, 2239 (2001);
et al.Philip D. Gingerich,
Feet of Eocene Protocetidae from Pakistan
Origin of Whales from Early Artiodactyls: Hands and
www.sciencemag.org (this information is current as of November 20, 2007 ):
The following resources related to this article are available online at
version of this article at:
including high-resolution figures, can be found in the online
Updated information and services,
can be found at:
Supporting Online Material
63 article(s) on the ISI Web of Science.
This article has been
related to this article
A list of selected additional articles on the Science Web sites
8 articles hosted by HighWire Press; see:
This article has been
This article appears in the following
in whole or in part can be found at:
permission to reproduce
of this article or about obtaining
Information about obtaining
registered trademark of AAAS.
is aScience 2001 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the Science
on November 20, 2007
phase environments (20) and, we may as-
sume, were also abundant in the solar nebula.
The alternative possibility exists that aliphat-
ic hydrocarbons in the meteorite underwent
condensation and aromatization as in terres-
trial kerogens. However, such thermal alter-
ation would not be consistent with the hydro-
gen abundance of the material, which implies
an incomplete condensation of rings, the
finding in the extracts of labile molecules, as
nitriles, and the presence of a substantial
aliphatic phase in more petrologically pro-
cessed chondrites, as CVs (7).
Far from disappointing, the relative simplic-
ity of Tagish Lake organic content provides
insight into an outcome of early solar system
chemical evolution not seen so far. In particu-
lar, the finding of just one suite of organic
compounds matching those of Murchison and
of some (but not all) the carbonaceous phases
and compounds seen in other chondrites dem-
onstrates the presence of distinct organic syn-
thetic processes in primitive meteorites. It also
implies that the more complex organic matter
of heterogeneous chondrites may result from
multiple, separate evolutionary pathways.
References and Notes
1. J. R. Cronin, S. Chang, in The Chemistry of Life’s
Origins, J. M. Greenberg, C. X. Mendoza-Gomez, V.
Pirronello, Eds. (Kluwer Academic, Dordrecht, Neth-
erlands, 1993), pp. 209–258, and references therein.
2. C. F. Chyba, C. Sagan, Nature 355, 125 (1992).
3. P. G. Brown et al., Science 290, 320 (2000).
4. S. Pizzarello, G. W. Cooper, Meteoritic Planet. Sci. 36,
5. J. R. Cronin, S. Pizzarello, S. Epstein, R. V. Krishnamur-
thy, Geochim. Cosmochim. Acta 57, 4745 (1993).
6. J. R. Cronin, S. Pizzarello, Geochim. Cosmochim. Acta
54, 2859 (1990).
7. ????, J. S. Frye, Geochim. Cosmochim. Acta 51,
8. The compounds were analyzed by GC-MS as isopro-
pyl esters. The GC columns were ChyrasilDEX
(Chrompack; 25 m by 0.25 mm) or Chirasil-Val (All-
tech; 50 m by 0.25 mm). The typical program was
40°C initial, 5 min; ?100 at 2°/min; and ?200 at
9. GC-C-IRMS: HP 6890 GC, Finnigan high-temperature
conversion interphase III, and Mat Delta?-XL MS. GC
conditions were as for GC-MS (8). Oxidation was at
940°C with a ceramic oxidation reactor bearing NiO/
CuO/Pt wires. Standard CO2
?10.07‰ (VPDB). Data were analyzed with Finnigan
ISODAT software, with ? ? ?0.3‰ for peaks ?0.5 V.
10. ?85‰ to ?5‰, up to ?5‰ for methanogenic
11. These acids were analyzed with DB-17 column (60 m
by 0.25 mm; Agilent Technologies). Other ?13C val-
ues were ?7.7 ? 0.3‰ for a methyl-phthalimide
and ?5.9 ? 0.7‰, ?10.5 ? 0.8‰, and ?17.5 ?
0.4‰ for three dimethyl-phthalimides.
12. J. G. Lawless, G. U. Yuen, Nature 282, 396 (1979).
13. G. W. Cooper, M. H. Thiemens, T. Jackson, S. Chang,
Science 277, 1072 (1997), and references therein.
14. M. H. Studier, R. Hayatsu, E. Anders, Geochim. Cos-
mochim. Acta 36, 189 (1972).
15. Airborne contamination can be absorbed readily by
meteorites (28) and should be considered for Tagish
Lake, which was recovered a week after its fall.
