatively identified P (L55017A) fits

the P+ alveolus of L45062 well and may even belong to this specimen. It is

similar to P,, but is slightly broader posteriorly, and with vestiges of the prin-

cipal cusp still evident.

The P, of A. africanum is much larger than the anterior premolars, and in

relative size and morphology it is not unlike its counterpart in the Ursinae.

It is a stout, double-rooted tooth, with a prominent principal cusp, and in

lateral view the anterior and posterior halves are almost mirror images of one

another. The tooth is broader posteriorly due to the presence of a postero-

lingual bulge in the cingular region. The anterior and posterior keels of the

principal cusp are well defined but rather blunt, and terminate ventrally on

cuspless horizontal shelves. The posterior shelf is fringed by a well-developed

cingulum. Much of the P, enamel is finely rugose, which is also the case in other

posterior cheek teeth of A. africanum. The apex of the principal cusp develops

a horizontal wear facet, which merges with an inclined facet on the posterior

keel.

The P* of A. africanum (Fig. 9) is an important tooth, since it was largely

on its characteristics that the species was identified (Hendey 1972). The P* of

the holotype (L2045) has been exhaustively described (Hendey 1972; Wolff

et al. 1973), while examples from the bed 3aN sample have been briefly discussed

(Hendey 1977). The P* of L2045 is unlike that of any previously described

Agriotherium because of the presence of a large accessory cusp antero-internally,

that is, the anterior end of the protocone lobe. It is now evident that this feature

also distinguishes it from the P*’s of the bed 3aN sample, in which this cusp may

be present but small (e.g. L45137), or very small (e.g. L45062), although it may

also be absent (e.g. L41404). The P* of the A. africanum holotype is thus atypical

of the species sample as a whole. It is, however, not necessarily an abnormal

specimen, but may well be a typical example from a population which predates,

and is therefore more primitive than that represented in bed 3aN.

The bed 3aN P*4 sample, which is comprised of nine specimens belonging

to six individuals, includes only one in which the antero-internal cusp is absent

(i.e. L41404). Consequently, this cusp can still be regarded as a characteristic

of the species. A similar situation does, however, exist in respect of the A. insigne

4 1

Fig. 9.

ANNALS OF THE SOUTH AFRICAN MUSEUM

AAU UT

Buccal, occlusal and lingual views of Langebaanweg Agriotherium P*’s.

A. L46074, from bed 3aN. B. L2045, from bed 3aS.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 25

from Montpellier in France. A specimen described by Gervais (1859) apparently

lacks an antero-internal cusp, but an undescribed specimen from the same

locality (NMB-—MP549) has a large antero-internal cusp on its P*. For this,

and other reasons, A. africanum should perhaps be regarded as a junior synonym

of A. insigne, a matter which will be pursued in a later section of this report.

There is apparently one feature of P* in which A. africanum may be unique.

In L2045 there is a small wear facet situated between the antero-internal cusp

and the apex of the protocone, which was interpreted as the vestiges of another

accessory cusp on the protocone lobe by Hendey (1972). Wolff et al. (1973)

dismissed this interpretation, and regarded this feature simply as a wear facet

caused by occlusion with M,. It is undoubtedly such occlusion which caused

the wear facet, but Wolff and his co-authors overlooked the fact that it is

impossible to get dentine exposed at the same level as enamel unless there had

previously been a small cusp in this position. The difference of opinion is con-

clusively resolved by the fact that the bed 3aN sample includes unworn examples

of this accessory ‘intermediate’ cusp. It has already been recorded elsewhere

(Hendey 1977: 114) that the P* of L41404, although lacking an antero-internal

cusp, does have ‘a small, more posteriorly situated cusp which apparently corres-

ponds to the “intermediate cusp” of the holotype’. Another specimen is now

known (L46074) in which both an antero-internal cusp and an ‘intermediate cusp’

are present and unworn. As with the antero-internal cusp, the ‘intermediate

cusp’ of L2045 must have been more prominent than that of any bed 3aN

specimen.

The buccal cusps of the A. africanum P* are unremarkable. All ten speci-

mens have prominent parastyles, which is characteristic of the genus and of

some advanced Jndarctos specimens (see p. 81). The P* has three roots, two

beneath the buccal cusps and one beneath the lingual ones. The M1 and M? of

A. africanum are similar in this respect, although the lingual root is larger in

these teeth.

The wear facets on P* were discussed by Hendey (1972) and Wolff et al.

(1973), and the only additional observation possible is that the parastyle develops

a crescentic facet which is inclined anterolingually.

The seven M,’s belonging to four A. africanum individuals are, in general,

similar to those of previously described Agriotherium (Fig. 10A). The trigonid

makes up the bulk of the tooth, with the protoconid being particularly large

and prominent. Although the shearing facets normally found in carnivore lower

carnassials are evident on the buccal surfaces of the paraconid and protoconid,

the apices of these cusps also develop horizontal facets. In addition, the single

anterior, and two divergent posterior keels of the protoconid may be obliterated

by inclined facets. Cingula are developed on both sides of the paraconid, with

that on the lingual side being more pronounced. There are usually only two

talonid cusps clearly developed, a prominent, rather bulbous one on the lingual

side, and a low, ridge-like one on the buccal side. The latter develops a hori-

zontal wear facet, while the large lingual cusp develops a posterobuccally

ANNALS OF THE SOUTH AFRICAN MUSEUM

6 1

Fig. 10. Lingual, occlusal and buccal views of Langebaanweg Agriotherium M,’s.

A. L50004, a male specimen of typical morphology. B. L50006, a female specimen with

vestigial ?metaconid.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA PA

inclined facet. There is a ridge of enamel posterior to this cusp which may take

the form of a low cusp, as in L50004.

There are two M,’s in the assemblage which deviate from the basic cusp

pattern described above. An isolated specimen, L50006, which is the smallest

of the M,’s, and which evidently belonged to a female, has a small additional

cusp situated between the protoconid and the large lingual talonid cusp (Fig.

10B). This specimen is also unusual in having only a lingual posterior keel on

the protoconid, the buccal one, which is usually less pronounced, being absent.

In addition, the ridge anterior to the buccal talonid cusp is directed towards

the additional cusp, rather than directly anteriorly as is usually the case. The

second unusual M, is L50446, which is distinct in having two small cusps

situated posterior to the large lingual talonid cusp. The possible significance

of these specimens will be discussed later (see pp. 83, 84).

Both M, and M, have two roots, a larger one supporting the trigonid and

the other supporting the talonid.

The M! of Agriotherium has the basic four-cusped pattern typical of all

later ursids, and this tooth of A. africanum is little or no different from its

counterparts in previously described Agriotherium. The paracone and metacone

are prominent conical cusps of similar size, which develop horizontal wear

facets on their apices, and inclined facets posterolingually. The protocone is a

ridge-like cusp directed anterobuccally at a slight angle to the anteroposterior

axis of the tooth. The hypocone is conical and less voluminous than the proto-

cone. The two lingual cusps are lower crowned than the buccal ones. There are

lingual and buccal cingula which are distinct largely because the enamel is

smooth, whereas that elsewhere tends to be rugose. The cingular region bulges

anterolingually, and this results in the length of the tooth measured over the

lingual cusps being close to the buccal (maximum) length. In addition, the

breadth measured over the two anterior cusps is comparable to the posterior

breadth measurement, and to the two length measurements. In other words,

the M! of A. africanum is more or less square in outline. Agriotherium is unique

amongst later ursids in this respect. In the Ursinae, Tremarctinae, Jndarctos

and Ursayus the M"’s are rectangular with lengths exceeding breadths, while in

Ailuropoda the situation is reversed and the M!? is broader than it is long.

The M, of A. africanum also has a basic four-cusped pattern, but in this

instance the two largest cusps (paraconid and protoconid) are situated

anteriorly, rather than buccally as in M! (Fig. 11A). There are, however, three

teeth belonging to at least two individuals which deviate from this pattern.

They are L45115 and L46563 (Figs. 5, 11B) which probably belong to one

individual, and L50007, which belongs to a smaller individual, probably a

female. In these specimens the anterolingual cusp is reduced or absent. Reduc-

tion is evident in L50007 where the anterolingual cusp is situated closer to the

anterobuccal cusp, with its anterior and posterior keels directed accordingly.

This gives the impression that it is merging with the anterobuccal cusp, rather

than simply reducing in its usual position. In L45114/L46563 the anterolingual

28 ANNALS OF THE SOUTH AFRICAN MUSEUM

A B

gy nu

Fig. 11. Occlusal and buccal views of Langebaanweg Agriotherium My,’s.

A. L50004, a specimen of typical morphology. B. L46563, a specimen lacking

the anterolingual cusp.

cusp is absent, and its posterior keel is linked to the apex of the anterobuccal

cusp. Only vestiges of the anterior keel remain. These three teeth are also

distinct in being relatively narrower than others in the M, sample. The M, cusps

develop horizontal wear facets, while there are also inclined facets antero-

buccally and on the buccal surface of the posterobuccal cusp.

The M? of A. africanum, like the P?, is an important tooth, and it is fortu-

nately well represented (Fig. 12). It is essentially similar to M? in its basic cusp

A B C

Fig. 12. Occlusal views of Langebaanweg Agriotherium M?’s. A. L47242 (reversed).

B. L48564. C. L47698.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 29

morphology. The paracone and metacone are the most prominent of the four

cusps, although in this instance the metacone is a little smaller than the para-

cone. The protocone is ridge-like, while the hypocone is more conical, although

it tends to be less distinctly developed than the other cusps. There is an expan-

sion of the lingual cingulum adjacent to the protocone, and sometimes also in

the posterolingual part of the tooth. The latter expansion is significant since it

represents the vestiges of the talon which was present in ancestors of Agrio-

therium (see page 87). As with the other molars, the cusps develop horizontal

wear facets, while there are also inclined facets developed anterolingually on

the paracone and metacone.

The M, of A. africanum is a relatively simple, single-rooted tooth, which

tends to be circular in occlusal view, with no distinct cusps developed (Fig. 13).

@ @

LUAU

114 115 1

UU ]

Fig. 13. A. Occlusal and lingual views of Langebaanweg Agriotherium M3, L50981.

B. Occlusal and posterior views of M;, L12561.

The occlusal surface is in the form of a shallow basin surrounded by a low

ridge of enamel. This ridge develops a horizontal wear facet, and the entire

occlusal surface would presumably be worn flat in older individuals. There is

also an inclined wear facet anterobuccally. This region of the tooth tends to be

slightly expanded, which emphasizes a posterior tapering which is best seen in

the specimen L50981.

More detailed comparisons between the teeth of A. africanum and other

taxa will be made in a later section of this report.

THE POSTCRANIAL SKELETON

The present study of A. africanum postcranial bones has been superficial,

largely because adequate modern comparative material was not available and

30 ANNALS OF THE SOUTH AFRICAN MUSEUM

waa

Fig. 14. Lateral and ventral views of Langebaanweg Agrio-

therium scapula, L50858. Arrow indicates postscapular fossa.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 31

because of the lack of direct access to relevant fossil material. Little has hitherto

been published on the postcranial skeleton of Agriotherium and other late

Tertiary ursids, and although meaningful interpretation of the evolutionary

and functional significance of A. africanum bones is no doubt possible, the

following account of them is essentially descriptive. In this account references

are made to bones of a male European Ursus arctos (ZM39056), a male Asian

Euarctos (ZM38805) and a female Helarctos (ZM36289), all of which are zoo

specimens. Reference is also made to the limb bones of Ailuropoda as described

by Davis (1964).

In general, the bones of A. africanum are larger and more stoutly pro-

portioned than those of the U. arctos comparative specimen, but they are

essentially similar morphologically.

The available vertebrae of A. africanum are all incomplete, and most are

very fragmentary. Except for their larger size, they are similar to those of the

available comparative specimens and Ailuropoda. No ribs definitely identified

with A. africanum are known.

The only identified A. africanum scapulae, L50857/8, may belong to one

individual, although L50858 differs in showing indications of mild osteo-

arthritis (Fig. 14). Both specimens lack most of the blades. The glenoid fossa

is anteroposteriorly elongated and tapers anteriorly, much like those in the

ursine comparative specimens and the Ai/uropoda specimen illustrated by

Davis (1964, fig. 46). The preserved parts of the blades of A. africanum scapulae

are also similar to those of ursines and Ailuropoda, except for one marked

difference which may be of great functional significance.

Davis (1964: 91) has recorded that there are differences in the nature of

the postscapular fossae in Ailuropoda and U. arctos, and, judging from the

available comparative specimens, there is appreciable variation of this fossa in

Ursinae. It is only the lower parts of this fossa which can be observed in

L50857/8, and these specimens are distinct in having the inferior scapula spine

terminating on the lateral surface of the blade 50 to 60 mm above the glenoid

fossa, instead of terminating posteriorly at the glenoid fossa. Consequently,

the postscapular fossa in A. africanum is confined to the lateral surface of the

scapula, whereas the lower part of this fossa is on the medial surface in Ailuro-

poda, U. arctos and Helarctos, and posteromedially in Asian Euarctos.

According to Davis (1964: 91, 173) the postscapular fossa ‘lodges the

subscapularis minor muscle’ which is the main ‘medial rotator of the arm’.

The functional significance of its distinct orientation in A. africanum is not

known.

No complete humerii of A. africanum are preserved, the best available

specimen being L45063, in which the proximal end is detached and slightly

crushed, and part of the proximal part of the shaft is lost (Fig. 15, Table 4). The

humerus is otherwise known from several fragmentary specimens, mainly distal

ends. All the important features of the humerus, except overall length, can be

observed. It is one of the many A. africanum bones in which available speci-

SV. ANNALS OF THE SOUTH AFRICAN MUSEUM

Fig. 15. Posterior, lateral and anterior views of Langebaanweg Agriotherium humerus, L45063.

mens exhibit appreciable size variation, larger specimens presumably belonging

to males and smaller ones to females. The latter include L45063.

In its basic morphology the humerus of A. africanum is similar to those of

Ursinae. Proximally it is similar to corresponding parts of ursine humerii in

all observable respects, but distally there are some differences, although like

the Ursinae, and unlike Tremarctinae and Ailuropoda, it lacks the entepicondylar

foramen. However, vestiges of the bar of bone enclosing this foramen, in the

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 33

TABLE 4

Dimensions of Langebaanweg Agriotherium humerii.

140040 145063

hl 7) 67,0

121,5 98,5

92,7 76,4

Max. ant.—post. diam., distal end

Max. transy. diam., distal end .

Max. transy. diam., distal articulation

form of a rugosity at its proximal termination, are more obvious in A. africanum

than Ursinae.

The olecranon fossa is deep and relatively narrower than its counterparts

in available ursine humerii, although it is closer to the Ursinae than Ailuropoda

in this respect (see Davis 1964: 95, fig. 49). There is a relatively greater antero-

posterior development of the A. africanum humerus distally, although once

again it is closer to Ursinae than Ailuropoda. The medial epicondyle is much

less prominent than in ursines, although the lateral epicondyle is similarly

developed.

The lateral epicondylar ridge is more constricted above the lateral epi-

condyle than in ursines. This ridge has a relatively greater length than those of

U. arctos and Helarctos, but that of Asian Evarctos is comparable in this respect.

This is a reflection of the relatively greater elongation of the humerii in A. afri-

canum and Euarctos.

The proximal termination of the lateral epicondylar ridge is more or less

opposite the point of convergence of the deltoid and pectoral ridges (the deltoid

tuberosity), as in ursines. The pectoral ridge is much like that of ursines, but

the deltoid ridge is more prominently developed, and proximally it is more

anteriorly situated. Consequently, the area between these two ridges faces

anteriorly over a greater distance than in the ursines. This is the area of insertion

of the cephalohumeral muscle, which is the chief extensor of the foreleg (Davis

1964: 95, 167). The functional significance of the more anterior insertion, and

apparently greater development of this muscle, is not known.

Some of the characteristics of the ulna of A. africanum have already been

discussed elsewhere (Wolff et al. 1973), but more and better specimens are now

available (Fig. 16, Table 5). Once again the most complete specimen is L45063,

which belongs to a smaller individual than the previously described specimen

(L2154). Apart from the size difference, these two specimens are similar in all

observable respects. The A. africanum ulna resembles those of available ursine

specimens, and other ursids, including the /ndarctos atticus specimen from

Samos described by Pilgrim (1931). The proportions of L45063 are similar to

those of the U. arctos comparative specimen.

In lateral view the A. africanum olecranon does not project as markedly

as in ursines, and its medial tapering is usually less pronounced. The olecranon

is, however, very broad transversely. The area for insertion of the triceps and

flexor carpi ulnaris muscles is therefore large, although that for the insertion of

the latter is less knob-like. In the Ursinae and Ailuropoda the area for insertion

of the most proximal part of the anconeus muscle is prominently developed (see

34 ANNALS OF THE SOUTH AFRICAN MUSEUM

badd

A B

Fig. 16. A. Medial view of Langebaanweg Agriotherium ulna, L46076. B. Anterior and

lateral views of ulna, L45063.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

0°0z = a

OE oe ae

8°89 a T'¥6

0°Or 09S 9°09

Spl Fo S‘P8

L‘6E€ — L‘0v

C8 y‘8E L‘0r

9°77 ae 8°Sz

0'9Ir ar re

£90SV7T PSIZIT = (7290S

S‘L8

09S

6°98

Orr

S'97

9LO9V'I

S‘Iv

9 TP

9°17

€8eclT

vrOOvVT 8080S

¢ alav I,

9°C ae

Liv >>

a 0°8s

ae 8°€6

ra 9'Sh

a OLY

cae 0°87

SO80SsT} £9L0ST

‘OUl[N WNLaYyjolisp SoMueeqosuLyT JO SUOISUSWIIG

1wsrA = Ca)

* ssod0i1d projAjs JO “wieIp y104S

ssos0i1d projAjs Jo ‘wieip 3u0T

* —_ ssed0i1d prouo.109 je wep ysod—juy

* —- YO}OU JeUNIIWOs ye ‘WeIpP ‘ysod—juYy

ssoooid jeouosue je ‘wieip jsod—juy

; * yoJOU IeUN]IWes Jo “WRId

p * yojOU [eIper

ye ssoooid plouol0d jo ‘“WieIp “AsueIy,

ssosoid jeouooue JO “WIRIpP “ASULIT,

: } ; Yj3Ue] [[eI0AO

36 ANNALS OF THE SOUTH AFRICAN MUSEUM

Davis 1964: fig. 50), but is less so in A. africanum, especially female specimens

such as L45063. Once again, the functional significance of these differences is

not known.

The semilunar notch differs from those of ursines and Ailuropoda in some

respects. In lateral view it is almost perfectly semicircular, with the inner

borders of the notch not divergent at their extremities as in ursines, and not

showing the beginnings of convergence as in Ailuropoda (see Davis 1964,

figs 50-51). In anterior view the dorsal part is nearly parallel-sided, rather than

sharply tapering as in ursines. The anconeal process of A. africanum is therefore

relatively broad. This, and the relatively narrow olecranon fossa of the humerus,

are probably directly related features. When the humerus and ulna of A. afri-

canum are articulated and fully extended the olecranon fossa is nearly com-

pletely filled by the anconeal process, and little lateral movement of the ulna is

possible. In the ursines only about half the olecranon fossa is filled and appreci-

able lateral movement is possible. Much the same evidently applies in the case

of Ailuropoda, which has an even wider olecranon fossa than ursines, while

Davis (1964: 96) noted that there was no protection against lateral shifting of

the elbow joint. The significance of these contrasting situations is not known.

The shaft of the’ A. africanum ulna is very stout, and although the areas

for attachment of the brachialis tendon and interosseous ligament vary both in

size and form, they are very pronounced features.

Distally the radial articular facet is relatively, and sometimes absolutely,

smaller than its counterpart in ursines. Medially between this facet and the

styloid process is a deep and almost circular depression, which is directly in

line with the prominent ridge on the shaft separating the areas of insertion of

the pronator quadratus and the distal part of the flexor digitorum profundus 5

muscles. This depression is absent in available ursine specimens, and apparently

also in Ailuropoda (Davis 1964, fig. 50). Its significance is not known.

The radius of A. africanum is represented by several proximal and distal

fragments, but only one that is complete (L40044), while another has been

restored (L45063) (Fig. 17, Table 6). The latter belongs to a female and L40044

to a male. These two specimens differ only in size. They are typically ursid in

their characteristics, the proximal end being particularly distinctive, and L40044

is very similar in its proportions to the J. atticus specimen described by Pilgrim

(1931). The latter specimen is, however, distinct in having a broad groove on

the anterior surface towards the distal end. Pilgrim (1931: 27) thought this

noteworthy since ‘a similar structure exists in a corresponding position in the

radius referred by Falconer to Agriotherium . . . sivalense’. According to Pilgrim

it is also present in Amphicyon, although it is evidently not characteristic of all

amphicyonines (see Ginsburg 1977, fig. 25). Since the radius of A. africanum is

otherwise unremarkable, it is not described in detail.

With one exception, all the carpal bones characteristically present in

Ursidae are represented by at least one, and as many as nine, complete speci-

mens (Fig. 18, Table 7). The exception is the radial sesamoid, which was evi-

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 37

LLU

fi ; \! | + I 2 1 iafe

oon

(ieee LF

6 L

hulu

lat

Fig. 17. A-B. Medial views of Langebaanweg Agriotherium radii. A. 145063. B. L40044.

C. Proximal view of L40044.

38 ANNALS OF THE SOUTH AFRICAN MUSEUM

Hui

1 sil

tui i =

wut

i-)

{

H I

Fig. 18. Carpal bones of Langebaanweg Agriotherium. A—G. L45063. H. 146134. I. L50767

(reversed). A. Proximal and dorsal views of scapholunar. B. Medial view of cuneiform.

C. Medial and dorsal views of unciform. D. Anterior view of pisiform. E. Proximal, dorsal

and lateral views of magnum. F. Proximal view of trapezoid. G. Lateral view of trapezium.

H.-I. Specimens illustrating size range of scapholunars.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

e90SV'1

SET ODE ¥H'8Z

THE O6E O'8E

661 90% 677

6€€ OLE SOE

y8rp OS ZS

£00071

C90SV1

SWNOAISIT

vLy

L’vEé

918081

Si8S a es OOS Oe Epes 9c an ne

ViL SiC MECC sacOG) OCC Ge YC GIES cs

CPEeaS St OSES 686 OTP 6 eV L8G

SNNNOVW

‘WIRIP JOYS 4SIq

‘WIRIP SUC] “}SIG

‘WIRIP J1OYS “XOIg

‘WIRIP SUC] *XOIg

YSU] [[V10AO

SUVYNOATOHdVOS

“s[edivo wnideyjolusp SsoMueeqosury] JO SUOISUSWIG

L a1avL,

L‘€9 == — S‘OL = BS Le

Str 9 = =a 0°6r sa 53 mag

rsp €‘0s 0'6r 9 L‘0S 8°7S (als y'9S

‘OE S‘LE ‘BE ‘9 oI = 8‘€V Ure

O'SHE eS =e o'r0r =a ae: ae

€90SVT 9C8EET 669471 vroOVI 790SPT LPL8VI 9080571

‘UIpel WNayjolisp Somueeqosuey] JO SUOISUSWIG

9 alav

“WILIP [PIJUSAOSIOP “xe

‘WIPIp “AsUPI}

‘WIP ‘sod—jue

"WUPIP [e1}USAOSIOp

“WIeIp “AsuUPIy

Jo0vj [eIpel JO ‘WIeIp -jsod—jue

: : “WIeIp ysod—"jue

pus [eIsIp “weip ‘Asued

puod [eISIp “weIp jsod—jue

pus jewixoid “wep ‘Asuen

puo jeuxoid “werp “jsod—yue

“Xe

“XPJA

“XP

“XE

“XP

“Xe

“XP

“XE

Yysug] [[eI9AO

40 ANNALS OF THE SOUTH AFRICAN MUSEUM

dently absent in A. africanum, since the scapholunar lacks the radial sesamoid

facet. In this respect A. africanum is very different from Ailuropoda, in which

the radial sesamoid is better developed than in any other arctoid carnivore

(Davis 1964: 99, 100). A small radial sesamoid is apparently usually, or always,

present in Ursinae, and is larger in Tremarctinae. Otherwise the carpals of

A. africanum are essentially similar to their counterparts in Ursinae.

The largest carpal, the scapholunar, is also the best represented. As with

other A. africanum bones, the scapholunars exhibit appreciable size differences,

the smallest (50767) being only about two-thirds the size of the largest (L46134)

(Fig. 18H-I). This is a slightly greater size difference than that observed in the

scapholunars of North American Tremarctos floridanus, an ursid of similar

overall size (Kurtén 1966, table 19), although in both instances sample sizes |

are small. There are no significant morphological differences between the ~

smallest and largest scapholunars, and they differ in only minor respects from

those of ursines. In the latter the cuneiform facet tends to be clearly distinct

from the unciform facet, but in A. africanum they merge and are distinguishable

only because the cuneiform facet is flattened and the unciform facet is concave.

In some specimens (e.g. L48021) the proximal (radial) articular surface is in

contact anteriorly with the articular facet of the trapezium, and in this respect

A. africanum is similar to Ailuropoda (see Davis 1964: 99). The proximal articular

surface of the A. africanum scapholunar is also Ailuropoda-like, and different

from Ursinae, in lacking the lateral depression which receives ‘the saddle on

the distal end of the radius’ (Davis 1964: 99).

The cuneiform of A. africanum is morphologically similar to that of

U. arctos, except that the scapholunar facet is more elongated. In addition, this

bone is relatively more flattened than that of U. arctos.

Such minor differences in morphology and proportions also distinguish

other A. africanum carpals from their U. arctos counterparts. For example,

in the A. africanum unciform, the magnum and scapholunar facets are not

confluent as in U. arctos, while the pisiform is a considerably stouter bone.

At least some of the distinctive features in A. africanum carpals may be due to

the large size of the species, but their possible significance in other respects was

not investigated.

Of the metacarpals of A. africanum only the second is not represented by

-a complete specimen, although there is one which lacks only the distal end

(L33160). All the metacarpals are similar to their counterparts in ursines in

terms of morphology, but are relatively more massive (Fig. 19, Table 8). In

spite of this, they are not necessarily much longer than those of the available

U. arctos specimen. One of the complete metacarpals I (L45448) is in fact

similar in length to that of the U. arctos comparative specimen, although it

has a much stouter proximal end and shaft. The A. africanum metacarpals

appear to be readily distinguishable from those of Ailuropoda, which are ‘short

and stout, relatively considerably shorter than in [other bears] of comparable

size’ (Davis 1964: 100).

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 41

<i (pa @

¢|

Fig. 19. Proximal, anterior and lateral or medial views of Langebaanweg Agriotherium meta-

carpals. A. I, L50775. B. III, L25862. C. IV, L33828. D. V, L45063.

ANNALS OF THE SOUTH AFRICAN MUSEUM

c90SV'T

8°P7Z

Sve

0°77

T9E

9EIT

8660¢T

Ill

6 IZI

C98STT

L‘€Z = O'EZ 2 0°97

6 IZ — VCC Ob

List = SEE PEG WISE

66¢ OLE €CE pa eve €9€ SLE

ZT90Or = = LLIT vOcl OT

le?

OST SET

S9E SLE

mo ie

3 @

[o-e) _

os [ony

(>) So

“s[edivovjoul WnlayjoUsp SoeMmueeqosueryT JO SUOIsSUOWIC

g ATEV IL,

Dene Iaae ‘4SIP “UWIRIpP “ASURI] “xeyAT

* worjelnoy1e

‘\SIp “WIeIp “jsod—jue “xey

* puo -xoid “uIRIp “AsueI) “xe

ies) poids “WeIp ‘jsod—jue -xepy

YI3Ug] [[PI0AO

Teorey THe 4SIp “WeIp “AsuRI} “xe/[

* uone[nonie

*1SIP “WeIp jsod—jue -xey

* pus -xoid “‘WieIp ‘Asuei} “xepy

ee) Sorel” ‘WeIp “jsod—jue -xey

: Yysu9gy [[e10AO

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 43

No complete innominate of A. africanum is known, the best specimen

(L33341) comprising only the posterior part of the ilium and the acetabular

region. The latter is similar to that of the U. arctos comparative specimen, and

is distinguished only by its slightly larger size. There is, however, a greater

dorsoventral constriction of the ilium anterior to the acetabulum, while the

dorsoventral diameter of the posterior parts of the sacroiliac articulation is

actually less than in the comparative specimen. A. africanum is more like

Ailuropoda in this respect (see Davis 1964, fig. 59). The iliopectineal eminence is

less prominent than in U. arctos, and in this respect it is also Ailuropoda-like.

No other important features are observable in L33341, but the fact that it is

more like Ailuropoda than U. arctos in at least two respects may be significant.

The pelvis of the former is very different from those of other bears (Davis 1964:

113), and it is possible that that of A. africanum was equally distinctive.

The femur of A. africanum is represented by several fragmentary specimens,

and a complete one (L40002) belonging to an aged and arthritic female (Fig. 20,

Table 9). The latter specimen is considerably smaller and more slender than

corresponding parts of male specimens, but is morphologically similar to them.

It differs from the femur of the U. arctos comparative specimen in being slightly

longer and relatively more slender.

The head of the femur is hemispherical, as in other ursids, while the neck

is distinct and slightly longer than in U. arctos, but similar to that of Asian

Euarctos and Ailuropoda. The greater trochanter is lower than the head and it

is similar to the U. arctos femur in this respect. The gluteal tuberosity is very

prominent in L40002 and terminates well below the level of the lesser trochanter,

which is also prominent. The area of attachment of the quadratus femoris

muscle between the lesser trochanter and gluteal tuberosity is well marked, as

is the area of attachment of the adductor muscle, which extends about three-

quarters the length of the very straight shaft. The distal end of the femur is

essentially similar to that of U. arctos.

The femur of A. africanum is very different from that identified with

A. sivalense by Lydekker (1884, pl. 29, fig. 1). This very curious specimen may

be pathological.

The tibia of A. africanum, of which only one complete specimen (L45062)

is known, is also very variable in size (Fig. 21, Table 9). As with the femur,

L40002, the tibia is longer than that of the U. arctos comparative specimen, but

in this instance it is also much stouter, evidently because L45062 belongs to a

male. The proximal end of the tibia differs most markedly from those of ursines

in having the tibial tuberosity and crest more prominent. This applies even in

the case of specimens belonging to females (e.g. L50842). In addition, the

proximal articular facets are relatively longer anteroposteriorly.

In Ailuropoda and ursines the lateral edge of the tibia shaft viewed anteriorly

is bowed, with the tibial crest paralleling the proximal curvature. This, together

with prominent lateral projections at the proximal and distal ends, ‘increases

the interosseous space between the tibia and fibula, and the total width across

Fig.

isabel

iat

sli

ANNALS OF THE SOUTH AFRICAN MUSEUM

A-B. Anterior and posterior views of Langebaanweg Agriotherium femora.

A. L40002. B. L40128. C. Medial view of L40002.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

Fig. 21. Anterior and medial views of Langebaanweg Agriotherium tibia, L45062.

ANNALS OF THE SOUTH AFRICAN MUSEUM

pUo [eISIp “UIP “AsUeN “xe

a, = = OS 17S * pus [eISIp “werp ysod—jue ‘xeyy

ZT 016 ‘2 0°66 °2 LOOT S801 * puso jeurxoid “ureip *Asueq} “xeyy

pes L'S8 €‘L8 1°26 — pus jewrxoid “weip ysod—jue ‘xe

a ae a = = peoy jo ‘weip ysod—jue *xeyy

— ae x O'8LE = : ; Sug] [[e19AO

Cy80ST s8c00VI 6c00VI c90SVI § <€E80ST

/T80sT

6886V1 cOO0OVI 8zcl0rT

AVIAL], VUOWAT

“OEIQN PUL BIOWOJ wniayjoUsp Somueeqosuey JO suoIsuOWIG

ACGCKAD

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 47

Da INIU UH

INH

inn

Itt

Fig. 22. Tarsal bones of Langebaanweg Agriotherium, 145062. A. Proximal, lateral, ventral

and distal views of astragalus. B. Lateral and dorsal views of calcaneum. C. Dorsal view of

cuboid. D. Proximal view of navicular.

48 ANNALS OF THE SOUTH AFRICAN MUSEUM

the leg’ (Davis 1964: 115). The tibia of A. africanum differs in having the proxi-

mal and distal lateral projections less prominent, and the shaft less bowed,

which gives it a much straighter and more bilaterally symmetrical appearance

in anterior view. The distal articular facet, like the proximal ones, has a rela-

tively greater anteroposterior diameter than in ursines. The same applies to

the proximal and distal fibula facets, while the fibula itself is much like those of

ursines.

All the tarsal bones of A. africanum, except the mesocuneiform, ecto-

cuneiform and tibial sesamoid, are known from at least one, and as many as

eight specimens (Fig. 22, Table 10). The tibial sesamoid, like the radial sesamoid,

was probably absent in this species. As with the carpals, the tarsals are essentially

similar to their counterparts in Ursinae, and they, too, exhibit appreciable

sexual dimorphism.

The astragalus is the best represented tarsal bone. It is distinguished

from those of ursines principally by a longer neck. The available ursine astragali

have a variably developed lip of bone projecting posteriorly from the base of

the tibial facet, and which is most prominent medially. This lip of bone is

absent in A. africanum, although in the specimens L45062 and L47533 there is

a ventrally projecting lip of bone in this position. Since it effectively inhibits

movement between astragalus and calcaneum, it is probably an abnormality

caused by osteo-arthritis.

The calcaneum of A. africanum is also ursine-like, differing principally in

having a relatively shorter and much stouter tuber calcis.

The remaining tarsal bones differ in only minor respects from their counter-

parts in U. arctos.

All the metatarsals of A. africanum are represented by at least one complete

specimen (Fig. 23, Table 11). They are in general much stouter but only slightly

longer than their counterparts in U. arctos. An exception is a small metatarsal I

(L50771), evidently that of a female, which is of similar length and which is

TABLE 10

Dimensions of Langebaanweg Agriotherium tarsals.

ASTRAGALI

145062 (R) 150765 147533 L50766 150770 147910

61,9 58,0

63,2 62,1 59,8 c. 56,5 51,2 43,7

Max. ant.—post. diam. .

Max. transy. diam.

Transv. diam. of tibial facet 44,9 c. 46,0 43,1 Shy 33,5 29,5

Transv. diam. of navicular facet 43,5 44,5 43,0 37,9 32,8 27,0

Max. dorsoventral diam. 40,4 — 37,2 35,3 28,8 27:5

CALCANEUM NAVICULARS

145062 L50773 145062 | L45062

48,7 43,3

38,7 38,2

20,5 17,6

. ant._post. diam. .

Max. transy. diam.

Max. dorsoventral diam. ;

. transy. diam. of tuber calcis

(R) = Right

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 49

Fig. 23. Proximal, anterior and lateral or medial views of Langebaanweg Agriotherium

metatarsals. A. I, L50787. B-E. II-V, L45062.

ANNALS OF THE SOUTH AFRICAN MUSEUM

a = TET CET : er ae

DONeNoIe [e\sIp “WeIp “ASURI} “Xe

= = OI~ L‘6L = A - = wvonrnonse [eisip “werp ysod—jue “xe

S*LZ ITE LOE OLZ €ZE : : * puso yecapxoid “UIRIP "ASUBI} “XE

ELT 9°67 SOE Ile VE ‘ Sei abe pice al “WieIp “jsod—jue “xe

= = S‘SII S‘6IT = é Pa ae al :

W13U9y [[eJeAO

BOIIy1 6c80ST 790SVE CLS8VI ZI80ST

coOSVI = HBOS

Vz TC * WONEfNose [e}SIp “UIEIP “ASUI} “Xe

y6l S07 uorefnonse [eysip “werp “jsod—yue “xe

SC (44 ates pus jeunxoid “urerp “Asue] “XE

Ive S‘9E

BES [Eanes ‘weIp ‘jsod—jue “xe

6 LOT 9OIT

: ysus] [[e19AO

SsOcOeT 90ST

Til

C90SV1

ELS8Vv1

ILLOSI 6 L8ELVI = L8 LOST

“S[esIe}e}OU wnidayjousy Zomueegosuey jo suorlsuauid

TI S1av.L

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 51

more slender than that of the U. arctos comparative specimen. Morphologically

there are no significant differences between the metatarsals of A. africanum and

U. arctos.

The patella, sesamoids and phalanges of A. africanum are also U. arctos-

like, although the Ist and 2nd phalanges of the former are relatively much

shorter and stouter (Fig. 24).

| Mn hn

Fig. 24. Dorsal and lateral or medial views of Langebaanweg Agriotherium phalanges, L45062,

possibly of one digit.

DISCUSSION

The overall impression gained from the study of the skull and postcranial

skeleton of A. africanum is of a large and heavily built animal which was unmis-

takably bear-like in its appearance (Fig. 25). It was undoubtedly plantigrade

since its postcranial bones exhibit most of the characteristics found by Ginsburg

(1961) to be indicative of this condition. The appreciable size variation observed

is to be expected of a bear of such large proportions, since marked sexual

dimorphism is characteristic of these animals (Kurtén 1955, 1966, 1967). Being

typical of large ursids in this respect, it is of interest to compare A. africanum

with better known species which belong in this category. They include Ursus

spelaeus, Tremarctos floridanus, and Arctodus simus.

Of these species A. africanum probably resembled the North American late

Pleistocene tremarctine, Arctodus simus, most closely. For example Kurtén

(1967: 49, fig. 28) found that the skull of this short-faced bear ‘shows a remark-

able convergence with the great cats’, which is also a feature of the skull of

A. africanum (see p. 9 and Fig. 3). In overall size the male skull of the latter

(L45062) is intermediate between those of female and male A. simus (see Table 1

herein and Kurtén 1967, table 5). In addition, the postcranial bones of A. afri-

canum are similar in proportions to those of A. simus, although they differ in

52 ANNALS OF THE SOUTH AFRICAN MUSEUM

Fig. 25. Reconstruction of Agriotherium, adapted from one of North American Arctodus simus

by J. Matternes Gin Guthrie 1972).

some morphological details. In actual size the bones of A. africanum males

compare closely with those of A. simus females (e.g. the Potter Creek Cave

sample—see Tables 4-11 herein and Kurtén 1967, tables 10-25).

Unfortunately the skeleton of A. africanum is less well represented than that

of A. simus and consequently cannot be analysed in as much detail. In view of

the marked individual size variation in A. africanum it may be misleading to

combine skeletal elements of different individuals in metric analyses. Never-

theless, this was done in several instances, one of which is presented here.

The isolated metatarsal I, L50787, which is evidently that of a male, was

combined with the metatarsals II to V of L45062, also a male, in order to com-

pare their relative lengths with the metatarsals of other ursids (Table 12). The

similarity to the Potter Creek Cave A. simus sample is striking.

On the other hand, the calcaneum length expressed as a percentage of the

longest metatarsal length of L45062 is 86,6, a figure which compares closely

with the 86,5 of an U. arctos sample, and which is considerably lower than the

figures for extinct tremarctines and U. spelaeus, which are over 100 (Kurtén

1966, table 36; 1967, table 27).

In spite of such deviations from the Arctodus pattern, A. africanum is like

this genus and different from other later ursids in having relatively long legs.

This characteristic, together with its specialized skull and dentition, is a highly

significant departure from the typical ursid condition. Kurtén (1967: 50)

interpreted A. simus as ‘a predominantly carnivorous form’, which may have

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 53

TABLE 12

Relative lengths of the metatarsals of some bears expressed as a

percentage of the length of metatarsal V.

I I Tl IV Vv

Agriotherium africanum . 71 82 96 101 £100

(L50787 + L45062)

Arctodus simus* . ;

(Potter Creek Cave)

Tremarctos floridanus*

70 82 95 99 100

69 80 91 103 100

(males)

Ursus spelaeus' . 63 78 88 98 100

(Salzofen)

Ursus arctos* 65 79 87 96 100

(Recent)

1 Kurtén 1967, table 26.

‘preyed on large contemporary herbivores’, and although not ‘truly cursorial

it may have been capable of bursts of speed exceeding those of U. arctos’. These

conclusions presumably apply equally well in the case of A. africanum.

The ecology and relationships of this species will be dealt with in more

detail in following sections of this report.

PALAEOECOLOGY

The Langebaanweg Agriotherium assemblage is comprised of over 330

specimens, which represent a minimum of 14 individuals. Females are much

less commonly represented than males, and no very young animals are known.

Those postcranial bones belonging to immature individuals are probably all

of young adults, while those of which teeth are known are all young or prime

adults. Some specimens (e.g. L40002) show signs of osteo-arthritis, which

suggests an advanced age for the individuals concerned. Bone pathology is

otherwise rare, one notable exception being the metacarpal, L45448, which

exhibits an osteitis of the proximal end, the cause of which is unknown.

All A. africanum specimens were found in, or closely associated with, river

channel deposits. They occurred together with a wide variety of terrestrial,

freshwater and marine vertebrates, which range in size from shrews to whales.

Lists of most associated mammals have been given elsewhere (Hendey 1976,

table 4; 1978, table 10). Associated birds will be listed by P.V. Rich (in pre-

paration). Lists of lower vertebrates, which include cartilaginous and bony

fish, amphibians and reptiles, have yet to be compiled. It is clear that A. afri-

canum was an element of a rich and diverse fauna, with resemblances to both

late Miocene (Turolian) faunas of Eurasia, and later African faunas. The

Varswater Formation fauna includes descendants of taxa typical of the Eurasian

late Miocene, with A. africanum included in this category, as well as ancestors

of species which are now typically African.

A. africanum is one of the Varswater Formation taxa not recorded from

the Quartzose Sand Member, the lowest of the three important fossil mammal-

bearing units of the succession (Hendey 1976) (Table 13). The Quartzose Sand

ANNALS OF THE SOUTH AFRICAN MUSEUM

jueurwmopeid

So}BIQOVWIOA | RII}Sel1o} i pues

onjeydsoyd ul UOUWIWOD sso]

Sului09eq jnq ‘“jURpUNQe s]Issoy

pues oiyeyd PeprAtp

-soyd ojur spieMdn 3ulpeis -qns

Se] jIssoy} pue pues zjienb yueq JOA pue jouueys IOAII jou

yusseid Ajgeqoid

sua][od yng ‘pep1OdeaI $9} ¥1q91.19A

AjUO ‘]JJ JO yYInos uoWUWOS

SSO] A[PAISSoIs 01d S]ISSOF

suoziioy Avid pue

pues zj1enb Jepun souilj}oWIoOs

“Keo pue pues snosdvuoqgies

puod pure ysieur

Aejo pue

pues sNOsdeUOGIeS O}UI SUIpeIs AI PUe I] UseM19q 9} eIPSUIE} UL

Sel jIssoy pue pues zji1enb

yueUIWOpeld so}e1q9}19A [eL}

-So119} PUL S[eOS f UOUIUOD S]ISsOy

yueurmopeid

SO} VIQOIOA [P1I}S91I19}

pue sjeos ‘juepunqe sjissoy

yoor ajeydsoyd jo sa] ur pue jouueys IdATI

‘uo 3x jIsso} pue pues zjrenb

UdaNdY ALIYOHdSOHG I1VLATWdg

yueulMopeid

$9}21Q9}19A [PII}SO1I0}

gjees ‘juepunqe s|Issoy

yoor ayeydsoyd uo pues Askejo yueq ISAII

pue

e1el AISA S3}e1QG

-O}LIOA ‘S[ISSOJOIOIUI SUIIUI SUIOS

poyenucieyipun

pues syeqdsoyd [ero] oulTeUr ‘Wdd

SAONFAANIIO TISSOFL SLNAWIGAS OILLSIAALOVAVH) SLNANWNOMIANA IVNOILLISOdId yaodai stu}

pure 9/6] Aepusyy

6L61

10 ja yIsuld

‘SomuLEgosURT ‘AlIengd .q, “UOMNeUIO, IoJeMSIVA OU} Ul SOOUSIINIDO [ISSOJ PUL SJUSUIIPSs SI}SIIO}OvIeY ‘sJUSWIUOIIAUS [eUOITISOdaqd

€[ J1av.L,

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

sjiuin AreJUWIIpas JO soroey oe (AT 01 [ INSO PUe AT O1 I N&E Wedd ‘S}UsTeAINboe I1ey} pue) py 0} Tq pue eq pue ZC SUOISIAIpgns oy,

jueulUoperd saye1q

-OJIOAUL PUL So}eIQ9}I0A SUTTeUT

‘yuepunge SOoUI}oWIOS s|Issoy

uMOUyY JOU

yueUIWIOpeld so}e1q

-O}IOAUL JoJeMYseIJ pue sUlIe

-njso “oulIeur ‘jUepUNge s|Issoy

WOUWIUIOS

$0} V1IQO}IOA [VII{Se1I0) “UeUTUIOp

-o1d suoyjod ‘yuepunge sjissoy

yueulwopeid

$0}81Q9}19A [PII}S01I9}

‘UOWWIOS SOUIjaWIOS s]ISSOy

yueulMopeid

SO}B1IQS}IOA [PII}S01I0}

‘yuepunqe sowiiewlos sjissoy

pues zj1enb

pue odes yoor 9sjeydsoyd

uMOUyY j0U

wis Appnut

Ae pue pues snosovuogieo

pues zj1enb

puvs zj1enb

(¢ “39 “6L61 (72 2 9]8UIC 90S)

yoveq Apues pure A011

(Aizenb Jo yjnos

pue jsvo-yjnos s}sixo A[qeqoid

yng posodxo ou) JouURYS JIOATI {

ysieur

JeuUeYd IOATI 0} Josopo ured

-pooy g ‘I Jo uorerres o[qissod

(jouuPvYyS sseuTeIp IOUTUI

‘puod °3'9) sjusUIUOIAUSOIOIUI

poyeioosse yA ‘urefdpooy

poyenuorsyipun

le)

ECINCI IAN

TAAVAD

wuasWayY, GNVS dASOZLUVNO

56 ANNALS OF THE SOUTH AFRICAN MUSEUM

Member deposits were laid down mostly on the floodplain of a river which then

met the sea to the south or south-west of the existing ‘E’ Quarry. Many of the

fossils from this unit are believed to represent the remains of animals which

lived in the immediate vicinity (Hendey 1974: 349-353; 1976: 223-226). By

contrast, most of the Pelletal Phosphorite Member fossils, including those of

A. africanum, were washed into the area by the river, which was then following

more northerly courses, first depositing bed 3aS and later bed 3aN (Hendey

1976: 226-230). Consequently, the A. africanum fossils are likely to be out of

their natural environmental context, and there is no way of certainly establishing

the nature of the preferred habitat of the species. Assuming that A. africanum

was a terrestrial species, the number of possibilities is, however, limited, and

there is some evidence which favours one of them.