INOVA400 spectrometer at 9.4 T with the VACP MAS
sequence, 50-kHz spectral width, 3.5-ms cross-polar-
ization contact time, recycle delay of 1.04 s, 6.7-?s
1H90 pulse width, 28,584 transients per spectrum,
55-kHz decoupling field, 16.0-kHz spinning rate, and
(six pulses) ?13C:
4-mm silicon nitride rotors with Torlon caps. Tagish
Lake repeat analyses were at 0.5-, 1.0-, and 2.0-ms
17. G. D. Cody, C. M. O’D Alexander, F. Tera, Lunar Planet.
Sci. 30 (abstract 1582) (1999) [CD-ROM].
18. A. Gardinier et al., Earth Planet. Sci. Lett. 184, 9
19. ?D values were determined by thermal conversion
(TC)/elemental analyzer (EA) interfaced to Finnigan
Delta?XL MS. Pyrolysis was at 1450°C to yield hy-
drogen gas quantitatively. Standardization was by
hydrogen gas pulse; sample analyses were repeated
20. L. J. Allamandola, S. A. Sanford, B. Wopenka, Science
237, 56 (1987).
21. J. F. Kerridge, S. Chang, R. Shipp, Geochim. Cosmo-
chim. Acta 51, 2527 (1987).
22. 1.3 ppm of fullerenes was isolated from the bulk
carbonaceous residues by extraction with 1,2,3,5 tet-
ramethylbenzene. The yield was incomplete, as for
other meteorites (23), because of limited organic
solubility of fullerenes.
23. L. Becker, R. J. Poreda, T. E. Bunch, Proc. Natl. Acad.
Sci. U.S.A. 97, 2979 (2000).
24. Helium within the extract was ?0.01% of total
helium in the bulk material. About 40% of this total
helium is released below 800°C, consistent with the
extracted fullerene carrier. No other carbon carrier
has been identified with similar low-temperature re-
25. E. T. Peltzer, J. L. Bada, Nature 272, 443 (1978).
26. I. Gilmour, M. A. Sephton, V. K. Pearson, Lunar Planet.
Sci. 32 (abstract 1969) (2001).
27. I. Gilmour, V. K. Pearson, M. A. Sephton, Lunar Planet.
Sci. 32 (abstract 1993) (2001).
28. J. Han, B. R. Simoneit, A. L. Burlingame, M. Calvin,
Nature 222, 364 (1969).
29. J. R. Cronin, S. Pizzarello, D. P. Cruikshank, in Mete-
orites and the Early Solar System, J. F. Kerridge, M. S.
Matthews, Eds. (Univ. of Arizona Press, Tucson, AZ,
1988), pp. 819–8857.
30. G. W. Cooper, J. R. Cronin, Geochim. Cosmochim.
Acta 59, 1003 (1995).
31. ?‰ ? (Rsample? Rstandard/Rstandard) ? 103. For car-
bon, R ?
?13C(VPDB) of isopropanol: ?29.3‰ (EA-IRMS). ?13C
mass balance equations: ?13Cdiacid? [?13Cder. diacid?
a(?29.3)]/b (a and b are fractional carbon abundances
of isopropanol and diacid, respectively). A proce-
dural succinic acid standard gave no substantial
32. We thank M. Zolensky, A. Hildebrand, and J. Brook for
providing the Tagish Lake sample; J. Cronin for much
help and suggestions; C. Moore for supporting a
student assistant; and four anonymous referees for
helpful reviews. We gratefully acknowledge conver-
sations with M. Bernstein, C. Chyba, J. Dworkin, I.
Gilmour, D. Hudgins, W. Huebner, and A. Primak. The
work was supported by NASA grants from the Exo-
biology (S.P., Y.H., and G.C.) and Cosmochemistry
(L.B.) Divisions and NSF grant CHE9808678 (NMR).
13C/12C and the standard is VPDB.
17 May 2001; accepted 14 August 2001
Published online 23 August 2001;
Include this information when citing this paper.
Origin of Whales from Early
Artiodactyls: Hands and Feet of
Eocene Protocetidae from
Philip D. Gingerich,1* Munir ul Haq,1,2Iyad S. Zalmout,1
Intizar Hussain Khan,2,3M. Sadiq Malkani2
Partial skeletons of two new fossil whales, Artiocetus clavis and Rodhocetus
balochistanensis, are among the oldest known protocetid archaeocetes. These
came from early Lutetian age (47 million years ago) strata in eastern Balo-
chistan Province, Pakistan. Both have an astragalus and cuboid in the ankle with
characteristics diagnostic of artiodactyls; R. balochistanensis has virtually com-
plete fore- and hind limbs. The new skeletons are important in augmenting the
diversity of early Protocetidae, clarifying that Cetacea evolved from early
Artiodactyla rather than Mesonychia and showing how early protocetids swam.