The environment in the vicinity of Langebaanweg and in adjacent areas

at the time of deposition of the Varswater Formation was clearly very different

from that of the present (Hendey 1973). Long-necked giraffes (Giraffa sp.) were

common, and, together with other large browsers such as a sivathere (Siva-

therium hendeyi), a palaeotragine (Palaeotragus cf. germaini), and primitive

proboscideans (Anancus sp., Mammuthus subplanifrons), indicate the presence

of trees, probably in substantial numbers, and perhaps in the form of a riverine

woodland. On the other hand, grazers such as alcelaphine antelopes (Gentry

1980 in press), an equid (Hipparion cf. baardi), and a _ rhinoceros

(Ceratotherium praecox) indicate the presence of grasslands as well. Although

there was evidently a variety of micro-environments in the area (Hendey 1976),

it is only the major terrestrial habitats of woodlands and grasslands which

need be considered in the case of Agriotherium.

Of the large herbivores, the one which occurs most commonly in the

Quartzose Sand Member and bed 3aS of the Pelletal Phosphorite Member is

Ceratotherium praecox, a grazer, while browsing giraffoids are very rare in

the Quartzose Sand Member, and only slightly more common in bed 3aS. By

contrast, C. praecox is either very rare, or absent, in bed 3aN, while giraffoids

are astonishingly well represented. The implication is that either woodlands

became a progressively more widespread habitat during deposition of the

Varswater Formation, or that taphonomic factors were such that woodland

species had their remains incorporated in the deposits with increasing frequency.

Either way, the fact that A. africanum is not recorded from the Quartzose Sand

Member, is rare in bed 3aS, and is relatively common in bed 3aN, suggests

that it was a woodland species.

This conclusion has also been reached in respect of Agriotherium elsewhere.

Kurtén (1968: 119) suggested that Agriotherium ‘was probably a forest animal

like most modern bears’, while Wolff et al. (1973: 226) concluded that ‘it does

seem that specimens of /ndarctos and Agriotherium are better represented at

localities which have a greater representation of woodland forms’. The possible

influence of habitat on the evolution of Agriotherium will be discussed later

(see p. 70).

=e ER.

= = -e ~ —aer

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 57

The habitat preference of Agriotherium may have been a factor which

contributed to its comparatively poor fossil record, but this was almost certainly

due largely to the habits of the animal. Agriotherium has long been recognized

as an atypical ursid because it was apparently carnivorous rather than omni-

vorous. Carnivorous species of such gigantic proportions would of necessity

have been rare animals, even under the most ideal conditions (Wolff et a/. 1973).

The same restriction would not apply in the case of more omnivorous bears,

and for later species the ‘fossil record is excellent; indeed, as regards the Pleisto-

cene bears of Europe, almost incomparable’ (Kurtén 1968: 119). An added

factor in the case of the latter was that extreme climatic conditions during the

Pleistocene led bears to use caves as retreats, with consequent concentration

of their remains in caves. Such circumstances did not apply to Agriotherium,

all remains of which are recorded from open sites dating from a climatically

moderate period.

Although A. africanum has been said to be a well represented species, it is

nevertheless one of the less common elements in the Langebaanweg assemblage,

which is now comprised of the remains of many thousands of animals. Pro-

portionately, Agriotherium may be no more common at Langebaanweg than it

is at localities elsewhere.

Many of the Langebaanweg Agriotherium specimens were recovered in

the course of mining operations, or by screening of bulk sediment samples

mechanically removed from the mine. In these instances the original condition

_Of specimens, associations of skeletal elements and body part representations

cannot necessarily be determined. More significant from a taphonomic point

of view is that material recovered from controlled excavations. Three such

excavations yielded Agriotherium remains (Fig. 26, Table 14), and the relevant

material is listed on pages 5-6.

The deposits in which this material occurred are noteworthy for the almost

complete absence of a very coarse lithic fraction. Occasional pebbles of quartz,

feldspar and the local phosphate rock do occur, but the sediments are generally

made up of medium- to coarse-grade sands. Fine sands and clayey sands are

also present. The larger elements of the lag gravels in bed 3aN are almost

exclusively bones and teeth of vertebrates (Fig. 27). Consequently, those fossils

transported by the river were not subjected to the destructive battering by, and

against, cobbles and boulders, which is often a feature of this sedimentary

environment. For about 30 km east of Langebaanweg there are few rock out-

crops, the area being largely covered by the generally sandy deposits of the

‘Sandveld’ (see Talbot 1947; Visser & Schoch 1973). This means that in its

lower reaches the river which was largely responsible for building up the Vars-

water Formation (Fig. 1) could have picked up little in the way of a coarse

lithic fraction. In addition, its generally sandy bed would have provided a

relatively smooth passage for organic materials in its load.

In fact the fossils of beds 3aS and 3aN show remarkably few signs of

abrasion which could be ascribed to transport in sand-charged water over a

58 ANNALS OF THE SOUTH AFRICAN MUSEUM

Fig. 26. Aerial view of past and present areas of exposure of bed 3aN in ‘E’ Quarry, Lange-

baanweg. 1—LBW-E 1975/1; 2—LBW-E 1976/1; 3—Dump 10; 4—LBW-E 1976/2;

5—Dump 9; 6—Dump 8; 7—Carbonaceous deposits; R—Phosphate rock outcrops;

W—Wet season river channel; D—Dry season river channel.

sandy substratum. What abrasion there is may have developed after deposition

rather than during transport (see below). This suggests that many of the fossils

reached the vicinity of ‘E’ Quarry still protected by soft tissue, perhaps even as

floating carcasses, and that disarticulation and dispersal of skeletal elements

took place locally.

The bed 3aN Agriotherium specimens were recovered from deposits laid

down in three distinct micro-environments. Those deposits exposed in the

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 59

TABLE 14

Controlled excavations in ‘E’ Quarry which yielded Agriotherium remains.

EXCAVATION DEPOSITIONAL STRATIGRAPHIC

ENVIRONMENT Unit?

LBW-E river bank and

1975/1 river channel,

with phosphate

rock substratum

PPM 3aN

river channel,

1976/1 with phosphate

rock substratum

LBW-E river channel in PPM 3aN

1976/2 lee of phosphate II

rock, with

unconsolidated

sand substratum

1 See Table 13.

2 Locality 5 of Dingle et al. 1979, fig. 2.

excavation LBW-E 1975/1 were laid down partly in the river channel and

partly on the north bank of the channel. Channel deposits were sampled in

both excavations LBW-E 1976/1 and 1976/2, the depositional environments

differing only in that the former had a rock substratum and the latter a sandy

one. In both the 1975/1 and 1976/1 areas the substratum was a phosphate rock

horizon of up to 0,75 m thick. The third micro-environment was not sampled

by controlled excavation, but was the source of a few Agriotherium specimens

recovered in the course of mining operations. This was an extensive area of

carbonaceous (peat-like) deposit over the southern (seaward) limit of the

river channel. It was probably the area of accumulation of plant debris washed

down by the river during flood times. During the dry season it formed the west

bank of the river and was probably a marshy area with appropriate vegetation.

The likely positions of the wet and dry season channels are indicated in Figure 26.

The bed 3aN deposits are overlain by the thick and extensive commercially

exploited phosphatic sand, which was deposited in a marine littoral environ-

ment (Tankard 1975), and from which some fossil vertebrates, not including

Agriotherium, are known (Hendey 1976: 230).

The fossils from LBW-E 1975/1, which were deposited close to or on the

north bank of the river channel, were generally better preserved and less frag-

mented than those from the other two excavations. This applied particularly

in the case of specimens not in direct contact with the phosphate rock sub-

stratum. In the area of LBW-E 1976/1 most of the deposit overlying the phosphate

rock had been mined away, and the material recovered came from on, or close

to the rock surface itself, particularly depressions therein. This material was

for the most part very fragmented.

The greatest concentration of fossils in bed 3aN was in the area of LBW-E

1976/2. Here the deposits were laid down on an unconsolidated substratum

(the Quartzose Sand Member) in the lee (west) of the phosphate rock exposed

60 ANNALS OF THE SOUTH AFRICAN MUSEUM

in the other two excavations. Immediately adjacent to the phosphate rock a

60 cm thick horizon of fossils was accumulated (Fig. 27). This thinned out

rapidly to as little as 10 cm westwards and southwards. Most of the fossils in

this area were highly fragmented and in a poor state of preservation. This applies

particularly in the case of the remains of larger species, of which giraffoids,

especially Sivatherium, were by far the most commonly represented. By contrast

the remains of aquatic vertebrates, with the seal, Prionodelphis capensis, being

exceedingly common, tended to be in good condition, although their remains

were often fragmented and skeletons disarticulated and dispersed.

As this fossil lag deposit thinned westwards and southwards, the clastic

matrix became finer-grained, with an increasing clay component. Westwards

the fossiliferous horizon terminated abruptly against another phosphate rock

outcrop, but southwards it once again thickened and also spread out laterally

in a south-westerly direction. There was also a rapid darkening in the colour

of the clastic matrix, which coincided with a diminution in the occurrence of

vertebrate fossils. These fossils occurred mostly as isolated, and often frag-

mented, teeth and bones, but associated parts of skeletons, including at least

one of Agriotherium, are recorded. In the case of the latter, it is not known how

complete they were, since all were chance discoveries made after disturbance

of the deposit by mechanical excavators.

The darkening of the deposits southwards was caused by an increasing

carbonaceous fraction, which was evidently derived from decomposed plant

remains. This deposit was not a pure peat, the carbonaceous material having

been mixed with a high proportion of sand and clay, but such peats might well

have overlain the remaining carbonaceous deposit (see below).

The picture which emerges is that of a river which in times of flood carried

in to the area remains of terrestrial vertebrates, sometimes as whole carcasses,

depositing some along its banks and others in the channel itself. A major part

of the load of vertebrate remains was deposited immediately after the channel

passed over the western edge of a southward projecting tongue of phosphate

rock. Another such outcrop about 30 m further west then deflected the channel

southwards, where it spread out over a wide and flat area. Here it dropped the

last of its organic load, this probably being comprised largely of easily trans-

portable plant material. Some vertebrate remains also reached this area, but

they are recorded only from the more northerly parts, that is, closest to the area

where the main vertebrate load was dropped.

Although the vertebrate assemblage of bed 3aN is comprised largely of the

remains of terrestrial species washed in by the river, marine vertebrates are also

represented. In the case of the seal the number of individuals involved is sub-

stantial. The marine vertebrate remains are generally better preserved than those

of terrestrial species, which suggests, not surprisingly, that they had suffered

less transport. The seals, cetaceans, marine birds, bony fish, and sharks may all

have been inhabitants of, or visitors to, the river estuary. This probably applied

particularly during flood times when the influx of carcasses of terrestrial species

*SUOISIAIP WO ¢ Ul BJVOG “JYSII oY} WO UOT}OOIIp

Ajioseo-yj10U eB ul spremdn Surdojs doroyno yoo syeydsoyd pure “jJo] pomosse jisodep BxI [Issog “A11end .q, JO vore 7/9/61 A-MAT ‘LZ ‘Stet

re]

ie}

;

62 ANNALS OF THE SOUTH AFRICAN MUSEUM

provided an abundant source of food for marine scavengers. Whether or not

the seal was included in this category is not known. Seals may simply have been

attracted by an increase in the numbers of scavengers such as fish and crus-

taceans, and would themselves have attracted predators such as sharks. Marine

vertebrates were certainly in a position to have their remains caught up in, and

dispersed by, the river’s floodwaters.

During the dry season when the river was not in flood it is unlikely to have

breached the phosphate rock outcrop which underlies the northerly and easterly

parts of bed 3aN. Instead it would have been deflected southwards by this

outcrop in the direction of bed 3aS, which was laid down during an earlier

phase of the Varswater cyclothem. This would account for the fact that bed 3aS

was abruptly truncated south of the phosphate rock outcrop. This truncation

was not due solely to flood periods in bed 3aN times because bed 3aS terminated ~

at least 25 m east of the first exposure of the carbonaceous deposit of bed 3aN,

which is an indication of the width of the dry season channel. There may have

been other distributaries of the river north of ‘E’ Quarry.

As indicated earlier, that area west of the dry season channel, where

floodwaters had dropped their load of organic materials, probably took the

form of a marsh.

Although the river probably still fed vertebrate remains into the area

during the dry season, the number of specimens involved is likely to have been

much lower. This raises a problem in connection with specimens collected

between the principal exposures of bed 3aS and bed 3aN. There are relatively

few such specimens, and their actual source is usually uncertain, since most

were collected by mine workers in the course of their activities. This material

has been recorded as being from bed 3aS, but it may actually belong with the

bed 3aN complex of deposits. Those Agriotherium specimens with numbers

between L40000 and L43000 fall into this category.

Fortunately there are no doubts about the source of Agriotherium specimens

from the bed 3aN excavations, and those from the bed 3aN carbonaceous

deposits, the latter being distinctive because of their dark colour. The only

other fossils from ‘E’ Quarry in a similar state of preservation are those from

the ‘peat bed’ of the Quartzose Sand Member (Hendey 1976: 218, table 2), and

there was no possibility of material from these two deposits becoming mixed.

Although there can be little or no doubt as to how the bed 3aN Agrio-

therium remains reached the ‘E’ Quarry area, it is of interest to consider the

nature of some of this material.

The partial skeleton, L45062, is interpreted as the remains of an animal

which reached the point of its discovery as a complete, or nearly complete

carcass. This carcass was deposited about 15 m west of the phosphate rock

outcrop in the LBW-E 1976/2 area. It is virtually certain that at least some,

and perhaps all, missing parts of L45062 were mined away, the parts recovered

having come from an area of 2 to 3 m* immediately adjacent to a vertical face

cut by a mechanical excavator. The remains were found in the lag deposit where

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 63

it. was about 20 cm thick, and were centred on a depression on the unconsoli-

dated substratum. The preservation of the remains was considerably better

than that of the majority of surrounding fossils. Their good preservation,

together with the unmistakable characteristics of Agriotherium bones and teeth,

facilitated their recovery from a mass of thousands of fragmentary fossils.

It appears that after deposition of the carcass of L45062, flowing water

scoured the depression beneath it. At the same time the carcass acted as an

obstruction to coarse debris, causing it to settle in the immediate vicinity.

Disarticulation and slight dispersal of skeletal elements followed decomposition

of soft tissue. L45062 is unusual in being one of the few instances in the LBW-E

1976/2 area where parts of one individual were found in a good state of preserva-

tion, and with skeletal elements in close association. This was otherwise notice-

able mainly amongst the seals, although in these instances the large number of

individuals involved made it impossible to separate their skeletons.

The Agriotherium forelimb, L45063, was another instance where there was

association of skeletal elements of one individual. This material was found in

the same depression as L45062, but was readily distinguishable, having belonged

to a smaller (female) individual. This forelimb must also have reached the area

held together by soft tissue, perhaps with the scapula and most phalanges

already detached. Although no two elements of this limb were found in

articulation, the individual elements were less dispersed than those of L45062.

The incomplete manus, L33160, from LBW-E 1975/1, was a similar occurrence

to L45063.

Other Agriotherium specimens from the bed 3aN controlled excavations

tended to be isolated occurrences of individual skeletal elements. Since they

are generally well preserved and show few, if any, signs of abrasion, they, too,

are likely to be from carcasses similar to that of L45062. However, in these

instances there was greater dispersal of skeletal elements and associations were

no longer obvious. The condition of isolated specimens was usually in keeping

with the nature of the depositional environment. Long bones were invariably

fragmented, but shorter and stouter bones, such as those of the manus and pes,

were usually intact. The hemimandible, L45114, had lost the single-rooted

teeth (incisors, P; and M3), which are easily separated from the jaw, and the

ascending ramus, which is more fragile than the mandibular corpus. The single-

rooted canine was still in position, because the root of this tooth is large and

slightly bulbous, and impossible to remove from the jaw without breaking the

root or the bone enclosing it.

Also of interest from a taphonomic point of view is the post-mortem damage

to specimens. This was clearly dependent on the micro-environment in which

specimens were deposited.

The incomplete manus, L33160, was deposited close to, or on the north

bank of the river in the LBW-E 1975/1 area. This specimen is remarkable

because it provides evidence of the activities of hyaenas in this area. Like many

of the fossils from ‘E’ Quarry, L33160 shows signs of having been chewed by

64 ANNALS OF THE SOUTH AFRICAN MUSEUM

—

=—_

thet

nein

canal

——

Hyaena-gnawed Agriotherium manus, 133160, from Langebaanweg. Arrows indicate

areas of major damage.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 65

a hyaena, and this was done while the bones were still articulated. The phalanges,

metacarpal I and distal ends of the metacarpals II to V are lost, probably having

been ingested by the hyaena concerned, and there are tooth-marks on some of

the metacarpals adjacent to the missing parts. There is also damage on the

scapholunar and unciform, but the magnum and trapezoid, which are largely

enclosed by the other bones, are intact (Fig. 28).

The remains of at least six small hyaenas (Ictitherium preforfex) were

recovered from LBW-E 1975/1. This species of hyaena, and perhaps others,

probably scavenged the banks of the river for the remains of animals washed

into the area. Hyaena-damaged bone was not uncommon in this area, another

notable example being the skull of a seal with double punctate marks on the

braincase. The lower canines of Ictitherium preforfex fit these punctate marks

well.

Another specimen from LBW-E 1975/1, the metatarsal L30205, is remark-

able in having on its shaft gnaw-marks apparently made by a small rodent

(Fig. 29). This type of damage is rare amongst the fossils from ‘E’ Quarry.

110 Mi 12

113

Fig. 29. Rodent-gnawed Agriotherium metatarsal, L30205, from Langebaanweg.

Arrow indicates area of damage.

Although rodent remains are generally common in the fossiliferous deposits

of the Varswater Formation, this was not the case in the LBW-E 1975/1 area,

where only bathyergids were represented in moderate numbers. Living bathy-

ergids, which are fossorial, are known to gnaw at objects encountered in their

tunnels, plastic water-pipes and telephone cables being items recently affected

in this way in the south-western Cape Province. It is possible that a bathyergid

tunnelling on the river bank was responsible for the damage to L30205.

Three distinct types of post-mortem damage are evident on the bones and

teeth of the partial skeleton L45062. Most common is simple fracturing of

bones, which is ascribed to movement after loss of soft tissue, and to subsequent

compression by overlying deposit. Not surprisingly, it was the relatively delicate

parts of the skull which were particularly affected by this fracturing. The second

type of damage is abrasion, which was probably caused by sand-charged water

66 ANNALS OF THE SOUTH AFRICAN MUSEUM

flowing over exposed parts of the partly buried skeleton. This type of damage

is not uncommon on the fossils from ‘E’ Quarry, and a good example was

mentioned elsewhere (Hendey 1970: 82, fig. 3).

The third type of damage was caused by fire, with affected bones and teeth

being blackened, more badly fractured and less well preserved than unburnt

bone (Fig. 30). This type of damage is also not uncommon on fossils from

‘E’ Quarry, and has been discussed elsewhere (Hendey 1974: 351; 1976: 224).

Fig. 30. Fire-damaged Agriotherium astragalus,

145062, from Langebaanweg.

It was previously assumed that fire-damage was caused by dry-season bush or

grass fires. While this may sometimes have been the case, there are instances

where the inferred depositional environment and nature of the damage is such

that fires of another sort are likely to have been responsible. The fire-damage

of L45062 is a case in point.

It is clear that this damage was done after the skeleton had been dis-

articulated, since severe fire-damage on one bone is not matched by damage to

immediately adjacent parts of the skeleton. For example, the symphyseal region

of the right hemimandible is fire-damaged and the incisors are lost, whereas

the left hemimandible is intact in this region and the incisors are present.

Similarly, the distal end and part of the shaft of the left ulna is damaged, but

corresponding parts of the left radius are unaffected.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 67

Assuming that the depositional environment of L45062 has been correctly

interpreted, it is highly improbable, if not impossible that the random burning

of parts of the skeleton could have been caused by a bush or grass fire. A more

plausible explanation is that the fire-damage was caused by peat fires, a

phenomenon which has been reported in a North American estuarine swamp

by Staub & Cohen (1979). It was indicated earlier that the horizon in which

L45062 occurred graded laterally into a peat-like deposit, and that pure peats

may have overlain this horizon. Such peats, if they did exist, may have been

prone to dry season fires like the American example cited above. Having burnt

away they would have left no trace obvious to observers unfamiliar with such

phenomena other than some underlying burnt bone.

Peat fires may also have led to the formation of hitherto unexplained

deposits capping the peat-like sediments immediately south of the LBW-E

1976/2 area. Immediately overlying the ‘peat’ was a horizon of sand a few

centimetres thick, which was recorded as being ‘orange-brown’ in colour and

in this respect unlike any other sand body recorded before or since in ‘E’ Quarry.

The ‘orange-brown sand’ was in turn overlain by a grey clay incorporating

scattered sand grains. In the examples of peat fires recorded by Staub & Cohen

(1979), and in other similar ones (e.g. Cypert 1961), ponds developed in depres-

sions left in the peat after burning, and such ponds become a new and distinct

depositional environment in the areas in question. If there was, indeed, a peat

fire in the LBW-E 1976/2 area, the orange-brown sand and grey clay may

represent sediment accumulated in a resultant pond, while the underlying

carbonaceous deposit represents an unburnt residue of the original peat deposit.

The carbonaceous deposit may have remained unburnt either because it was

waterlogged or because its high mnon-carbonaceous content made it

incombustible.

The grey clay is finely laminated, suggesting slow accumulation in still-

water conditions, with individual sediment particles perhaps having been

transported to the pond by wind. The origin of the orange-brown sand is not

known. It was initially thought that it may represent the unburnt residue of

the peat fire, but the occasional vertebrate fossils incorporated in the sand

show no signs of having been burnt.

The existence and effects of peat fires at Langebaanweg are largely specu-

lative, but they do provide a plausible explanation for hitherto unexplained,

or unsatisfactorily explained, aspects of the fossils and the deposits.

The ecological role of Agriotherium is also relevant here. It was mentioned

earlier that Agriotherium was a carnivorous animal and, judging from its

dentition, it was better adapted to carnivory than almost all other later ursids.

It is the nature of the posterior cheek teeth which are particularly significant

in this respect, and since these teeth are better represented in A. africanum than

in previously recorded Agriotherium, it is useful to examine them from a func-

tional viewpoint.

The P*, M! and M? of A. africanum, and other Agriotherium, resemble one

68 ANNALS OF THE SOUTH AFRICAN MUSEUM

another in having a series of prominent buccal cusps and less prominent lingual

ones (Fig. 31). In all there are seven buccal cusps, three on P* and two on each

of the molars, and they are in the form of a smooth curve, with the convexity

directed buccally. In addition to being lower-crowned, the lingual cusps are

less distinct and fewer in number, only five excluding the vestigial accessory

cusps on the P* protocone lobe. These cusps are more or less in a straight line.

TULA LULL

Fig. 31. Langebaanweg Agriotherium specimens showing prominence of buccal cusps.

A. Lingual view of P* and M1. B. Anterior view of M?.

Judging from wear facets, the buccal cusps have a combined shearing and

crushing function, with the former being predominant. The lingual cusps also

have this double function, but in this instance the crushing function is

predominant.

Much the same applies in the case of P, and the lower molars, in which

the more prominent cusps are situated buccally. An exception is the large

posterolingual cusp of the M, talonid. This cusp is, however, like the more

buccal ones in developing an inclined shearing facet on its buccal side. The

other lower teeth or cusps which develop inclined shearing facets buccally are

the paraconid, protoconid and hypoconid of M,, the protoconid and hypoconid

of M,, and the anterobuccal part of Ms.

A most significant development relating to the emphasis of the buccal

cusps in A. africanum is the reduction or loss of the anterolingual cusps of the

M,.’s, L45114/L46563 and L50007. As far as is known this has not previously

been observed in Agriotherium. It creates the impression that the M, of A. afri-

canum was tending to become a second lower carnassial. Although smaller

than the true carnassial, L45114/L46563 and L 50007 resemble this tooth in

having the trigonid large and functioning essentially as a shearing element.

To have carried this evolutionary experiment to its logical conclusion

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 69

would have required continued suppression of the lingual cusps of P*, M! and

M2, continued emphasis of the buccal cusps of these teeth, and a similar emphasis

on the more buccal parts of the lower molars. This would have resulted in a

sectorial dentition unique amongst Carnivora, but since no Agriotherium is

known in which there was an advance on the A. africanum condition, it evidently

served the needs of the genus adequately.

It is worth noting in this connection that a similar evolutionary path has

been followed by the polar bear, Thalarctos maritimus. In this species the P4

protocone is reduced or absent, the buccal cusps of M! and M? are relatively

higher crowned than those of other ursines, while the lingual ones, including

the M? talon, are correspondingly reduced. These dental characters reflect ‘the

carnivorous habits of the species and diverges from the omnivorous dentition

of most other ursids’ (Kurtén 1964: 4).

Whether Agriotherium was a predator or a scavenger is not known. Its

large and cumbersome build suggests that the former alternative is less likely,

although it may have been capable of hunting down at least some of the larger

contemporary herbivores, such as the giraffids. The P* of Agriotherium has

long been referred to as hyaenid-like (Falconer & Cautley 1836; Wagner 1837),

and recently Wolff (1978: 4) has written that this tooth ‘bears a striking resem-

blance in robustness to the teeth of the bonecrushing hyaenas’. This implies a

belief that Agriotherium may have been a scavenger. However, in the hyaenas it

is the more anterior cheek teeth which are enlarged to perform a bone-crushing

function, and the enlarged cheek teeth of Agriotherium may have been too

posteriorly situated to perform this function efficiently.

Irrespective of how its food was acquired, there can be no doubt that an

abundance must have been available in the Langebaanweg area when it was

inhabited by Agriotherium. There is evidence, however, which suggests that the

Varswater Formation fauna dates from the latter part of the period which

Kurtén (1971: 152) termed the ‘climax of the Age of Mammals’, and that even

in its heyday it was becoming an anachronism. In fact, it is possible that the

circumstances which led to the evolution of Agriotherium ultimately also caused

its extinction.

The late Tertiary was a period of world-wide environmental change, with

a general lowering of mean annual temperatures (Butzer 1971, fig. 2), which

heralded the glacial-interglacial oscillations of the Pleistocene. Over wide

areas of the Old World and the Americas, forests and woodlands were giving

way to savannas and grasslands, and these changes had a profound effect on the

character and composition of terrestrial vertebrate faunas (Webb 1977).

The Varswater Formation fauna provides evidence of these changes. For

example, it includes two of the earliest alcelaphine antelopes yet recorded

(Gentry 1980), and it is these bovids with their high-crowned teeth which are

characteristic of the African savannas today. Similarly, an early ancestor of

Africa’s grazing rhinoceros (Ceratotherium simum) is recorded from Langebaan-

weg (Hooijer 1972). Palynological evidence from Langebaanweg, and elsewhere

70 ANNALS OF THE SOUTH AFRICAN MUSEUM

in the south-western Cape Province, indicates that the modern Cape macchia

(fynbos) vegetation was becoming established at this time, and that the climate

was changing from ‘Cool Wet’ to ‘Colder Drier’ (Coetzee 1978: 121, fig. 2).

There is even some as yet unpublished evidence to suggest that there was a

marked fall in local sea temperatures during deposition of the Varswater

Formation.

The immediate cause of the climatic and environmental deterioration in the

Langebaanweg area and adjacent regions was the development of the Benguela

Current System off the west coast of southern Africa and the consequent

aridification of the adjacent land mass (Siesser 1978; Tankard & Rogers 1978).

The development of the Benguela Current was in turn the result of major

glaciation in Antarctica later in the Miocene (Kennett et al. 1975).

This period of climatic and environmental change must have influenced

the Ursidae as much as any other mammalian group. During the Vallesian |

the ursids which inhabited Europe were generally relatively small omnivorous

animals inhabiting forests and woodlands. That lineage which was to give rise

to the Ursinae, and probably also the Tremarctinae, continued to be repre-

sented by small species during the Turolian. However, in the Indarctos lineage

there was a marked increase in the size of the species during this period. In the

following section of this report, evidence will be presented which indicates that

Indarctos was the ancestor of Agriotherium, and consequently it is the Indarctos

lineage, and the changes it underwent, which are relevant here.

By the Turolian (and its equivalents) the environment in mid-latitudes was

in the process of change, with woodlands giving way to more open country

(Kurtén 1971). Consequently, woodland plant foods such as fruits, nuts and

berries, favoured by omnivorous ursids, were probably becoming less readily

available. It is the larger species which would have been most adversely affected

by this development, and thus Turolian Jndarctos, and its counterparts else-

where, may increasingly have been forced to adapt their diets to more abundant

food sources. In this instance, the response was evidently in the direction of

increased carnivory, with adaptations culminating in the evolution of

Agriotherium.

During the Pleistocene a similar situation arose in respect of Thalarctos.

At the onset of one of the glaciations, populations of Ursus arctos did not

retreat southwards in advance of the ice, but instead adapted to the new con-

ditions. Amongst other things, this involved a growing reliance on carnivory

to replace the plant foods which had previously been part of its diet (Kurtén

1964; Hendey 1972).

The early ancestors of the giant panda, Ailuropoda, evidently responded to

late Tertiary environmental change in a different way. They adapted their diet

to include more readily available plant foods which had not necessarily figured

in their diets previously. Thus arose the most herbivorous of all living ursids,

whose main diet of bamboo shoots is varied by other plant foods when this is

possible (Ewer 1973). :

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

be!

Rex :

“ff

‘OL6[ Bulidg ‘somuvegesuryT ‘AjIJedo1d ourur py] sojwioyo sul “ze “SI

72 ANNALS OF THE SOUTH AFRICAN MUSEUM

By becoming a highly specialized herbivore, Ai/uropoda ensured its survival,

although latterly in diminishing numbers and in increasingly remote areas.

On the other hand, by adopting carnivory, Agriotherium placed itself in direct

competition with smaller and better adapted predators and scavengers for a

declining food source (i.e. large woodland browsers), or one which was increas-

ingly difficult for a large ambulatory carnivore to acquire (i.e. cursorial grazers

of the savannas and grasslands). According to this hypothesis, the fate of Agrio-

therium was sealed by the factors which led to its origins.

The picture of the Langebaanweg Agriotherium which emerges is of a

gigantic carnivore living at a time when vertebrate life flourished in the south-

western Cape Province, and when the Langebaanweg area was well-watered

and richly vegetated. The semi-arid environment of Langebaanweg today,

devoid of indigenous trees and scarred by man’s activities, is a poor reflection ~

of the past (Fig. 32). Only some of the smallest of the carnivorous cousins of

Agriotherium, namely foxes, polecats, mongooses and wildcats, still occupy

the area and the prospects for their survival are limited.

RELATIONSHIPS OF AGRIOTHERIUM

The first Agriotherium specimens to be collected and recorded were several

isolated teeth of an aged individual found at Montpellier in France early in the

nineteenth century. These specimens, which are preserved in the Museum

d’Histoire Naturelle, Geneva, were described by Cuvier (1822) under the name

of ‘Lophiodon de Montpellier’. Stehlin (1907) gave an account of this historic

material, whose true identity remained a mystery for many decades, and pro-

vided an indication of the difficulties originally encountered in correctly identi-

fying and classifying specimens belonging to Agriotherium.

Although its ursid affinities were recognized by Falconer & Cautley (1836)

on the basis of material from the Siwalik Hills of India, Agriotherium is in many

respects an atypical member of the family. Of all the genera of Ursidae, it is

Agriotherium and its herbivorous counterpart, Ailuropoda, whose relationships

have been most controversial. The fact that Agriotherium has a poor fossil

record, while that of its nearest relatives is not necessarily any better, has

aggravated the problem. In addition, there has been a tendency to compare it

with the ursine bears, which have long been well known, and this has further

obscured matters, since the ursines and Agriotherium are only distantly related.

Much the same applies in the case of Ailuropoda.

Hyaenarctos Owen, 1845, a junior synonym of Agriotherium Wagner, 1837,

was the generic name in common use during the last century and the earlier

part of the present one. Prior to the identification of Indarctos by Pilgrim (1913),

“Hyaenarctos’ was the name applied to species now referred to both Agrio-

therium and Indarctos. This is an indication that these two genera have many

characters in common, a point made obvious by studies such as those of Frick

(1926) and Matthew (1929). Pilgrim (1931, 1932) did much to clarify the situa-

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 5

tion, and thereafter, apart from occasional lapses (e.g. Viret 1939), the name

‘Hyaenarctos’ fell into disuse, and Pilgrim’s interpretation of the genera Agrio-

therium and Indarctos was generally accepted.

Pilgrim (1932: 42, 43) listed half a dozen characters which he believed

distinguished these genera and concluded that ‘Indarctos appears to represent

a distinct line of development from Agriotherium, in some ways more and in

others less advanced than the latter’, while ‘J. punjabiensis seems to some extent

to bridge over the gulf between [them]’. In spite of the attention given by

Pilgrim to the problem of distinguishing Agriotherium from Indarctos, and in

determining their relationships to one another, and to other ursids, uncertain-

ties have remained. Nevertheless, it has become common practice to include

these genera in the subfamily Agriotheriinae, together with ‘the basal genus of

the family, Ursavus’ (Kurtén 1966: 7).

Since Agriotherium and Indarctos fossils are not common, they, and the

question of their relationships, have received only infrequent attention since

Pilgrim’s (1932) review. Such attention has been prompted on the one hand by

descriptions of new material (e.g. Viret 1939; Kretzoi 1942; Tobien 1955;

Thenius 1959; Hendey 1972; Crusafont & Kurtén 1976; Wolff 1978), and on

the other by reviews of ursid inter-relationships and phylogeny (e.g. Erdbrink

1953; Thenius & Hofer 1960). On those rare occasions when the relationship

between Agriotherium and Indarctos has been discussed, Pilgrim’s (1932)

opinion has been favoured. For example, both Erdbrink (1953, fig. 61) and

Thenius & Hofer (1960, figs. 34-35) place these genera on separate lineages,

and indicate derivation from an unknown common ancestor in the Ursavus

group.

This theory is based on the belief that Agriotherium is the more ‘primitive’

(i.e. canid-like) of the two genera, which, since it survived later than Indarctos,

must represent a distinct lineage. However, there is an alternative hypothesis.

In previous studies on the Langebaanweg Agriotherium it was suggested that

this genus was directly descended from Jndarctos (Hendey 1972, 1977), an

idea which was apparently first conceived by Schlosser (1899). Before examining

this alternative in more detail it is worth noting what is known of the temporal

ranges of the two genera.

In Europe Agriotherium is first recorded from the Ruscinian, while Indarctos

is known only from the Vallesian and Turolian (Table 15). In North America

Agriotherium is restricted to the late Hemphillian (i.e. 4,5-6,0 Ma) and Indarctos

is recorded only from the latter part of the early Hemphillian (i.e. 6-7 Ma)

(R. H. Tedford 1979, pers. comm.). In Africa the Langebaanweg Agriotherium

is from a Ruscinian-equivalent fauna whereas the cf. Agriotherium from Sahabi

(Boaz et al. 1979) is probably an advanced Indarctos in terms of current defi-

nitions, and is almost certainly a little older than the Langebaanweg species.

The situation in Asia is still obscure, although it is now known that the Dhok

Pathan of the Siwaliks, from which important Indarctos and Agriotherium

specimens are recorded, spans an appreciable period during the late Miocene

74 ANNALS OF THE SOUTH AFRICAN MUSEUM

TABLE 15

The occurrence of Jndarctos and Agriotherium in Europe and the Middle East.

APPROX.

AGE IN

m.y.t

TAXON LOCALITY MAMMAL AGE! | MEIN ZONE?

Ruscinian

Turolian

Montpellier

Concud

Samos

Pikermi

Maragha

Montredon Late

Orignac Vallesian —

Pfaffstetten

Westhofen

Can Llobateres Early 12

Can Purull Vallesian

Agriotherium insigne .

Indarctos atticus

Indarctos arctoides

Indarctos vireti

1 Berggren & Van Couvering 1974; 7? Mein 1975.

and Pliocene (Pilbeam et al. 1977). Consequently, the Dhok Pathan repre-

sentatives of these genera, whose taxonomy is controversial (see below), were

not necessarily contemporaneous with one another. A similar situation may

well exist in the case of Jndarctos and Agriotherium from Chinese late Tertiary

localities, with the former known from Localities 30, 31, 43 and 52, while

Agriotherium is tentatively identified from Locality 13 (Kurtén 1952).

Indications are, therefore, that Agriotherium and Indarctos were not

contemporaries, and when their age is known the former appears later in the

fossil record. Consequently, their known temporal ranges are in accord with the

theory that Agriotherium was descended from Indarctos.

While its relationships to Agriotherium may be controversial, the history

of Indarctos itself is now reasonably well documented. It apparently had its

origins in Europe during the Vallesian, having stemmed from an Ursavus, and

subsequently spread through Asia and into North America (Thenius & Hofer

1960), as well as to Africa (see above). Differing interpretations of inter-generic

relationships are possible. For example, Crusafont & Kurtén (1976) suggested

that J. vireti was ancestral to both I. arctoides and I. atticus, whereas I. arctoides

is here regarded as an intermediate between the other two species. Crusafont

& Kurtén (1976: 15) further suggested that J. anthracitis ‘may be a precociously

specialized form’. This species is, indeed, unusual, perhaps because it evolved

in isolation on a Tethyan island, but it is largely irrelevant to present con-

siderations. Recorded Asiatic Indarctos are either close to I. atticus or more

advanced, while North American /ndarctos is more advanced than J. atticus.

Irrespective of the actual inter-relationships of Indarctos species, there is

no doubt that J. vireti is a generalized and early form, whereas J. atticus, and

Asian and North American Jndarctos are more specialized and younger in age.

For example, Crusafont & Kurtén (1976: 15) noted that J. atticus is ‘more

advanced [than other European /ndarctos] with larger cheek teeth, more reduced

premolars, heavy and powerful jaws, and ursine limb proportions’.

a ee — rf

-—- ——e

> — a J

, nn

ta ~— eee 6 ee | A oe.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA US

It follows that if Agriotherium were descended from Jndarctos, then the

evolutionary trends evident in the /. vireti-J. atticus lineage are likely to have

been continued in the hypothetical Indarctos—Agriotherium lineage. This is,

indeed, the case, and Crusafont & Kurtén’s comments quoted above apply

equally well to Agriotherium relative to J. atticus. Such general trends do not

necessarily constitute proof of a direct phylogenetic relationship between the

two taxa.

Much more convincing evidence comes from an examination of certain

tooth and skull characters in advanced Jndarctos (i.e. those between 6 and 9 m.y.

old), and early Agriotherium (i.e. those that are, or probably are, about 5 m.y.

old). Examples of the latter are A. africanum from Langebaanweg, A. insigne

from Europe (Montpellier), and A. palaeindicum from the Siwaliks. Advanced

Indarctos includes J. atticus from Europe (Samos) and Iran (Maragha), J. pun-

jabiensis from the Siwaliks, and North American Indarctos. All recorded speci-

mens of the latter are here referred to the species J. oregonensis (see p. 101).

The classification of material referred to some of the above species has

long been complicated by specimens which exhibit a combination of characters

‘typical’ of both genera. Pilgrim (1932: 4446) discussed such a problem involv-

ing three incomplete mandibles from the Dhok Pathan of the Siwaliks (GSI-D8,

D9, D10), and although he found it ‘difficult to give a definite answer’, he

decided to reverse the identifications suggested by Lydekker (1884). Pilgrim’s

identifications are accepted here, with D8 assigned to Agriotherium palae-

indicum, while D9 and D10 are assigned to Jndarctos punjabiensis. The latter

has since been recognized as one of the more advanced representatives of

Indarctos (Thenius 1959), while A. palaeindicum is one of the more primitive

representatives of Agriotherium (Hendey 1977).

It is highly likely that more will yet be written on the identity of the Siwaliks’

Indarctos and Agriotherium. Both Pilgrim and Lydekker may have been incorrect

in their interpretation of GSI-D8, D9 and D10 (and other specimens) since

they could represent a single species which was no more variable than the

Langebaanweg Agriotherium. The anomalous situation in respect of J. pun-

Jabiensis and A. palaeindicum will be repeatedly evident in the discussions which

follow. The study of new and well provenanced material from the Siwaliks

may resolve a situation which is beyond satisfactory resolution on the basis of

available evidence.

Whatever the final outcome of this controversy, the fact that specimens

can with some justification be identified with either Indarctos or Agriotherium

is here regarded as highly significant, with specimens such as GSI-D8, D9 and

D10 being interpreted as ‘intermediates’ between ‘typical’ Jndarctos and ‘typical’

Agriotherium. The present study has shown that such ‘intermediates’ are more

common than has hitherto been supposed. The characters which the two genera

have in common, coupled with apparent evolutionary trends linking them,

provide surprisingly good evidence for the transition of Indarctos to Agrio-

therium considering the relatively poor fossil record of the taxa concerned.

76 ANNALS OF THE SOUTH AFRICAN MUSEUM

Indarctos—Agriotherium ‘intermediates’ are included in the A. africanum

assemblage, while an undescribed A. insigne specimen from Montpellier

(NMB-MP549) and the recently described Indarctos skull from Florida (Wolff

1978) are also in this category. The Florida skull is perhaps the best single

recorded specimen which is ‘intermediate’ between ‘typical’ Indarctos and

‘typical’ Agriotherium. The age of this specimen is ‘late Hemphillian’ (Wolff

1978: 1), and it probably dates back about 6 m.y. It is certainly younger than

the Samos J. atticus skulls described by Helbing (1932) and Thenius (1949,

1959), and it is much younger than the skull of J. vireti from Spain (Crusafont

& Kurtén 1976). On the other hand, it is probably a little older than the skull

of the Langebaanweg A. africanum.

Judging from the few measurements given by Wolff (1978), and from

illustrations, the Florida skull is appreciably larger than that of J. atticus, which

in turn is larger than that of J. vireti (Crusafont & Kurtén 1976). The trend of

increasing size with time in the Jndarctos lineage is clearly illustrated by these

specimens. This trend was continued with the evolution of Agriotherium,

since the skull of A. africanum is, in some respects at least, still larger than that

of the Florida Jndarctos. In respect of size the Florida skull is probably closer

to that of A. africanum (and A. sivalense) than the three described skulls of

European Jndarctos.

The general similarity between the skulls of the Florida Jndarctos and

A. africanum was discussed earlier (see pp. 9-16), while the differences between

them can all be ascribed to the more advanced condition of the latter. The

advances are probably all related to the larger size of the A. africanum skull,

and to modifications of its masticatory apparatus.

Judging from the Florida Indarctos and A. africanum skulls, the following

are the most significant cranial characters shared by advanced Indarctos and

Agriotherium:

1. Snouts relatively short and broad.

2. Zygomatic arches very stout and of similar shape.

3. Sagittal crest very high.

4. Overall similarity of the basicranial regions, particularly the absence

of the alisphenoid canal, and the positions of the oval, postglenoid and

other foramina.

In respect of the development of the snout, zygomatic arches and sagittal

crests, the Florida Jndarctos is intermediate between J. atticus and A. africanum.