Whales are marine mammals grouped in the
order Cetacea (1). Most mammals live on land
and the fossil record of early mammals is ter-
restrial. Thus, it has long been reasonable to
infer that the origin of whales involved an evo-
lutionary transition from land to sea. This is one
of the most profound changes of adaptive zone
amenable to study in the fossil record. Living
whales are so distinctive, and intermediate fos-
sils sufficiently rare, that a half-century ago
Simpson regarded Cetacea as “on the whole the
most peculiar and aberrant of mammals,” insert-
ing them arbitrarily in his classic 1945 classifi-
Gemeroy (3) applied innovative immunological
precipitin tests and found much higher cross-
reactivity of Cetacea to Artiodactyla than to
other living orders of mammals. Van Valen (4)
attempted to reconcile close relationship of
whales and artiodactyls with the then-known
fossil record by proposing that whales originat-
ed from Paleocene mesonychid condylarths,
1Department of Geological Sciences and Museum of
Paleontology, The University of Michigan, Ann Arbor,
MI 48109–1079, USA.2Geological Survey of Pakistan,
Sariab Road, Quetta, Pakistan.3Department of Earth
Sciences, University of New Hampshire, Durham, NH
*To whom correspondence should be addressed. E-
R E P O R T S
www.sciencemag.orgSCIENCEVOL 29321 SEPTEMBER 2001
on November 20, 2007
whereas artiodactyls originated from closely
related Paleocene arctocyonid condylarths.
Most morphologists and paleontologists have
favored a mesonychid origin of whales (5–
14), but further immunological, DNA hybrid-
ization, and molecular sequencing studies
support close relationship of Cetacea to artio-
dactyls (15–18) and more specifically to hip-
popotami within Artiodactyla (19–27). Here
we provide paleontological evidence showing
that whales evolved from early artiodactyls
rather than mesonychid condylarths.
Artiodactyla (Greek artios, entire or even-
numbered, and dactylos, finger or toe) are
named for the even number (two or four) of
manual and pedal digits (fingers and toes)
found on each hand (manus) and foot (pes) in
extant taxa. Ankle or tarsal bones are the
most diagnostic elements of the artiodactyl
skeleton (28, 29). In contrast, mesonychids
have comparatively generalized mammalian
tarsals (29). Most cetaceans lack hind limbs,
but recovery of reduced tarsal bones in mid-
dle-to-late Eocene archaeocetes Dorudon and
Basilosaurus from Egypt (30) raised hope
that earlier archaeocetes might retain tarsals
that could be compared with artiodactyls and
Primitive archaeocetes are best known from
shallow marine sedimentary rocks of early and
middle Eocene age deposited in the eastern
Tethys Sea in what is now India and Pakistan.
In the year 2000, we found two partially artic-
ulated skeletons of early Protocetidae preserv-
ing fore- and hind limbs. Protocetidae are gen-
erally considered to be ancestral to later Basi-
losauridae and, hence, on the main line of ceta-
cean evolution (1). One of the new skeletons
represents a new genus and species, and the
second represents a new species within an ex-
isting genus. Both come from transitional beds
of the Domanda Formation in Lakha Kach syn-
cline near Rakhni, in the eastern part of Balo-
chistan Province, Pakistan. Habib Rahi lime-
stones represent the early Lutetian high sea
stand in the Indus Basin, and the age of the
transitional zone yielding the whales described
here is about 47 million years ago (Ma) (31,
32). For comparison on the same time scale, the
ages of pakicetids Himalayacetus (12) and Pa-
kicetus (6) are about 53.5 and 48 Ma, respec-
tively; the age of ambulocetid Ambulocetus (7)
is between 48 and 47 Ma; and the age of
basilosaurids Basilosaurus and Dorudon in
Egypt (30) is about 37 Ma.
Artiocetus clavis, new genus and species
(33), includes a virtually complete skull (Fig. 1)
that increases the range of cranial forms and
feeding specializations known in early Proto-
cetidae. Artiocetus is distinctive in that its ex-
ternal nares are positioned well forward on the
dorsal surface like those of land mammals, it
has a narrower frontal shield than that of other
protocetids (relative to both the length of the
cranium and breadth of the cranial base), and it
has a wide cranial base relative to its skull
length (29). Artiocetus is slightly older geolog-
ically (circa 0.5 million years) than other early
protocetids (8, 34), and its proportions are like-
ly to be primitive for the family.