The basicranial region of J. atticus is similar to those of the other two skulls,

and, according to Crusafont & Kurtén (1976), the basicranial region of /. vireti

is similar to that of J. atticus. I. vireti is more primitive, however, in having

the ‘facial part of the skull. . . relatively much longer than in J. atticus’ (Crusa-

font & Kurtén 1976: 10).

The relatively short snout of Agriotherium goes together with a relatively

short palate, which is of interest because Kurtén (1964: 22) found that the

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA Ti

palate of the polar bear, Thalarctos, is slightly shorter than that of the brown

bear, U. arctos, a pattern which he was not ‘able to match... . in any other bear

population’. Evidently the Indarctos—Agriotherium example was not taken into

account, but it is a parallel to U. arctos-Thalarctos in this respect. Kurtén (1967)

subsequently noted that, like Thalarctos, a short and broad snout characterizes

Arctodus, another ursid which is convergent with Agriotherium (see p. 51).

- Another apparent similarity between the skulls of Agriotherium and

Thalarctos is that in lateral view they have a relatively straight profile, with the

sagittal crest prominent (Erdbrink 1953). It was on this basis that Wagner (1837)

proposed the name Agriotherium for the Ursus sivalensis of Falconer & Cautley

(1836) (see Erdbrink 1953: 557). Thalarctos is like Agriotherium and Arctodus in

being a carnivorous animal descended from an omnivorous ancestor, and other

parallels between them will be mentioned below.

The presence of a premasseteric fossa in the mandible of Agriotherium, and

its absence in Jndarctos, is an important distinguishing characteristic, the sig-

nificance of which has prompted much published and unpublished comment.

In one recent account it was erroneously stated that the premasseteric fossa is

also characteristic of Ursavus (Hendey 1977), an error stemming from Frick

(1926: 99), citing Wegner (1913). In fact, Ursavus, like Indarctos, lacks this

fossa, and it is thus peculiar to Agriotherium in the hypothetical Ursavus—

Indarctos—Agriotherium lineage.

The premasseteric fossa is important from a phylogenetic point of view,

since it, like other ‘characteristics’, may not be an invariably diagnostic feature

of Agriotherium. For example, it is absent from at least one mandible which

has been assigned to Agriotherium, namely, the A. palaeindicum specimen

GSI-D8. On the other hand, in the Samos J. atticus specimen described by

Thenius (1959) (NMW-Samos 1912, 29), there is a slight depression in the pre-

masseteric region which could be an incipient fossa. No such fossa, however,

has been reported in more advanced Jndarctos, such as that from North America.

Thenius (1959) pointed out that most of the Jndarctos mandibles then

known were either incomplete or belonged to immature individuals, and that

the premasseteric fossa was either not observable or absent. This fossa clearly

is an ontogenetic character, since in the Langebaanweg assemblage it is well

developed only in older individuals (e.g. L45062), and is shallow in the mandible

of a young adult (L45114) (Fig. 33).

A premasseteric fossa is otherwise known amongst ursids in Hemicyoninae

and Tremarctinae. Although the early history of tremarctines is not well known

(Thenius 1976), there is nothing to indicate that they and the hemicyonines are

closely related. Consequently, it is certain that the premasseteric fossa in these

two groups was evolved independently. There is thus no reason to suppose

that its presence in Agriotherium is indicative of a close relationship with either

the Hemicyoninae or the Tremarctinae. Since the significance of the premas-

seteric fossa in ursids is not known (Davis 1955), there is no way of knowing

why it should develop in some lineages and not in others.

78 ANNALS OF THE SOUTH AFRICAN MUSEUM

A B

Fig. 33. Dorsoventral cross-sections of Lange-

baanweg Agriotherium hemimandibles with arrows

indicating premasseteric fossae. A. 145114.

B. L45062.

The teeth of Indarctos and Agriotherium provide much evidence in support

of the theory that they are directly related.

As is often the case with carnivore incisors and canines, little of signifi-

cance emerged from a study of these teeth in Indarctos and Agriotherium, except

that they are essentially similar morphologically The [' and I? are perhaps the

most distinctive of the anterior teeth in A. africanum, and they are closely

matched by those of the /. atticus specimen from Samos described by Helbing

(1932, fig. 2) (NMB-Sam31). The anterior teeth of Agriotherium are distinguished

from those of Indarctos principally by their larger size, this being a reflection of

the overall size differences between the two genera.

It is worth noting in this connection that since the Agriotheriinae, like other

ursids, exhibit appreciable sexual dimorphism, it is possible that large males

of advanced Indarctos were of similar size to, and perhaps even slightly larger

than, small Agriotherium females. For example, in terms of overall size the

maxillary fragment of a small A. insigne specimen from Montpellier (NMB-—

MP549) is virtually identical in size to corresponding parts of the Vienna

I. atticus specimen (NMW-Samos 1912, 29) (Fig. 34). Similarly, the Florida

Indarctos skull is in some respects as large as that of the Langebaanweg Agrio-

therium, which belongs to a male, and would therefore have been larger than

those of A. africanum females. Thus size alone may not necessarily be a reliable

criterion for distinguishing the two genera.

One of the general trends in ursid evolution has been the emphasis on the

development of the posterior cheek teeth, and the. reduction in size or even loss

of the anterior premolars (Pj-P3). More reduced premolars is one of the charac-

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 719

ee eee C IT)

Fig. 34. Upper cheek tooth rows of Indarctos

atticus, NMW-Sam1912/29 (left) and Agriotherium

insigne, NMB-MP549 (right). Drawn from casts.

ters which distinguishes J. atticus from the earlier J. vireti (Crusafont & Kurtén

1976). In I. atticus P} and P§ are apparently always single-rooted, while P3 usually

have two roots, although P, may sometimes have only one root (Thenius 1959).

In the Florida Indarctos P! and P? are single-rooted, and P? is double-rooted

(Wolff 1978).

Judged on the basis of the Langebaanweg sample, the anterior premolars

of Agriotherium had undergone even further reduction, since in this instance

all are invariably single-rooted and as many as two of a series may be lost

(Table 16). This sample does not, however, exhibit one of the extremes in

anterior premolar development encountered in Agriotherium. The Montpellier

A. insigne specimen, NMB-MP3549, is like advanced Indarctos in having a

80 ANNALS OF THE SOUTH AFRICAN MUSEUM

TABLE 16

Anterior premolars in Agriotherium from Langebaanweg.

L41404

150003 = = p p p

150445 — == — p p p

140044 — —_— p a p

L50004 —_— — p a p

150903 = = p a p

p p

145062

145114

p = present; a = absent.

double-rooted P? set transversely in the jaw. It is especially reminiscent of the

Vienna J. atticus specimen in this respect (Fig. 34), the only difference being

that the principal cusp in the latter is slightly more distinct. The P? of NMB—

MP549 is unlike that of any other recorded Agriotherium.

By contrast, the Vienna J. atticus specimen is Agriotherium-like in having

the P, to P; reduced to single-rooted teeth.

Thus, in respect of anterior premolar development conditions typical of

advanced Jndarctos may occasionally be found in Agriotherium, and vice versa.

In addition, the anterior premolars of Agriotherium are morphologically similar

to those of advanced Indarctos, except for the slightly more distinct principal

cusps in the latter. This is a primitive characteristic since the principal cusps of

the anterior premolars of /. atticus are in turn less well developed than those of

I. vireti.

The P,’s of advanced Indarctos and Agriotherium are also almost indis-

tinguishable in terms of their basic morphology. Differences in detail are probably

no greater than those in the Langebaanweg Agriotherium sample, which includes

one specimen (L50445) with the principal cusp configuration resembling that

in the J. atticus specimen, NMB-Sam31 (Fig. 35). Of particular significance is

the tendency in Indarctos for the development of a postero-internal bulge in

the cingular region. This feature is well developed in the Maragha J. atticus

specimen (De Mecquenem 1925). In Agriotherium the postero-internal bulge

on P, is comparably developed, whereas it is absent or much less pronounced

in all other ursids.

The principal differences between the P,’s of advanced Indarctos and

Agriotherium are that the latter are higher crowned and tend to be larger in

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 81

Fig. 35. Buccal and occlusal views of P,’s.

A. Indarctos atticus, NMB-Sam31 (after Helbing

1932, fig. 3). B. Langebaanweg Agriotherium,

L50445.

overall size. All the posterior cheek teeth of Agriotherium are higher crowned

than their counterparts in Indarctos, although the Florida Indarctos specimen

may be an exception in this respect. Apparently the crown height increase in

the Indarctos—Agriotherium lineage was more or less in proportion to the overall

increase in the size of the taxa concerned.

A likely parallel of this situation is that involving Thalarctos, which has

relatively higher crowned posterior cheek teeth than Ursus arctos (Kurtén 1964).

In the case of the upper carnassial (P*), it is the nature of the protocone

lobe and the development of the parastyle which are of particular phylogenetic

significance.

As a general rule the P* of Agriotherium is distinguished from that of

Indarctos by the presence of a parastyle. However, some specimens of Jndarctos

do have a P* parastyle, although it is usually less prominent than that of Agrio-

therium. For example, this cusp is present in J. atticus from Concud (Crusafont

& Kurtén 1976) and Samos (Helbing 1932; Thenius 1959), I. punjabiensis from

the Siwaliks (Lydekker 1884), and North American Jndarctos (Merriam et al.

1925; Merriam & Stock 1925; Wolff 1978). The parastyle is particularly well

developed in some North American specimens (e.g. the Florida skull), which

are younger than European J. atticus. This cusp is absent in the still older and

more primitive J. arctoides and I. vireti. There was thus a tendency for the

development of a P* parastyle in the Jndarctos lineage, and this was continued

in the presumed descendant, Agriotherium, in which this cusp is always present

and well developed.

As far as is known the only other ursids in which the P? parastyle is present

are Ursavus depereti and its possible descendant, the giant panda, Ailuropoda

melanoleuca (see p. 96).

The situation in respect of the P* protocone lobe is slightly different,

although equally informative. According the Crusafont & Kurtén (1976: 8)

82 ANNALS OF THE SOUTH AFRICAN MUSEUM

there was a tendency for J/. vireti to develop an antero-internal cusp on the P*,

a situation which ‘is not uncommon in J. arctoides and I. atticus’. This cusp is

also present in the J. punjabiensis specimen, GSI—D6, although in this instance

it is rather small (Lydekker 1884). In the younger and more advanced North

American Jndarctos this cusp is either reduced (Wolff 1978) or absent (Merriam

& Stock 1925). This indicates that while a double-cusped protocone lobe is

characteristic of most /ndarctos, there was a tendency for the antero-internal

cusp to be reduced or lost in advanced forms.

By contrast, an antero-internal cusp is found only in some of the earlier

Old World representatives of Agriotherium, namely, I. insigne from Mont-

pellier (NMB-MP549) and the Langebaanweg A. africanum. In the case of the

latter, this cusp is best developed in the holotype, L2045, from bed 3aS, which

predates those A. africanum P*’s from bed 3aN in which the antero-internal

cusp is small or absent.

Indications are, therefore, that early forms of European and African

Agriotherium still had an antero-internal cusp on P*, but this was soon reduced

and lost. On the other hand, no Asian or North American Agriotherium is

known to have had this cusp, which had already been reduced and lost in the

advanced Indarctos of these continents. This suggests that the Indarctos—

Agriotherium lineage may have had at least two geographically separated

branches, one in Europe and Africa and the other in Asia and North America,

which in respect of their P* protocone lobes evolved at different rates. The impli-

cations of this possibility will be discussed later (see pp. 101—4).

Reduction of the P* protocone is a characteristic of the Ursus—Thalarctos

lineage (Kurtén 1964), and in South American Arctodus (Kurtén 1967). This is

another example of a parallel development in Agriotherium, Thalarctos and

Arctodus.

Once again a double-cusped protocone lobe is otherwise known only in

Ursayus depereti and Ailuropoda, although in these taxa the morphology of the

lobe is somewhat different (see p. 96).

As with the upper carnassial, the lower one (M,) is an important tooth in

indicating the origins of Agriotherium. There is, however, a complication with

M,, although in the final analysis its phylogenetic significance is not diminished.

The complication concerns the identification of the posterolingual cusps of this

tooth.

The cusp of the Agriotherium M, which has invariably been identified as

the metaconid may in reality be the entoconid. If this is, indeed, the case, then

the metaconid of the Agriotherium M, is either reduced or absent, usually the ,

latter.

Both the metaconid and entoconid are present and well developed in |

Ursavus and Indarctos, although with time the metaconid becomes a progres-

sively less prominent feature of My,, since it is reduced in size relative to the

entoconid and is increasingly overshadowed by the protoconid. In U. primaevus ~

the metaconid is larger and more prominent than the entoconid, and is only ~

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 83

slightly less high-crowned than the protoconid (e.g. Crusafont & Kurtén 1976,

fig. 13). Much the same applies in the case of J. vireti and J. arctoides, except

that in these species the metaconid and entoconid are of comparable size

(Crusafont & Kurtén 1976, fig. 2; Tobien 1955, fig. 3). In I. atticus the metaconid

is much less prominent than the protoconid, but is still of similar size to the

entoconid (Helbing 1932, fig. 3).

In the smallest of the A. africanum M,’s (L50006) what is interpreted as a

vestigial metaconid is still present. It is completely overshadowed by the proto-

conid, and, unlike its counterpart in J. atticus, is smaller than the entoconid.

The metaconid is absent in all other M,’s from Langebaanweg, and, apparently

also from all other recorded Agriotherium M,’s. The progressive reduction and

eventual loss of the M, metaconid in the Ursavus—Indarctos—Agriotherium lineage

is illustrated in Figure 36.

Fig. 36. M,’s. A. Ursavus primaevus (after Crusafont & Kurtén 1976, fig. 14). B. Indarctos

arctoides (after Tobien 1955, fig. 3). C. Indarctos atticus (after Helbing 1932, fig. 3). D. Lange-

baanweg Agriotherium (D1—L50006; D2—L50004; D3—L50446). All are lingual views

except C, which is a buccal view with the buccal talonid cusps omitted. e—entoconid;

m—metaconid; p—protoconid.

If the largest of the posterolingual M, cusps in Agriotherium is indeed the

entoconid (i.e. a talonid cusp), rather than the metaconid (i.e. a trigonid cusp),

this would account for Tobien’s (1955: 14) observation that the “Metaconid’ of

Agriotherium is ‘niedriger und starker zuriickgeschoben’, while in /ndarctos it is

84 ANNALS OF THE SOUTH AFRICAN MUSEUM

‘héher und naher an das Protoconid gestellt’. The cusp which Tobien believed

to be the metaconid in Agriotherium may simply be a well-developed entoconid

in more or less its usual position, while in Jndarctos the metaconid was correctly

identified as such, and it, too, is in its usual position.

The metaconid in /ndarctos has a counterpart on the buccal side of the

talonid, this being a small cusp situated between the protoconid and hypoconid.

It is present in J. vireti (Crusafont & Kurtén 1976), I. atticus (Helbing 1932),

I. oregonensis (Dalquest 1969), and other specimens, although traces of its

presence may be obliterated by wear. A vestige of this cusp may be represented

in the A. palaeindicum specimen, GSI-—D8, by a slightly inclined ridge anterior

to the hypoconid. A similar ridge is present in the M, of A. africanum, although

in this species it takes the form of an undemarcated horizontal extension of the

hypoconid. It meets with the posterobuccally directed keel of the protoconid.

L50006 is also unusual in lacking this keel and in having the hypoconid ridge

linked directly with the vestigial metaconid.

The reduction and loss of the M, metaconid in the Indarctos—Agriotherium

lineage thus appears to have been accompanied by reduction and loss of its

buccal counterpart. The impression gained is that these cusps were ‘absorbed’ by

the protoconid, which is considerably enlarged in Agriotherium. This is a

manifestation of the development of the shearing cusps (paraconid and proto-

conid) at the expense of the crushing cusps (talonid cusps) in the Indarctos—

Agriotherium lineage.

It is possible, however, that the traditional interpretation of the large

posterolingual M, cusp in Agriotherium as the metaconid is correct. Another

of the Langebaanweg specimens, L50446, is unusual in having the largest of the

posterolingual cusps flanked posteriorly by two smaller cusps, whereas in all

other specimens there is only one such cusp. An almost exact match of L50446

is the M, of the A. palaeindicum specimen, GSI-D8 (Lydekker 1884). If the

largest of the posterolingual cusps in these specimens are metaconids, then the

smaller cusps immediately posteriorly would be the entoconids. It would,

therefore, be the entoconid which is lost in other Agriotherium M,’s, and this,

rather than the loss of the metaconid, would have characterized the Indarctos—

Agriotherium lineage. This hypothetical link would then be supported by the

observation that the entoconid is as well developed in Agriotherium specimens

such as L50446 and GSI-D8 as it is in some advanced Jndarctos (e.g. I. oregon-

ensis—Dalquest 1969, fig. 4).

In the case of the first alternative suggested above, the two smaller postero-

lingual cusps in L50466 and GSI—D8 would be interpreted as a duplication of

the single cusp in this position in other Agriotherium M,’s.

A first-hand examination of all relevant specimens may be necessary before

deciding which of the above alternatives is likely to be correct. Irrespective of

which applies, a transition from the typical Indarctos condition to that typical

of Agriotherium is documented by specimens from the Siwaliks and

Langebaanweg.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 85

Another trend evident in the lower carnassials of Indarctos and Agrio-

therium is that of a reduction in their relative lengths with time (Table 17).

Only the problematical Siwaliks’ specimens GSI-D8 and D9 are anomalous in

terms of their length: breadth ratios.

The arrangement and morphology of the M! cusps in advanced Jndarctos

and Agriotherium is virtually identical. The M? of A. africanum differs from that

of I. atticus only in being relatively shorter and higher crowned, although it may

be indistinguishable from more advanced Indarctos (e.g. the Florida specimen)

in these respects.

In most, and perhaps all, Indarctos M?’s the posterior keels of the meta-

cone and hypocone are linked across the posterior end of the tooth by a low

ridge of enamel. This feature is present but less obvious in A. africanum, and

probably all other Agriotherium, and this region of the tooth also differs in

being noticeably shorter than in the corresponding part of the J. atticus M?.

In other words, there is a very short ‘talon’ region in the latter, and it is reduction

of this feature which contributes to the overall shortening of M! in Agriotherium.

TABLE 17

Length : breadth ratios of Indarctos and Agriotherium My,’s.

Loca ity and/or

TAXON NUMBER LENGTH | BREADTH L:B MEANS

Agriotherium africanum .\ { L45062 1 1,82: 1

145062 2 il

150004 oat |

L50006 21

150446 Bil

145114 al

L50005 Bil

Agriotherium insigne .\ Montpellier! oa 17a

Montpellier? aI

Agriotherium

palaeindicum . .| GSI-D8? Balt 2,00 : 1

Indarctos punjabiensis .| GSI-D9? gt 1,88 : 1

Indarctos atticus . .| NMB-Sam31 Bal! 1,90:1

NMW-Saml 912/29? v1

Concud? al

Indarctos arctoides* .| Westhofen il 1,93: 1

Pfaffstetten |

Montredon ori

Indarctos viretit . .| Can Llobateres

VP633 $ 2,01 : 1

VP647

Can Purull

Type (pontiensis)

Type

1 Viret 1939; * Lydekker 1884; ° Thenius 1959; +4 Crusafont & Kurtén 1976.

86 ANNALS OF THE SOUTH AFRICAN MUSEUM

This is a significant difference, because once again a specimen of one genus

is known which has the character of the other. An Jndarctos-like post-metacone

lengthening is one of the few features visible on the incomplete and badly

restored M? of the Montpellier A. insigne specimen, NUB-MP549.

The ‘primitive’ M! of this remarkable specimen is yet another of its Indarctos

‘characteristics’, others being its relatively small size, double-rooted and trans-

versely orientated P’, and double-cusped P* protocone lobe. This specimen also

has the P*, M! and M2? lower crowned than any of their counterparts in the

Langebaanweg assemblage. However, it is identified with Agriotherium because

its M? is of the Agriotherium type and distinct from that of all specimens referred

to Indarctos (see below). Had the M? of this specimen not been preserved, it

may well have been referred to Jndarctos. Montpellier could therefore have

erroneously acquired the distinction of being the only locality where Indarctos

and Agriotherium occurred together. This imaginary situation is mentioned

here to indicate the importance of NMB-MP549 as an Indarctos—Agriotherium

‘intermediate’, and to illustrate how easy it is to misidentify specimens belonging

to late Indarctos and early Agriotherium. The possibility of similar confusion

with certain Siwaliks specimens was mentioned above.

Before dealing with the next tooth in the dentition, another parallel between

Agriotherium and Thalarctos is mentioned. The M? of the latter is relatively

shorter and higher crowned than that of U. arctos, its ‘structural’ ancestor

(Kurtén 1964). This also applies in the case of at least some species of Arctodus

(Kurtén 1967).

The M, of A. africanum, and other Agriotherium, usually consists of two

trigonid cusps side by side, flanked posteriorly by two similarly positioned

talonid cusps. The M, of Jndarctos differs only in having two lingual talonid

cusps, and in being relatively longer and narrower. There are, however, Agrio-

therium specimens which are Indarctos-like in both these respects. The M, of

the A. palaeindicum specimen GSI—D8 has two lingual talonid cusps, while

vestiges of a second cusp are visible in the Langebaanweg specimens L54114/

L46563 and L50007. The length:breadth ratios of the M,’s of GSI-D8 and

L45114/L46563 are comparable to those of I. atticus M,’s (Table 18). By

contrast, this ratio in the J. punjabiensis specimen GSI-D9 is the same as that

of the mean of the A. africanum sample.

Curiously, the A. africanum M,’s which are most Jndarctos-like in respect

of talonid cusps and proportions are those which are most specialized in terms

of their trigonid development (see p. 27). Consequently, these specimens should

not be considered as good Jndarctos—Agriotherium intermediates. In addition,

it could be argued that Pilgrim (1932) was incorrect in reversing Lydekker’s

(1884) identifications of GSI-D8 and D9 and that they, too, are not ‘inter-

mediates’ in the sense claimed above. This may, indeed, be the case, but the

fact remains that however these (and other) Siwaliks specimens are identified,

they exhibit a combination of Indarctos and Agriotherium characters.

The dentitions of Indarctos and Agriotherium are perhaps most clearly

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 87

TABLE 18

Length : breadth ratios of Indarctos and Agriotherium M,’s.

Locatiry and/or

NUMBER LENGTH | BREADTH L:B

145062

150004

150003

150007

145114

146563

Montpellier?

Agriotherium africanum .

Cie} 'afel (int hele) nie oie

pnh | pemed pemek pemek peek feek feed fe

Agriotherium insigne

Agriotherium

palaeindicum .

Indarctos punjabiensis

Indarctos atticus

GSI-D8?

GSI-D9?

NMB-Sam31

NMW-Sam1 912/298

Montredon

Westhofen

Pfaffstetten

Can Llobateres

VP633

VP640/1

VP647

Can Purull

Type

Type (pontiensis)

1,39: 1

1,29:1

1,39:1

Indarctos arctoides* 1,41:1

Indarctos vireti*

1 Viret 1939; * Lydekker 1884; * Thenius 1959; 4 Crusafont & Kurtén 1976.

distinguished from one another by the fact that the M? of the latter lacks a talon,

whereas in Jndarctos this feature, although variably developed, is always present.

The postero-internal (talon) region of the Agriotherium M? is also variably

developed, the Langebaanweg assemblage being useful in indicating the varia-

tion possible in a single population (Fig. 12). Of particular interest is the isolated

M?, L47698, in which there is a marked posterior projection of the postero-

internal part of the tooth. This is here interpreted as the vestiges of the talon

characteristically present in the M? of Indarctos. L47698 is remarkably similar

to the M?, GSI-D12, referred to J. punjabiensis by Lydekker (1884, fig. 6), and

these two specimens represent an intermediate between the conditions typical

of Indarctos and Agriotherium. The progressive shortening and broadening of

M”’s in the Jndarctos—Agriotherium lineage is indicated by the data in Table 19.

Erdbrink’s (1953: 582) view that there is ‘at best a beginning of a [M?]

talon .. . in A. insignis’ is here regarded as the reverse of the true situation.

A. insigne, like A. africanum, sometimes has the vestiges of a M? talon.

As with M}!, the arrangement and morphology of the four principal cusps

of M? in advanced Indarctos and Agriotherium are very similar. The paracones

88 ANNALS OF THE SOUTH AFRICAN MUSEUM

TABLE 19

Length : breadth ratios of Indarctos and Agriotherium M*’s.

LOCALITY or

NUMBER LENGTH} BREADTH L:B

Agriotherium africanum .|_ 48577 0,84: 1

148564 0,84 : 1

141404 0,85 : 1

L45062 0,87: 1

145062 0,88 : 1

L45137 0,93 : 1

L45137 0,94 : 1

L47242 1,00: 1

L47698 1,03 : 1

Agriotherium insigne NMB-MP549 0,89 : 1 ail

Type* 1,03: 1

Indarctos punjabiensis GSI-D12? 1,06: 1 ou

Indarctos atticus NMW-Sam1912/298 1,13:1 cat

NMB-Sam31 1,28: 1

Indarctos arctoides Montredon* 1,29:1 al

Montredon? 1,43 :1

Orignac* 1,28:1

Gau—Weinheim® 1,37:1

Indarctos vireti VP6334 1,47: 1 |

VP646? Sil

1 Frick 1926; *Lydekker 1884; *Thenius 1959; 4 Crusafont & Kurtén 1976; *°Tobien 1955.

and metacones are conical, with distinct anterior and posterior keels, the

protocone is ridge-like, and, with the small hypocone, is lower than the buccal

cusps.

The M*’s of Indarctos and Agriotherium illustrate very clearly a progressive

development from /. vireti, through I. arctoides, I. atticus, I. punjabiensis to

Agriotherium, with A. africanum taken as an example of its genus (Fig. 37).

The recorded M?’s of advanced North American Indarctos all have well-

developed talons, that of the J. cf. oregonensis specimen recorded by Merriam

& Stock (1925) being the most prominent. It is, however, from the same deposits

as another J. oregonensis M?, which is a larger tooth but with a relatively smaller

talon (Merriam et al. 1925). The latter is here interpreted as belonging to a male,

while the smaller specimen with the larger talon belongs to a female.

This raises the possibility that the teeth of female Indarctos and Agrio-

therium tend to be more ‘primitive’ than those of the larger males. This explana-

tion applies in the case of the ‘primitive’ A. insigne specimen, NMB-MP549,

discussed earlier, which represents a small individual, apparently a female.

Another example is the A. africanum M, with the vestigial metaconid, L50006.

The female I. cf. oregonensis is also ‘primitive’ compared with other North

American Jndarctos in having a relatively small P* parastyle and a relatively

elongated M1. It does not, however, have a double-cusped P* protocone lobe.

This, together with the fact that NMB-MP549 does not have a talon on M?,

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 89

A B c

D1 D2 D3

Fig. 37. Ms. A. Indarctos arctoides (after Helbing 1932, fig. 7). B. Indarctos atticus (after

Helbing 1932, fig. 1). C. Indarctos punjabiensis (after Lydekker 1884, fig. 6). D. Langebaan-

weg Agriotherium (D1 —L47698; D2—L45137; D3 —L41404).

indicates that females were not necessarily ‘primitive’ in all respects. Never-

theless, since increasing size was a characteristic of the Indarctos—Agriotherium

lineage, and consequently large size was itself an ‘advanced’ character, it is to be

expected that larger individuals would also be ‘advanced’ in other respects, and

vice versa. This is a further indication that allowance must be made for appre-

ciable variation in characters in studies on Indarctos and Agriotherium. The

fact that known North American Indarctos M*’s appear more ‘primitive’ than

the I. punjabiensis specimen, GSI-D12, does not necessarily mean that they

must be older.

There is other evidence to support this opinion. Once again there is a

parallel between Agriotherium and Thalarctos, since the latter also has the M?

talon reduced, and it may sometimes even be absent. The study by Kurtén

(1964) has shown that there is considerable variation in Thalarctos in this

respect even within a single population. Particularly remarkable is a specimen

in which the right M? talon is reduced, while that of the left M? is completely

absent (Kurtén 1964: 17, pl. 4A). Such examples are exceptional, but presumably

if samples of Indarctos as large as Kurtén’s Thalarctos sample (n = 113) were

available, similar specimens may be found. This suggests that undue reliance

may have been placed on the value of the M? talon as a distinguishing charac-

teristic of Indarctos.

The M, of Indarctos and Agriotherium is less commonly represented than

other posterior cheek teeth, evidently because it is single-rooted and easily

90 ANNALS OF THE SOUTH AFRICAN MUSEUM

separated from the mandible. It is nevertheless clear that this tooth was reduced

in length in the /ndarctos—Agriotherium lineage to a degree comparable to that

of M®. In J. vireti the Mg is markedly elongated anteroposteriorly and has a

pronounced posterior tapering (Crusafont & Kurtén 1976). The M, of J. atticus

is similar, but is less elongated, while that of Agriotherium is usually only as

long as it is broad. The latter, however, is still Jndarctos-like in having a pro-

nounced posterior tapering. The A. africanum Mg, L50981, is very like those of

the Vienna and Basel /. atticus specimens (Thenius 1959; Helbing 1932) in shape.

The reduction of Mg is another development shared by Agriotherium and

Thalarctos.

Although mention has been made of differences in proportions in certain

of the posterior cheek teeth discussed above, this matter warrants more detailed

attention, since these differences are directly related to changes in tooth mor-

phology and are as significant from a phylogenetic point of view. In addition,

metrical data have the advantage of being easily represented diagrammatically

and can thus give a visual impression of certain evolutionary changes in the

hypothetical Ursavus—Indarctos—Agriotherium lineage. Five Old World species

were selected to represent successive stages in this lineage. They are Ursavus

primaevus, I. vireti, I. arctoides, I. atticus, and A. africanum. Individual species

do not necessarily represent the stock from which the next in the series was

derived, but are simply structurally suited to such a role. The lengths and

breadths of their posterior cheek teeth are plotted in Figures 38—40.

It is clear from these figures that U. primaevus is in an isolated position,

whereas there are similarities between J. vireti and J. arctoides on the one hand,

10 SE

10 -E

p4 BREADTH / p4 LENGTH

2—/ -D 2 -D

1 -C 2 -c

2 -B 2 -B

-A 3 -A

(ee ee ee a ee ee eee ee ee ee ee SS Sd

8 10 12 14 #16 18 20 22 24 26 1214 16 18 20 22 24 26 28 30 32 34 36

M, LENGTH

=) 3

-C 3

-B

-A 5

s40 12-4416 8202222 1) 20 22 3A 28 20, S02 34, 36> GU TAUEROEAS

Fig. 38. Lengths and breadths of P* and M,. A. Ursavus primaevus. B. Indarctos vireti.

C. I. arctoides. D. I. atticus. E. Agriotherium africanum. Sample sizes, ranges and means are

indicated. Data from Crusafont & Kurtén (1976), Thenius (1959) and this report.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 91

6 -E 6 -E

mM! BREADTH mM! LENGTH

2 -D -D

3 -C -C

-B —B

4 -A 4 -A

Le en Se Eee eee ee ee eee (ee ee eS eee ee eee ee

10 12 14 16 18 20 22 24 26 28 30 12 14 16 18 20 22 24 26 28 30 32

-E -E

=D -D

-c -c

—B -B

-A —-A

ee ee Se ES eS ee es ee ee Ee ee

8 10 12 14 16 18 20 22 24 26 14.16 18 20 22 24 26 28 30 32 34

Fig. 39. Lengths and breadths of M' and M, of some Ursidae, (see Fig. 38 for key).

and J. atticus and A. africanum on the other. The isolated position of U. primae-

vus is not surprising since it is a contemporary of J. vireti (Crusafont & Kurtén

1976), and its phylogenetic connection with the /ndarctos—Agriotherium lineage

must be indirect. The fact that this lineage is divisible into two parts on the

basis of cheek tooth size (and other evidence) is significant. The first is comprised

of earlier, smaller and more primitive forms (i.e. J. vireti-I. arctoides), and the

second of later, larger and more advanced forms (i.e. J. atticus—A. africanum).

9 -E 9 -E

M2 BREADTH M> LENGTH

2 -D =D

3 -C -c

—B8 -B

3 =A. 3 -A

ee ee ee ee eee a SS SS eS ee ee es ee ees |

10 12 14 16 18 20 22 24 26 28 30 32 14.16 #18 20 22 26 26 28 30 32 34

-E -E

M, LENGTH

=D 2 -D

-B 4 -B

-A 2 -A

(aa) js ts

8 10 12 14 #16 «18 «20 8 10 12 14 16 18 20 22

Fig. 40. Lengths and breadths of M? and M; of some Ursidae, (see Fig. 38 for key).

92 ANNALS OF THE SOUTH AFRICAN MUSEUM

It is only in the last segment of the lineage that there are deviations from

otherwise general trends in the lineage as a whole. The most obvious of these

general trends was the increasing breadth of the posterior cheek teeth. This

was probably more or less in proportion to the overall size increase of the taxa

concerned. However, the situation in respect of the lengths of these teeth is

more complex.

Only in the case of P* was there a tendency to increase length throughout

the Ursavus—Indarctos—Agriotherium lineage. This is also a reflection of the

increasing size of successive members of the lineage. In addition, it may be

interpreted as indicating the increasing functional importance of the principal

shearing tooth in the upper dentition, with the increased lengths of P* in the

last stage of the lineage (i.e. J. atticus—A. africanum) being largely due to the

development of the parastyle. It might be expected that there would have been

a corresponding increase in the length of the lower carnassial. The fact that

there is a barely perceptible increase in the mean length of the A. africanum M,

over that of J. atticus may simply be due to the composite nature of this tooth,

the increased length of the shearing element (i.e. the trigonid) being obscured

by the reduction of the talonid.

By contrast, in the case of M? and M, the lengthening trend evident in most

of the lineage is reversed in the /. atticus—A. africanum segment. This is a mani-

festation of the reduced importance of the crushing function of the most pos-

terior cheek teeth in the last stage of the lineage.

The situation in respect of the lengths of the intermediate teeth, M! and

M,, was itself intermediate, with lengths remaining static once the J. atticus stage

was reached.

The overall impression gained from the study of the morphology and

dimensions of the teeth of successive members of the Ursavus—Indarctos—

Agriotherium lineage is that this lineage can be separated into two parts. Up to

the I. atticus stage the taxa concerned were generalized ‘omnivorous’ ursids,

but thereafter there were modifications of both tooth morphology and pro-

portions being manifested which indicate adaptation to a more carnivorous

diet. It may thus be more appropriate to think of ‘advanced Jndarctos’ as being

“primitive Agriotherium’.

Since the transition from Jndarctos to Agriotherium was gradual in the

sense that there are specimens which exhibit ‘characteristics’ of both genera,

it is far from obvious on what grounds the distinction between these genera is

to be made. Earlier attempts to find distinguishing characters are now seen to

have been inadequate. For example, lists of such characters given by Pilgrim

(1932: 42) and Tobien (1955: 14) can be misleading and should be used with

caution, if they are to be used at all. The formulation of mutually exclusive

diagnoses for these genera has become difficult, if not impossible. There would,

of course, be no problem if these diagnoses were to be based on a primitive

Indarctos, such as I. vireti, and any of the species presently identified with

Agriotherium. However, when all recorded intermediate forms are taken into

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 93

account, the diagnoses break down under a welter of qualifying statements. The

situation could only become more confused if all the Indarctos and Agriotherium

assemblages from the critical 5-8 Ma period were as large, or larger, than that

from Langebaanweg. Even if this is never the case, the Langebaanweg assem-

blage has clearly shown that agriotheriines are as variable as other ursids, and

consequently there can no longer be complacency about supposed ‘charac-

teristics’ of individual representatives of this group.

This situation is here interpreted as indicating an ancestor—descendant

relationship between Indarctos and Agriotherium beyond all reasonable doubt.

In concluding his study of the Florida Indarctos, Wolff (1978: 11) stated that

there are ‘several cranial features [which] may indicate a relatively closer

relationship between Ailuropoda and Indarctos than with other bears, although

other possibilities exist’. In the light of the preceding discussion and the fact

that the Florida Indarctos skull is even more like that of A. africanum than

Ailuropoda, the ‘other possibilities’ must certainly include Agriotherium. The

similarities between the skulls of advanced Jndarctos, particularly the Florida

specimen, and Agriotherium are so great that it is inconceivable that they are

only distantly related. Wolff’s concluding remarks refer specifically to ‘great

enlargement of the cheek teeth’, ‘expansions of the zygomatic arches and

sagittal crests’, and the ‘absence of the alisphenoid canal’ as shared characters

indicative of the ‘close relationship’ between Jndarctos and Ailuropoda. The fact

that Indarctos and Agriotherium also share them is equally significant, especially

taken in conjunction with all the other evidence cited above.

It seems superfluous at this stage to consider alternative theories on the

origin of Agriotherium, but one is mentioned here since it apparently still has

some support.

The fact that Agriotherium and the Hemicyoninae have a premasseteric

fossa on the mandible in common has contributed to the belief that they are

closely related (e.g. Frick 1926). There is also a superficial similarity between the

dentitions of Agriotherium and hemicyonines, and both are supposedly canid-

like, rather than ursine-like. This is indeed so in the case of hemicyonines, but

in terms of the arrangement, morphology and size of individual cusps, the teeth

of Agriotherium are far more like those of advanced Indarctos than any hemi-

cyonine. It could, of course, be argued that the rather generalized hemicyonines

are structurally suitable in both cranial and postcranial characters to be ancestral

to Agriotherium. This more tenuous hypothesis is considered less likely than the

alternative suggested here.

It was indicated earlier that the known temporal ranges of Indarctos and

Agriotherium are in accord with the theory that the latter was derived from the

former. In the case of the Hemicyoninae—Agriotherium alternative the situation

is much less convincing because of an apparent, or actual, gap between the

recorded histories of the taxa concerned.

In Europe Agriotherium is first recorded from the Ruscinian, whereas

hemicyonines were extinct by the end of the Vindobonian (Table 20). This

94 ANNALS OF THE SOUTH AFRICAN MUSEUM

TABLE 20

The occurrences of Hemicyoninae and Agriotherium in Europe.

LOCALITY MAMMAL AGE! | MEIN ZONE?

Ruscinian

MeL ka

Moc ee

Hemicyon sansaniensis, Wintershof- Burdigalian

Hemicyon goeriachensis | West, Sansan, to

etc.® Goriach, La Vindobonian

Grive-St-Alban,

APPROX,

Agriotherium insigne

Steinheim etc®

1Van Couvering 1972, Fahibusch 1976; * Mein 1975, Fahlbusch 1976; *Heizmann 1973.

means that there was a period of about 8 m.y. for which no possible inter-

mediate between hemicyonines and Agriotherium is known in Europe. Much the

same applies in Asia and North America. It is unlikely that this could be due

to a defective fossil record, since the period in question covers the Vallesian

and Turolian (and their equivalents), of which the faunas are moderately to

very well known. It is possible that the Hemicyoninae—Agriotherium transition

took place in Africa, where 5-13 Ma faunas are poorly known, but negative

evidence is hardly convincing support for a theory.

It is much more likely that the characters Hemicyoninae and Agriotherium

have in common are due to convergent evolution. Parallel and convergent

evolution often complicate studies of relationships, and the Ursidae are a group

where this complication definitely exists. Early in the history of the Ursidae,

the Amphicyoninae and Hemicyoninae evolved along parallel lines, while later

the same applied to the Ursinae and Tremarctinae. Agriotherium is of particular

interest in this connection, since not only is it intermediate in age between

amphicyonines/hemicyonines and ursines/tremarctines, but it has in certain

respects paralleled members of both sets of subfamilies.

The superficial similarities between Agriotherium and large amphicyonines

and hemicyonines (e.g. Amphicyon major, Dinocyon thenardi) are striking

enough to suggest that these animals were ecological vicars. On the other hand,

Agriotherium, A. major and D. thenardi (and better known hemicyonines) also

differ from one another in certain respects, which can be explained by their

having evolved at different times from different ancestors.

Amongst the Ursinae, Thalarctos has now been mentioned several times

as having evolved characters comparable to some in Agriotherium. In this

instance there is no possibility of the shared characteristics being due to a close

relationship. They are simply explained by the fact that both Agriotherium and

Thalarctos are essentially carnivorous forms which evolved from omnivorous

ancestors (Hendey 1972: 122). The relationship between Thalarctos and the

Ursus arctos group has been well documented (Thenius 1953; Kurtén 1964),

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 95

and is here regarded as a parallel of the relationship between advanced Indarctos

and Agriotherium. A more detailed search for parallelisms between these two

sets of taxa may well be worth while. The obvious differences between Agrio-

therium and Thalarctos are not unexpected, since their ancestral forms are

markedly different from one another. In addition, Thalarctos has as yet had a

relatively brief history compared with that of Agriotherium. Given time it would

no doubt become increasingly distinct from U. arctos, and perhaps become

even more Agriotherium-like.

Agriotherium and advanced Indarctos are also paralleled in some respects

by tremarctines, especially the large extinct species. For example, several

references have been made above to similarities between the skull and skeleton

of A. africanum and Arctodus simus. Merriam & Stock (1925: 5) found that

there are certain characters ‘in which Tremarctos and [Arctodus] show distinctly

closer affinity to [Agriotherium] and its allies of the Pliocene than is seen in

Pleistocene and Recent bears of the genus Ursus’. Others have thought this

significant. For example, Erdbrink (1953) suggested that Tremarctos was

closely related to the ursines, while the larger tremarctines were derived from

advanced Jndarctos. Kurtén (1966: 7) disagreed, and concluded that although

the ‘earlier history of Arctodus is poorly documented . . . there can be little

doubt that it is a tremarctine, and not a member of the Agriotheriinae’.

The Indarctos—Agriotherium lineage was but one of several evolutionary

developments amongst the Ursidae during the latter part of their history. As

indicated above, it was a development which paralleled that undergone earlier

by large amphicyonines and hemicyonines, but it, too, was ultimately unsuc-

cessful and by the end of the Tertiary Agriotherium was on the verge of extinc-

tion, if not already extinct. It had no descendants.

Other later ursid lineages, however, were more successful. This applies

particularly in the case of the one which gave rise to the Ursinae, which under-

went a spectacular radiation beginning in the Pliocene. As with Indarctos—

Agriotherium, the Ursinae also stemmed from Ursavus, and Thenius (1977 and

earlier papers) has suggested U. ehrenbergi as the likely ancestral form.

The Tremarctinae were less successful than the Ursinae, having been con-

fined to the Americas, and eventually being supplanted in North America by

the Ursinae (Kurtén 1966, 1967). Their origins were recently considered by

Thenius (1976), who suggested that they, too, stemmed from an Ursavus.

Unfortunately, little is known of early tremarctines, but no doubt more will

yet be learnt of the origins and early history of this group.

The origin of the giant panda, Ailuropoda melanoleuca, has long been a

controversial issue. The earlier views that Ailuropoda is an ursid and not a

procyonid, and that it had an agriotheriine ancestor (Hendey 1972), are main-

tained here, although it now appears that it may have stemmed from Ursavus

and not /ndarctos as previously suggested.

As indicated earlier, the skulls of Ailuropoda and advanced Indarctos have

many characters in common which suggest that they are more closely related

96 ANNALS OF THE SOUTH AFRICAN MUSEUM

to one another than to other bears. However, the latter had already undergone

specializations in the direction of Agriotherium which render it structurally

unsuitable as a stem form for Ailuropoda. For example, advanced Jndarctos

has reduced anterior premolars, while those of Ai/uropoda are unusually large

for an ursid. This objection does not apply in the case of earlier, unspecialized

Indarctos (see Hendey 1972, table 1), but since such species have much in

common with Ursavus, the latter must also be taken into account when con-

sidering the origin of Ailuropoda.

One of the less well-known species of Ursavus is U. depereti of the European

Turolian, which is in some respects atypical of the genus (see Heizmann 1973).

This applies particularly to the P*, and it is the unusual characteristics of this

tooth which suggest that U. depereti may have a direct phylogenetic connection

with Ailuropoda.

Casts of upper teeth of U. depereti from Soblay in France were recently

examined in the Naturhistorisches Museum, Basel. The originals are in Lyon

and were described by Viret (1949) and Viret & Mazenot (1949). Two P?’s are

represented and they are remarkable for two reasons. Firstly, unlike the P*’s

in other Ursavus species, the Soblay specimens have large parastyles. This cusp

is proportionately even larger than those in advanced Indarctos and in Agrio-

therium, in which the parastyles are overshadowed by large and high-crowned

paracones. The large parastyle and relatively low-crowned paracone of U.

depereti are reminiscent of the situation in Ailuropoda, and differ from that in

all other ursids. Secondly, the U. depereti P* has an enlarged, double-cusped

protocone lobe, which differs from that in Indarctos and Agriotherium in being

more anteroposteriorly elongated and regular in occlusal outline. In addition,

in the Soblay specimen, AA52 (Viret & Mazenot 1949, pl. 1 (fig. 6)), both cusps

are ridge-like rather than pointed and are of more or less equal size. In all these

respects the U. depereti P* protocone lobe is Ailuropoda-like, and once again

it differs from that in all other ursids.

The M? and M? of U. depereti are also basically similar in morphology to

their counterparts in Ailuropoda. However, the upper molars are also similar

to those of other Ursavus species and Jndarctos, and are thus less significant

in indicating a relationship with Ailuropoda. A possibly important charac-

teristic of the M? figured by Viret & Mazenot (1949, pl. 1 (fig. 4)) is the rugose

enamel of the occlusal surface, which could foreshadow the ‘richly tuberculate’

condition of the Ailuropoda M? (Davis 1964: 127).

The M, and M, of U. depereti also have the basic morphology charac-

teristic of other Ursavus and Indarctos, although the M, is distinct, and more

Ailuropoda-like, in having the metaconid more anteriorly situated (Schlosser

1902, pl. 2 (figs 20, 23)).

Slender though the evidence may be, U. depereti appears to be structurally

better suited than any other recorded fossil ursid to fill the role of ancestor to

Ailuropoda. In addition, it is also temporarily and geographically well suited to

this role since it is from the continent on which the early evolution of living

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 97

bears was centred and it dates from a period when the radiation of later bears

was just beginning. The absence of any intermediate forms may simply be due

to the relatively poor Pliocene record in Asia, the continent to which Ailuropoda

melanoleuca is confined (Chorn & Hoffmann 1978). Relevant fossils of Plio-

cene age, as well as better knowledge of U. depereti, are required to test this

theory of Ailuropoda origins.

While the ultimate origins of both Ailuropoda and the Indarctos—Agrio-

therium lineage are uncertain, it is evident that these genera are more closely

related to one another than to the Ursinae and Tremarctinae.

The radiation of the Ursidae, like that of some other mammalian families,

was thus characterized by repetitious evolutionary developments, and the

correct identification of relationships requires careful study of the fossil record.

Ignorance of the details of this record may well lead to misinterpretation

of the significance of similar, but independently evolved characteristics in

ursid taxa. Kurtén (1967: 5) has found, for example, the analogy between

European Miocene Jndarctos and North American Pleistocene Arctodus to be

‘truly astonishing and an example of the déja vu experiences so familiar to the

student of fossil bears’.

The relationships suggested here are indicated in Figure 41. The named

late Tertiary species are all from Europe and consequently the European sub-

divisions of this period are used.

URSINI TREMARCTINI AILUROPODINI

HOLOCENE & (Ursus, Thalarctos, Euarctos, Helarctos.

‘

PLEISTOCENE Melursus)

‘Ursavus’ Indarctos

x |

depereti atticus

TUROLIAN Nas

Ursavus sp. Ursavus

(Soblay) ehrenbergi ,

’ A

| as

URSAVINAE Ursavus is Indarctos Indarctos

primaevus +: anthracitis arctoides

Ursavus 4, Indarctos

brevirhinus Z soras- ss ViretL:

ee areas

VINDOBONIAN ek

Ursavus

brevirhinus

HEMICYONINAE eee

elmensis

Foueocene | AMPHICYONINAE oN

Cephalogale

Fig. 41. Suggested relationships of some Ursidae.

(Plionarctos. Arctodus, (Ailuropoda)

Tremarctos)

Agriotherium

insigne

AGRIOTHERIINAE

PLIOCENE

MIOCENE

98 ANNALS OF THE SOUTH AFRICAN MUSEUM

NOMENCLATURE

The suprageneric classification of the Ursidae adopted here (Table 21) is a

modification of an arrangement suggested earlier (Hendey 1972: 119), and is

based on the phylogenetic relationships indicated in Figure 41.

TABLE 21

A classification of the Ursidae.

SUBFAMILY TRIBE GENERA, SPECIES

Amphicyoninae . : ; Not subdivided t Amphicyon, + Cynelos and tothers

Hemicyoninae . ; 4 : Not subdivided {| Hemicyon, | Dinocyon and t others

Ursavinae Not subdivided t Ursavus

: a Agriotheriini {Indarctos, + Agriotherium

Agriotheriinae . . . . : — - - —

Ailuropodini t‘Ursavus’ depereti, Ailuropoda

Weise Tremarctini {Plionarctos, + Arctodus, Tremarctos

Ursini Ursus, Thalarctos, Euarctos,

Helarctos, Melursus

+ = extinct

In preceding sections of this report the generally accepted practice of

allowing the ursines and tremarctines subfamilial status has been followed.

With this as a precedent, it follows that the agriotheriines and Ailuropoda each

warrant similar status. Some of those who have included Ailuropoda in the

Ursidae have, indeed, placed it in a separate subfamily (e.g. Pilgrim 1932).

Similarly, the agriotheriines have been widely recognized as a valid subfamily

(e.g. Kurtén 1966).

A disadvantage with this arrangement is that it does not indicate the

apparently close relationships between the ursines and tremarctines on the

one hand (Kurtén 1966, 1967; Thenius 1976), and the agriotheriines and

Ailuropoda on the other (Hendey 1972; Wolff 1978; this report). This problem

is overcome by reducing the status of the four subfamilies to tribes, and then

separating them into two subfamilies, namely, the Agriotheriinae (Agrio-

theriini and Ailuropodini) and the Ursinae (Ursini and Tremarctini). Except

for the position of the Tremarctini, this was the arrangement suggested earlier

(Hendey 1972), and for which there has been some support (e.g. Chorn &

Hoffmann 1978).

There is also a difference now in the taxa constituting the Agriotheriinae.

Firstly, in the classification suggested here, Ursavus depereti is included in

the Ailuropodini, since it is regarded as the likely ancestor of Ailuropoda. This

means that it can no longer be referred to Ursavus, but must either be given a

new generic name, or be referred to Ailuropoda itself. The latter course is

favoured here, but in view of the uncertainties still surrounding this species

neither of these alternatives is followed, and the doubtful generic identity of

‘Ursavus’ depereti is indicated by quotation marks.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 99

Secondly, if the Ursini, Tremarctini and Ailuropodini did, indeed, stem

from the Ursavus group and are valid tribes, then the Indarctos—Agriotherium

group of the Agriotheriini must be given the same status. This means that

Ursavus, which is generally regarded as an agriotheriine (e.g. Kurtén 1966;

Hendey 1972), has to be excluded from the tribe.

The various species of Ursavus, excluding ‘Ursavus’ depereti, are accordingly

placed in a new subfamily, the Ursavinae. Since this subfamily is not established

on the same criteria as the others, it constitutes an unsatisfactory element in

the classification as a whole. When their relationships are better understood,

it may be possible to assign the species of Ursavus to other tribes, just as ‘Ursavus’

depereti has been assigned to the Ailuropodini. The need for this additional

subfamily will then fall away.

The classification of more primitive ursids, or ursid-like carnivores, namely,

the hemicyonines and amphicyonines, was not investigated since they are

largely irrelevant to the present study. They are mentioned here for the sake of

completeness, but the conclusions regarding their status are tentative.

The Hemicyoninae are generally regarded as an early off-shoot of primitive

ursid stock (probably Cephalogale—see Erdbrink 1953, fig. 61; Hendey 1972,

fig. 1), and they are here included in the Ursidae.

The connection between the Amphicyoninae and later ursids is more

remote, and the classification of this group is more controversial. They have

been variously classified as a separate family (e.g. Hunt 1972), a subfamily of

the Canidae (e.g. Kuss 1965), and as an ursid subfamily (e.g. Ginsburg 1977).

The latter course is followed here.

The nomenclature of the species constituting the Agriotheriini is a far

more problematical matter than the suprageneric classification of the Ursidae.

The conclusion that Agriotherium is directly descended from Jndarctos

with certain specimens exhibiting a combination of ‘characteristics’ of both

genera, raises the possibility that they are congeneric, with Agriotherium the

senior synonym. This situation had been foreseen even before Indarctos became

well established in the literature (Pilgrim 1914; Merriam ef al. 1916). It can be

considered in relation to a similar situation involving the brown and polar bears.

Ewer (1973) and Van Valen (1978) have recently discussed the relative

merits of opposing opinions on the generic identity of the polar bear. According

to one widely held opinion the polar bear and brown bear are congeneric

(i.e. both belong to the genus Ursus), while the opposing view is that the polar

bear represents a separate genus (i.e. Thalarctos). Ewer and Van Valen favoured

the latter alternative, a view which is supported here. A problem with this

arrangement is that by one widely accepted criterion Thalarctos does not merit

separate generic status. Thalarctos maritimus and Ursus arctos are known to

produce fertile hybrids (see Van Gelder 1977). Van Valen (1978: 292) dismissed

this objection in the grounds that ‘lack of intersterility per se’ is of ‘low evolu-

tionary importance’ a fact for which Vrba (1979) has found supporting evidence

amongst the Bovidae.

100 ANNALS OF THE SOUTH AFRICAN MUSEUM

It is clear from their habitats and habits that Thalarctos and U. arctos are

set on distinct evolutionary paths, and that given time they may well warrant

generic separation by any standards. It is fortuitous that at present their diver-

gence is not far advanced, although there is no doubt that this divergence does,

in fact, exist. There will always be a problem in classifying species, both living

and extinct, which are on separate lineages, but which are still close to a common

ancestor. In instances where divergence is certain, and in the case of the brown

and polar bears it is, then classification by ‘clade’ rather than ‘grade’ is

preferable.

As indicated earlier, Thalarctos and Agriotherium are similar in the sense

that both are essentially carnivorous forms derived from omnivorous ancestors.

There is an apparent difference, however, in the evolutionary histories of these

two genera. In the case of Thalarctos there was a divergence from the U. arctos

group after which two lineages evolved independently, one with essentially

carnivorous elements, and the other continuing with essentially omnivorous

ones. In the case of Indarctos—Agriotherium it has been suggested above that

the essentially omnivorous ancestor gradually adapted to give rise to the more

carnivorous descendant forms and did not itself continue to exist as a separate

entity. In other words, successive members of what may be regarded as a single

lineage adapted their habits in response to a gradually changing environment

(see p. 70).

This may, of course, be a misinterpretation of the fossil record and there

may have been a dichotomy during the history of Indarctos—Agriotherium

similar to that of U. arctos-Thalarctos. For example, Crusafont & Kurtén (1976)

may be correct in their interpretation of the early history of Indarctos, and

I vireti may have given rise to both I. arctoides and the more Agriotherium-like

I. atticus. According to this interpretation J. vireti-I. arctoides wouid be an

equivalent of the U. arctos lineage, while J. atticus was the counterpart of

Thalarctos. The nomenclatural implication is that J. atticus and other later

Indarctos should be referred instead to Agriotherium, leaving only J. vireti,

I. arctoides and the aberrant J. anthracitis as representatives of Indarctos. This

arrangement is supported by the earlier suggestion that those species referred

to throughout this report as ‘advanced Indarctos’, could as well be regarded as

‘primitive Agriotherium’. A minor difficulty which arises is that Indarctos would

have to be replaced by another name, because the genotype (J. salmontanus)

would then be identified as an Agriotherium.

It could also be argued that since J. vireti is readily distinguishable from

contemporary Ursavus, the evolutionary changes which were to culminate in

Agriotherium were already being manifested early in the Vallesian. Thus J. vireti

relative to contemporary Ursavus was the counterpart of the present-day

situation involving Thalarctos and U. arctos. According to this interpretation

all species presently identified with Jndarctos should be referred instead to

Agriotherium.

Both the above arrangements are less arbitrary than the existing one in

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 101

which it is only the most advanced agriotheriines which are referred to Agrio-

therium. This is unsatisfactory because it does not reflect the realities of the

situation. Nevertheless, no changes in this arrangement are proposed here.

Undescribed Jndarctos and Agriotherium material from Asia and North America

is available, and a decision regarding the status of these genera is best left until

it, too, can be taken into account.

The same applies in the case of nomenclature at the species level, although

some changes which reflect opinions on relationships are proposed here.

An appropriate starting point is with the European species, since they are

amongst the best known and they represent the primary group from which

agriotheriines elsewhere evolved. Four late Miocene species are recognized as

valid. They are J. vireti, I. arctoides, I. atticus and I. anthracitis. I. atticus was

apparently the first species to spread into Asia. J. maraghanus (De Mecquenem

1925) and J. lagrelii (Zdansky 1924) were apparently broadly contemporaneous

with J. atticus and are here regarded as junior synonyms of this species.

The situation in respect of Siwaliks species is more problematical. The

identifications of Pilgrim (1932), although regarded as unsatisfactory in some

respects, are accepted here. Various possibilities will have to be considered in a

revision of the Siwaliks material. For example, J. punjabiensis and I. salmontanus

may represent a single species which is conspecific with J. atticus. Alternatively

this species, or perhaps only J. punjabiensis, may be more advanced. The possi-

bility that advanced Siwaliks Indarctos is conspecific with A. palaeindicum will

also have to be considered. A. sivalense will be mentioned below.

Advanced Jndarctos is also represented in Africa (Sahabi—see p. 73), and

in North America, where material has been identified as J. oregonensis (e.g.

Dalquest 1969), J. nevadensis (MacDonald 1959), or not identified as to species

(e.g. Wolff 1978). There is no reason to believe that recorded North American

Indarctos represents more than one species. Whether it should be identified as

I. oregonensis, or referred to one of the Old World species, is not certain. The

North American species was an immigrant from the Old World during the

Hemphillian (Repenning 1967), and must, therefore, be closely related to a

contemporary Asian species. It is here regarded as definitely distinct from

I. atticus, but if I. punjabiensis is a valid species, it might be conspecific with,

and the senior synonym of J. oregonensis.

Largely because of the uncertainties surrounding the Siwaliks species,

I. oregonensis is here retained as a distinct species. In addition, the possibility

was mentioned earlier that in certain respects North American /ndarctos evolved

independently of, and at a more rapid rate than its counterparts in Europe and

Africa (see p. 82). If this were a development peculiar to North America,

rather than both Asia and North America, then it would be another reason for

recognizing [. oregonensis as a distinct species.

The situation in respect of recorded species of Agriotherium is complex,

although the final solution with these species may be very simple. This solution,

which is not advocated yet, is for all species of Agriotherium to be regarded as

102 ANNALS OF THE SOUTH AFRICAN MUSEUM

junior synonyms of A. sivalense. Judged on the basis of the variation observed

in the Langebaanweg Agriotherium, there may be no size or morphological

grounds for recognizing more than one species of Agriotherium, However, it

was decided to draw at least some distinctions on a geographical basis.

European A. insigne, which is here taken to include A. intermedium (Stach

1957), is tentatively regarded as valid. It was suggested earlier that African

Agriotherium was an offshoot from the primary European lineage, and although

A. africanum is almost indistinguishable from A. insigne, they are not regarded

as conspecific, since the former is likely to be a descendant of north African

Indarctos and was thus probably only indirectly related to European A. insigne.

There are some distinctive features in the dentitions of recorded A. insigne

and A. africanum specimens which suggest that their phylogenetic connection

may, indeed, have been indirect. For example, the European A. insigne speci-

men, NMB-MP549, has a double-rooted P®, a condition unknown in the

Langebaanweg sample. On the other hand, no European specimen is known

which has an ‘intermediate cusp’ on the P* protocone lobe or a vestigial M,

metaconid, while none is known to lack the anterolingual cusp of M,. The

possibility that European and African Agriotherium evolved independently will

be discussed again below.

The situation in respect of Agriotherium in Asia and North America is

similar to that with the Indarctos of these continents. A. sivalense is definitely

a valid species, while A. palaeindicum is only tentatively regarded as such.

Chinese Agriotherium is probably referable to A. sivalense. There is almost

certainly only one species of Agriotherium represented in North America, and

this may also be referable to A. sivalense. However, the North American species

name which has priority, A. schneideri (Sellards 1916), is provisionally retained.

It has generally been assumed that North American Agriotherium, like

Indarctos, was an immigrant from Asia (e.g. Repenning 1967). This is clearly

a simple and logical interpretation of a situation where one genus supersedes

another closely related one on a continent known to have received immigrants

during the period in question. However, this is not the only interpretation

possible. The fact that North American Jndarctos includes some of the most

Agriotherium-like specimens known suggests that the Indarctos—Agriotherium

transition may have taken place in North America. If this were so then either

the subsequent dispersal of Agriotherium started in North America, or Agrio-

therium evolved independently from advanced Indarctos in North America as

well as in Europe and Africa.

The former alternative is unlikely in view of what is known of the Old

World history of agriotheriines, whereas the independent evolution theory is a

distinct possibility. This alternative appears to be implausible only in that it

involves the polyphyletic origin of one genus as it is presently conceived. If, as

indicated earlier, ‘advanced Jndarctos’ is interpreted instead as ‘primitive

Agriotherium’, it would mean that the Indarctos—Agriotherium transition took

place once only (probably in Europe), and that once this evolutionary course

103

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA

sisuauobaJo sojosepu

uapleuyos wnueyjouby

VOIHAWV HLYON

‘goeds pu SUII} UI TUIMIOYJONIZY JO JUOWOSURIIL SAIIEIUN} YW “Zp “SIT

eaceetes hae ee sv

snioesyjue sojosepu|

| sapiojose sojosepu]

‘ |

ears Snowe sojosepuy

sisuaigelund sojo1epuy

| é

wnoipulaejed wnvayjouby

‘ds sojosepul;

ANSOOIW ALV1

asuajeais wnueyjouby aubisul wnuayjouby

visvana

wnueowe wnueyjouby

VOldsv

ANSOONMd

104 ANNALS OF THE SOUTH AFRICAN MUSEUM

had been set the development of Agriotherium (senso stricto) could take place

irrespective of the geographical location of the populations concerned.

Thus European A. insigne may have been descended from ‘J.’ atticus,

A. africanum from the north African Jndarctos-like agriotheriine, A. sivalense

from the ‘J.’ punjabiensis/A. palaeindicum complex in Asia, while in North

America ‘J.’ oregonensis was ancestral to A. schneideri. It is nevertheless clear

that there was a close connection between advanced agriotheriines in Europe

and Africa on the one hand, and in Asia and North America on the other, and

the above scenario may therefore be unrealistic. Instead A. insigne and A. afri-

canum may share a common ancestor and be conspecific, while the same may

apply to A. sivalense and A. schneideri. These alternatives are indicated in

Figure 42. The possibility that all Agriotherium (senso stricto) evolved from a

single source is not favoured here, but it cannot yet be dismissed.

The nomenclature of this genus will ultimately have to be established by

further testing of the hypotheses presented here, and perhaps by reference to

conclusions reached in respect of other taxa with similar wide distributions

and generally uniform characteristics. Amongst the latter are the living brown

and black bears, as well as a wide variety of other carnivores, notably canids

and felids.

SUMMARY

Available material of Agriotherium africanum from the latest Miocene/early

Pliocene Varswater Formation at Langebaanweg is described. This species

was found to be as variable as other bears of comparable size, this being in part

due to marked sexual dimorphism.

The depositional environment and likely habitat and habits of the Lange-

baanweg Agriotherium are discussed. It is suggested that it was a large wood-

land carnivore, remains of which were transported to their points of discovery

by a river which then met the sea in the vicinity of Langebaanweg. The trans-

porting agent, and hyaenas, rodents, and fires contributed to the post-mortem

damage exhibited by specimens.

Agriotherium was evidently descended from late Miocene Jndarctos, the

more significant evidence in support of this hypothesis being as follows:

1. There was an increase of size with time in the /ndarctos—Agriotherium

lineage, and advanced Indarctos may be little or no different’in size to

Agriotherium.

2. The skull characters of advanced Jndarctos (e.g. I. atticus from Samos;

Indarctos from Florida) are shared by Agriotherium (e.g. the Langebaanweg

specimen L45062). Observable differences reflect further development of trends

already evident in the Indarctos lineage. Significant shared characteristics are a

relatively short and broad snout, with the posterior palatine incisure in line

with the M? metacones, large zygomata and sagittal crest, and similar basi-

cranial region, including absence of the alisphenoid canal.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 105

3. There is little or no difference in the dentitions of advanced Indarctos

and Agriotherium. For example, the incisors, canines, P,’s, M?’s, M,’s and M,’s

may be morphologically indistinguishable. Differences that do exist reflect

trends already evident in the Jndarctos lineage. For example:

(i) There was a progressive reduction in the size and number of anterior

premolars in the Indarctos—Agriotherium lineage, with their principal cusps

becoming progressively lower crowned and indistinct. However, Agrio-

therium sometimes has Indarctos-like upper premolars (e.g. A. insigne,

NMB-MP549), while Indarctos may have Agriotherium-like lower anterior

premolars (e.g. J. atticus, NNUW-Samos 1912, 29).

(ii) The shearing elements of the carnassials in the Indarctos—Agriotherium

lineage are progressively developed, as are the buccal cusps of M! and M3,

which also have a shearing function. There is a corresponding reduction

in the crushing elements of the posterior cheek teeth. For example, while

the M2? of Indarctos is distinguished from that of Agriotherium by the

presence of a talon, this feature is progressively reduced in time. In advanced

Indarctos it may be small (e.g. I. punjabiensis, GSI-D12), while vestiges of

a talon may be present in Agriotherium (e.g. the Langebaanweg specimen

L47698).

(iti) As a general rule the P* of Agriotherium is distinguished from that of

Indarctos by the presence of a parastyle, but this cusp may be present in

advanced Indarctos (e.g. Samos I. atticus; most North American specimens).

4. A double-cusped P* protocone lobe is characteristic of European

Indarctos, and is also found in some European and African Agriotherium

specimens (e.g. A. insigne, NMB-MP549; Langebaanweg L2045 and others),

although the P* antero-internal cusp is sometimes absent (e.g. A. insigne,

Gervais (1859) specimen; Langebaanweg, L41404). The fact that this cusp is

sometimes absent in A. insigne, and that in those specimens postdating L2045

it is reduced or absent, indicates that there was a tendency in European—African

Agriotherium to lose the antero-internal cusp. A similar tendency is evident in

advanced Asian and North American Indarctos (e.g. I. punjabiensis, GSI-D6;

Florida Jndarctos), while the antero-internal cusp is always absent in the

Agriotherium from these continents. This common tendency in Jndarctos and

Agriotherium indicates a close relationship, while the examples cited suggest

that the European/African and Asian/North American Indarctos—Agriotherium

lineages evolved independently and at different rates later in their history.

The primary /ndarctos—Agriotherium lineage was European and comprised

I. vireti, I. arctoides, I. atticus and A. insigne, with I. anthracitis as an aberrant,

probably island-dwelling offshoot. Indarctos emigrated to Africa, probably late

in the Miocene, and an Agriotherium (A. africanum) very similar to the European

A. insigne was evolved on this continent. Advanced Indarctos also migrated

eastwards and the Asian/North American lineage is comprised of species of

106 ANNALS OF THE SOUTH AFRICAN MUSEUM

Indarctos and Agriotherium postdating J. atticus. Pending further studies, most

existing species names of these genera are retained.

It is suggested that either all /ndarctos, or all later Indarctos (I. atticus

and younger species) should be referred instead to Agriotherium, although the

status quo is maintained for the present.

A modified suprageneric classification of later Ursidae is proposed. The

species of Ursavus are included in a new subfamily, the Ursavinae. The Agrio-

theriinae are divided into two tribes, namely, Agriotheriini (Indarctos—Agrio-

therium) and the Ailuropodini (‘Ursavus’ depereti—Ailuropoda melanoleuca).

The Ursinae and Tremarctinae are reduced to the status of tribes within a

reconstituted subfamily Ursinae.

ACKNOWLEDGEMENTS

I am greatly indebted to the many persons who assisted directly and

indirectly in the study of the Langebaanweg Agriotherium. They include

Drs G. de Beaumont (Geneva), B. Engesser (Basel), A. W. Gentry (London),

L. Ginsburg (Paris), J. Jewell (London), C. A. Repenning (Menlo Park, Cali-

fornia) and E. Thenius (Vienna) who provided access to, and/or casts of,

specimens in their care. Drs A. J. Tankard (Knoxville, Tennessee) and R. H.

Tedford (New York) provided information on geological and palaeontological

matters respectively. Almost all the photographs of specimens were made by

Miss J. Nolte, while Mr A. Byron printed Figures 27 and 32. Miss L. Scott

prepared Figure 34, Mrs P. Eedes typed the manuscript, and the South African

Air Force provided the aerial photograph for Figure 26. Drs R. G. Klein

(Chicago) and B. Kurtén (Helsinki) reviewed early drafts of the manuscript.

The Langebaanweg Research Project is supported by Chemfos Ltd, the

South African Council for Scientific and Industrial Research, and the Wenner—

Gren Foundation for Anthropological Research (Grant no. 2752-1834), and

the assistance of these organizations is gratefully acknowledged.

REFERENCES

BERGGREN, W. A. & VAN COUVERING, J. A. 1974. The Late Neogene. Amsterdam: Elsevier.

Boaz, N. T., GAziry, A. W. & EL-ARNAUTI, A. 1979. New fossil finds from the Libyan Upper

Neogene Site of Sahabi. Nature, Lond. 280: 137-139.

BuTzer, K. W. 1971. Environment and archaeology. Chicago: Aldine.

Cuorn, J. & HOFFMANN, R. S. 1978. Ailuropoda melanoleuca, Mammalian Species 110: 1-6.

CoETZEE, J. A. 1978. Late Cainozoic palaeoenvironments of southern Africa. In: VAN ZINDEREN

BAKKER, E. M. ed. Antarctic glacial history and world palaeoenvironments: 115-127.

Rotterdam: Balkema.

CRUSAFONT, M. & KurTEN, B. 1976. Bears and bear-dogs from the Vallesian of the Vallés-

Penedés Basin, Spain. Acta. zool. fenn. 144: 1-29.

Cuvier, C. 1822. Recherches sur les ossemens fossiles. Paris: E. D’Ocagne.

CypertT, E. 1961. The effects of fires in the Okefenokee Swamp in 1954 and 1955. Am. Midl.

Naturalist 66: 485-503.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 107

DA Quest, W. W. 1969. Pliocene carnivores of the Coffee Ranch. Bull. Texas Mem. Mus. 15:

1-43.

Davis, D. D. 1955. Masticatory apparatus in the spectacled bear Tremarctos ornatus. Fieldiana,

Zool. 37: 25-46.

Davis, D. D. 1964. The giant panda. Fieldiana, Zool. Mem. 3: 1-339.

De MECQUENEM, R. 1925. Contribution a l’étude des fossiles de Maragha. Ann. Paléont. 13/14:

135-160, 1-36.

DINGLE, R. V., Lorp, A. R. & HENDEY, Q. B. 1979. New sections in the Varswater Formation

(Neogene) of Langebaan Road, south-western Cape. Ann. S. Afr. Mus. 78: 81-92.

ERDBRINK, D. P. 1953. A review of fossil and Recent bears of the Old World. Deventer: Jan

de Lange.

Ewer, R. F., 1973. The carnivores. London: Weidenfeld & Nicolson.

FAHLBUSCH, V. 1976. Report on the International Symposium on mammalian stratigraphy of

the European Tertiary. Newsl. Stratigr. 5: 160-167.

FALCONER, H. & CAUTLEY, T. P. 1836. On Ursus sivalensis, a new fossil bear from the Sewalik

Hills. As. Res. 19: 193-200.

ForstEn, A. 1978. Hipparion and possible Iberian—North African Neogene connections.

Ann. zool. fenn. 15: 294-297.

Frick, C. 1926. The Hemicyoninae and an American Tertiary bear. Bull. Am. Mus. nat. Hist.

54: 1-119.

Gentry, A. W. 1980. Fossil Bovidae (Mammalia) from Langebaanweg, South Africa. Ann.

S. Afr. Mus. 79. (In press.)

GerVAIS, P. 1859. Zoologie et Paléontologie francaises. Paris: A. Bertrand.

GinsBurG, L. 1961. Plantigradie et digitigradie chez les Carnivores fissipédes. Mammalia 25:

1-21.

GinspurG, L. 1977. Cynelos lemanensis (Pomel), carnivore ursidé de l’Aquitanien d’Europe.

Ann. Paléont. (Vertébrés) 63: 57-104.

Grecory, W. K. 1936. On the phylogenetic relationships of the giant panda (Ailuropoda) to

other arctoid Carnivora. Am. Mus. Novit. 878: 1-29.

Guturig, R. D. 1972. Re-creating a vanishing world. National Geographic 141: 294-301.

HEIZMANN, E. P. J. 1973. Die Carnivoren des Steinheimer Beckens. B. Ursidae, Felidae,

Viverridae sowie Erganzungen und Nachtrage zu den Mustelidae. Palaeontogr. Suppl.

8 (GB): 1-95.

HELBING, H. 1932. Ober einen Jndarctos-schadel aus dem Pontien der Insel Samos. Abh.

schweiz. palaeont. Ges. 52: 1-18.

HENDEY, Q. B. 1970. A review of the geology and palaeontology of the Plio/Pleistocene

deposits at Langebaanweg, Cape Province. Ann. S. Afr. Mus. 56: 75-117.

HENDEY, Q. B. 1972. A Pliocene ursid from South Africa. Ann. S. Afr. Mus. 59: 115-132.

HENDEY, Q. B. 1973. Fossil occurrences at Langebaanweg, Cape Province. Nature, Lond. 244:

13-14.

HENDEY, Q. B. 1974. The late Cenozoic Carnivora of the south-western Cape Province.

Ann. S. Afr. Mus. 63: 1-369.

HENDEY, Q. B. 1976. The Pliocene fossil occurrences in ‘E’ Quarry, Langebaanweg, South

Africa. Ann. S. Afr. Mus. 69: 215-247.

HENDEY, Q. B. 1977. Fossil bear from South Africa. S. Afr. J. Sci. 73: 112-116.

HENDEY, Q. B. 1978. Late Tertiary Mustelidae (Mammalia, Carnivora) from Langebaanweg,

South Africa. Ann. S. Afr. Mus. 76: 329-357.

Hoover, D. A. 1972. A late Pliocene rhinoceros from Langebaanweg, Cape Province. Ann.

S. Afr. Mus. 59: 151-191.

Hunt, R. M. 1972. Miocene amphicyonids (Mammalia, Carnivora) from the Agate Springs

Quarries, Sioux County, Nebraska. Am. Mus. Novit. 2506: 1-39.

Hunt, R. M. 1977. Basicranial anatomy of Cynelos Jourdan (Mammalia: Carnivora), an

Aquitanian amphicyonid from the Allier Basin, France. J. Palaeont. 51: 826-843.

KENNETT, J. P., Houtz, R. E., ANDREWS, P. B., EDWARDS, A. R., GosTIN, V. A., HANos, M.,

HAmpToNn, M.., JENINS, D. G., MARGOLIS, S. V., OVENSHINE, A. J., & PERCH-NIELSEN, K.

1975. Cenozoic paleo-oceanography in the southwest Pacific Ocean, Antarctic glaciation,

and development of the Circum-Antarctic Current. In: KENNETT, J. P. et al. eds. Initial

reports of the DSDP 29: 1155-1169.

108 ANNALS OF THE SOUTH AFRICAN MUSEUM

Kretzot, M. 1942. Zwei neue Agriotheriiden aus dem ungarischen Pannon. Fold. Kézlény 72:

350-353.

KurTEN, B. 1952. The Chinese Hipparion fauna. Commentat. biol. 13: 1-82.

KurtTEn, B. 1955. Sex dimorphism and size trends in the cave bear, Ursus spelaeus Rosenmiiller

and Heinroth. Acta zool. fenn. 90: 1-48.

KurtTEN, B. 1964. The evolution of the polar bear, Ursus maritimus Phipps. Acta zool. fenn.

108: 1-30.

KurTEN, B. 1966. Pleistocene bears of North America. 1. Genus Tremarctos, spectacled bears.

Acta zool. fenn. 115: 1-120.

KurTEN, B. 1967. Pleistocene bears of North America. 2. Genus Arctodus, short-faced bears.

Acta zool. fenn. 117: 1-60.

KurtEN, B. 1968. Pleistocene mammals of Europe. London: Weidenfeld & Nicolson.

KourtEn, B. 1971. The Age of Mammals. London: Weidenfeld & Nicolson.

Kuss, S. E. 1965. Revision der europaischen Amphicyoninae (Canidae, Carnivora, Mamm.)

ausschlieflich der voroberstampischen Formen. Sitz. heidelb. Akad. Wiss. 1965 (1): 1-168.

LYDEKKER, R. 1884. Indian Tertiary and post-Tertiary Vertebrata. Siwalik and Narbada

Carnivora. Palaeont. indica (10) 2: 178-354. é

MacDOonaL_Lb, J. R. 1959. The Middle Pliocene mammalian fauna from Smiths Valley, Nevada.

J. Palaeont. 33: 872-887.

MATTHEW, W. D. 1929. Critical observations upon Siwalik mammals. Bull. Am. Mus. nat.

Hist. 56: 437-560.

Mein, P. 1975. Proposition de biozonation du Neogene Mediterraneen a partir des mammi-

féres, Trab. Neogeno-Cuaternario Madrid 4: 112-113.

MERRIAM, J. C., Stock, C. & Moopy, C. L. 1916. An American Pliocene bear. Univ. Calif.

Publ. Bull. Dept. Geol. 10: 87-109.

MERRIAM, J. C., Stock, C. & Moopy, C. L. 1925. The Pliocene Rattlesnake Formation and

fauna of Eastern Oregon, with notes on the geology of the Rattlesnake and Mascall

Deposits. Carnegie Inst. Wash. Publ. 347: 45-92.

MerRIAM, J. C. & Stock, C. 1925. A hyaenarctid bear from the later Tertiary of the John Day

Basin of Oregon. Carnegie Inst. Wash. Publ. 347: 41-44.

Owen, R. 1845. Odontography. London: Bailliere.

PILBEAM, D., BARRY, J., MEYER, G. E:, IBRAHIM SHAH, S. M., PICKFORD, M. H. L., BisHop,

W. W., THOMAS, H., & JAcoss, L. L. 1977. Geology and palaeontology of Neogene strata

of Pakistan. Nature, Lond. 270: 684-689.

Pitcrim, G. E. 1913. The correlation of the Siwaliks with mammal horizons of Europe.

Rec. geol. Surv. India. 43: 264-326.

Pitcrim, G. E. 1914. Further description of Indarctos salmontanus Pilgrim, a new genus of

bear from the Middle Siwaliks, with some notes on the fossil Indian Ursidae. Rec. geol.

Surv. India 44: 225-233.

PitGrim, G. E. 1931. Catalogue of the Pontian Carnivora of Europe. London: British Museum

(Natural History).

PILGRIM, G. E. 1932. The fossil Carnivora of India. Palaeont. indica (n.s.) 18: 1-232.

REPENNING, C. A. 1967. Palaearctic-Nearctic mammalian dispersal in the Late Cenozoic. In:

Hopkins, D. M. ed. The Bering Land Bridge: 288-311. Stanford: University Press.

ScHLosseR, M. 1899. Ueber die Baren und barenahnlichen Formen des europaischen Tertiars.

Palaeontogr. 46: 95-146.

ScHLOossER, M. 1902. Beitrage zur Kenntnis der Séugetiere aus den siiddeutschen Bohnerzen.

Geol. paldont. Abh. 9: 117-258.

SELLARDS, E. H. 1916. Fossil vertebrates from Florida: a new Miocene fauna; new Pliocene

species; the Pleistocene fauna. Florida State geol. Surv., 8th Ann. Rep., 77-121.

StEsserR, W. G. 1978. Aridification of the Namib Desert: Evidence from oceanic cores. In:

VAN ZINDEREN BAKKER, E. M. ed. Antarctic glacial history and world palaeoenvironments:

105-113. Rotterdam: Balkema.

STACH, J. 1957. Agriotherium intermedium n.sp. from the Pliocene bone breccia of Weze.

Acta Palaeont. polonica 2: 1-18.

Straus, J. R. & CouEN, A. D. 1979. The Snuggedy Swamp of South Carolina: A back-barrier

estuarine coal-forming environment. J. sedim. Petrol. 49: 133-143.

STEHLIN, H. G. 1907. Les Types du ‘Lophiodon de Montpellier’ de Cuvier (Hyaenarctos insignis

P. Gery.). Bull. Soc. geol. Fr. (4) 7: 219-223.

AGRIOTHERIUM FROM LANGEBAANWEG, SOUTH AFRICA 109

TALBOT, W. J. 1947. Swartland and Sandveld. Cape Town: Oxford University Press.

TANKARD, A. J. 1975. Varswater Formation of the Langebaanweg-Saldanha area, Cape

Province. Trans. geol. Soc. S. Afr. 77: 265-283.

TANKARD, A. J. & RoGers, J. 1978. Late Cenozoic palaeoenvironments on the west coast of

southern Africa. J. Biogeog. 5: 319-337.

THENIUS, E. 1949. Uber die GehGrregion von Indarctos (Ursidae, Mamm.). Sber. mathem.-

naturw. Kl., Akad. Wiss. Wien 158: 647-653.

THENIUS, E. 1953. Zur Analyse des Gebisses des Eisbaren, Ursus (Thalarctos) maritimus Phipps,

1774. Saugetierkundl. Mitteil. 1953: 1-7.

THENIUS, E. 1959. Indarctos arctoides (Carnivora, Mammalia) aus dem Pliozain Osterreichs

nebst einer Revision der Gattung. Neues Jb. Geol. Paldont. Abh. 108: 270-295.

THENIUs, E. 1976. Zur stammesgeschlichtlichen Herkunft von Tremarctos (Ursidae, Mam-

malia). Z. Saugetierk. 41: 109-114.

THENIUS, E. 1977. Zur systematischen Stellung von Protursus (Carnivora, Mammalia). Sitz.

mathem.-naturw. KI., Osterreich. Akad. Wiss. 1977 (3): 1-5.

THENIUS, E. & Horer, H. 1960. Stammesgeschichte der Sdugetiere. Berlin: Springer.

TosiENn, H. 1955. Neue und wenig bekannte Carnivoren aus den unterpliozénen Dinotherein-

sanden Rheinhessens. Notizbl. Hess. Landesamt. Bodenforsch. (6) 6: 7-31.

VAN COUVERING, J. A. 1972. Radiometric calibration of the European Neogene. Jn: BisHop,

W. W. & Mirter, J. A. eds. Calibration of hominoid evolution: 247-271. Edinburgh:

Scottish Academic Press.

VAN GELDER, R. 1977. Mammalian hybrids and generic limits. Am. Mus. Novit. 2635: 1-25.

VAN VALEN, L. 1978. Why not to be a cladist. Evol. Theory 3: 285-299.

VirET, J. 1939. Monographie paléontologique de Ja faune de vertébrés des Sables de Mont-

pellier. III. Carnivora Fissipedia. Trav. Lab. géol. Fac. Sc. Lyon 37: 1-26.

Viret, J. 1949. Observations complémentaires sur quelques mammiféres fossiles de Soblay.

Ecl. geol. Helv. 42: 469-476.

VireET, J. & MAzeENoT, G. 1949. Nouveaux restes de mammiféres dans le gisement de lignite

pontien de Soblay (Ain.). Ann. Paléont. 34: 1-42.

Visser, H. N. & Scuocn, A. E. 1973. The geology and mineral resources of the Saldanha Bay

area. Mem. geol. Surv. S. Afr. 63: 1-150.

VrBA, E. 1979. Phylogenetic analysis and classification of fossil and recent Alcelaphini Mam-

malia: Bovidae. Biol. J. Linn. Soc. 11: 207-228.

Wacner, A. 1837. Palaontologische Abhandlungen. Gel. Anz. bayer. Akad. Wiss. Miinchen 5:

334-335.

WEGNER, R. N. 1913. Tertiar und umgelagerte Kreide bei Oppeln (Oberschleisen). Palae-

ontogr. 60: 175-274.

Wesp, S. D. 1977. A history of savanna vertebrates in the new world. Part 1. North America.

Ann. Rey. Ecol. Syst. 8: 355-380.

Wo ttr, R. G. 1978. Function and phylogenetic significance of cranial anatomy of an early

bear (UIndarctos) from Pleistocene sediments of Florida. Carnivore 1 (3/4): 1-12.

Wo rr, R. G., SINGER, R. & BisHop, W. W. 1973. Fossil bear (Agriotherium Wagner 1837)

from Langebaanweg, Cape Province, South Africa. Quaternaria 17: 209-236.

ZDANSKY, O. 1924. Jungtertiare Carnivoren Chinas. Palaeont. sinica (C) 2: 1-149.

| 6. SYSTEMATIC papers must conform to the /nternational code of zoological nomenclature

| (particularly Articles 22 and 51). :

| Names of new taxa, combinations, synonyms, etc., when used for the first time, must be

| followed by the appropriate Latin (not English) abbreviation, e.g. gen. noy., sp. nov., comb.

| novy., Syn. NOv., etc.

An author’s name when cited must follow the name of the taxon without intervening

| punctuation and not be abbreviated; if the year is added, a comma must separate author’s

| name and year. The author’s name (and date, if cited) must be placed in parentheses if a

| species or subspecies is transferred from its original genus. The name of a subsequent user of

a scientific name must be separated from the scientific name by a colon.

Synonymy arrangement should be according to chronology of names, i.e. all published

| scientific names by which the species previously has been designated are listed in chronological

| order, with all references to that name following in chronological order, e.g.:

Family Nuculanidae

Nuculana (Lembulus) bicuspidata (Gould, 1845)

| Figs 14-1SA

| Nucula (Leda) bicuspidata Gould, 1845: 37.

Leda plicifera A. Adams, 1856: 50.

Laeda bicuspidata Hanley, 1859: 118, pl. 228 (fig. 73). Sowerby, 1871: pl. 2 (fig. 8a—b),

Nucula largillierti Philippi, 1861: 87.

| Leda bicuspidata: Nicklés, 1950: 163, fig. 301; 1955: 110. Barnard, 1964: 234, figs 8-9.

Note punctuation in the above example:

comma separates author’s name and year

“semicolon separates more than one reference by the same author

full stop separates references by different authors

figures of plates are enclosed in parentheses to distinguish them from text-figures

dash, not comma, separates consecutive numbers

Synonymy arrangement according to chronology of bibliographic references, whereby

the year is placed in front of each entry, and the synonym repeated in full for each entry, is

not acceptable.

In describing new species, one specimen must be designated as the holotype; other speci-

mens mentioned in the original description are to be designated paratypes; additional material

| not regarded as paratypes should be listed separately. The complete data (registration number,

depository, description of specimen, locality, collector, date) of the holotype and paratypes

must be recorded, e.g. :

| Holotype

| SAM-A13535 in the South African Museum, Cape Town. Adult female from mid-tide region, King’s Beach

| Port Elizabeth (33°51’S 25°39’E), collected by A. Smith, 15 January 1973. 3

Note standard form of writing South African Museum registration numbers and date.

|7. SPECIAL HOUSE RULES

Capital initial letters

(a) The Figures, Maps and Tables of the paper when referred to in the text

| e.g. *... the Figure depicting C. namacolus...’; *. . . in C. namacolus (Fig. 10)...’

|(b) The prefixes of prefixed surnames in all languages, when used in the text, if not preceded

by initials or full names

e.g. Du Toit but A.L.du Toit; Von Huene but F. von Huene

|(c) Scientific names, but not their vernacular derivatives

| e.g. Therocephalia, but therocephalian

| Punctuation should be loose, omitting all not strictly necessary

| Reference to the author should be expressed in the third person

| Roman numerals should be conyerted to arabic, except when forming part of the title of a

| book or article, such as

‘Revision of the Crustacea. Part VIII. The Amphipoda.’

| Specific name must not stand alone, but be preceded by the generic name or its abbreviation

to initial capital letter, provided the same generic name is used consecutively.

| Name of new genus or species is not to be included in the title: it should be included in the

abstract, counter to Recommendation 23 of the Code, to meet the requirements of

Biological Abstracts.

Q. B. HENDEY

AGRIOTHERIUM (MAMMALIA, URSIDAE)

FROM LANGEBAANWEG, SOUTH AFRICA, AND

RELATIONSHIPS OF THE GENUS

E /

{ , OC

JOLUME 81 PART 2 FEBRUARY 1980 ISSN 0303-2515

ANNALS

OF THE SOUTH AFRIC

MUSEU

CAPE TOWN |

| .

INSTRUCTIONS TO AUTHORS

1. MATERIAL should be original and not published elsewhere, in whole or in part.

2. LAYOUT should be as follows:

(a) Centred masthead to consist of

Title: informative but concise, without abbreviations and not including the names of new genera or species

Author's(s’) name(s)

Address(es) of author(s) (institution where work was carried out)

Number of illustrations (figures, enumerated maps and tables, in this order)

(b) Abstract of not more than 200 words, intelligible to the reader without reference to the text

(d) Introduction

(e) Subject-matter of the paper, divided into sections to correspond with those given in table of contents