The skeleton of Artiocetus was located
after pieces of ankle bones, including a com-
plete left astragalus, partial left calcaneum,
and complete left cuboid, were found on the
surface. Other bone pieces were traced up-
slope about 2 m to a point where a pelvis and
lumbar vertebrae were lying in situ on a
bedding plane. Excavation yielded the artic-
ulated thorax, cervical vertebrae, and, lastly,
the skull (29). All are parts of the same
specimen: no land mammals are known from
this stratigraphic interval; no other mamma-
lian specimens were found in the vicinity; all
are similar in size, color, and preservation;
and no parts are duplicated. The astragalus
and cuboid (Fig. 2A) have characteristics that
are clearly diagnostic of Artiodactyla. On the
astragalus, the head has a well-developed
navicular trochlea, the sustentacular facet is a
proximodistally elongate oval on the ventral
surface of the astragalus, and the ectal facet is
a relatively small oval concavity on the lat-
eral surface. The cuboid has the distinctive
artiodactyl combination of a concave astraga-
lar facet paired with a characteristically
notched, convex, calcaneal facet.
Rodhocetus balochistanensis, new species
(35), is referred to Rodhocetus because the new
specimen has a femur with a relatively short,
mediolaterally expanded shaft like that of the
type species R. kasrani (8) (femoral shafts
known for other early protocetids, including
Artiocetus, are more cylindrical). The skull of
smaller but probably similar in proportions to
that of R. kasrani (29). Axial skeletons of the
two species can be compared, and that of R.
balochistanensis is about 13% smaller in linear
dimensions. Fore- and hind limbs are virtually
complete and complement what was known of
Rodhocetus previously. The wrist is that of a
generalized mammal, broad and gently arched,
with no carpal fusion and no centrale (Fig. 2B).
Carpals are alternating rather than serial. The
manus is mesaxonic, with the central digit (III)
being longest and most robust. Proximal pha-
langes of digits II through IV were constrained
to be habitually slightly extended by flat artic-
ular surfaces and large sesamoids on metacar-
pals II through IV. These are the digits with
relatively short, broad, hoof-bearing distal pha-
langes (preserved in II and IV) (29). On land,
Rodhocetus walked on a digitigrade hand, with
the central digits II through IV bearing weight.
Digit I is shorter than the others and more
slender, whereas digit V is virtually the same
length as II through IV and also slender. The
lateral digits were not weight-bearing, but ap-
pear to have been retained to broaden a webbed
Ankle bones of Rodhocetus (Fig. 2C) are
long and narrow, with a deep tibial trochlea on
Fig. 1. Cranium of the holotype
of Artiocetus clavis, GSP-UM
3458, new genus and species (A
through C: dorsal, right lateral,
and palatal views, respectively).
Note anterior position of exter-
nal nares, relatively narrow fron-
tal shield, and broad cranial base
of skull (29). Illustration: Bonnie
R E P O R T S
21 SEPTEMBER 2001 VOL 293SCIENCEwww.sciencemag.org
on November 20, 2007
the astragalus constraining lateral mobility, and
a substantial calcaneal tuber providing leverage
for powerful foot extension. The astragalus and
cuboid have the artiodactyl characteristics men-
tioned above, and, in addition, the calcaneum
of artiodactyls. The pes is paraxonic, with un-
usually long and thin metatarsals and phalanges
(including terminal phalanges). Matching in-
terosseous surfaces bordering the proximal
halves of the metatarsal shafts show that these
could be tightly compressed. Articulations both
with the tarsus and with the phalanges indicate
that the metatarsals were sometimes well sepa-
rated. Contraction of intrinsic muscles nar-
rowed the foot when it was tightly flexed. Un-
usual flanges of bone on the proximomedial
base of middle phalanges II and III and on the
proximolateral base of middle phalanges IV
and V (arrows in Fig. 2C) provided leverage for
opening the feet to maximum breadth during
extension. Pedal phalanges cannot have been
weight-bearing, but were elongated to stiffen a
large webbed foot. On land Rodhocetus evi-
dently walked on the plantar surface of the foot,
with the calcaneum, plantar processes of other
tarsals, and metatarsal sesamoids bearing
weight, somewhat like eared seals do today.
The structure of the hand is consistent with
limited locomotion on land, but the foot shows
that Rodhocetus was predominantly aquatic
rather than terrestrial.
A skeletal restoration of Rodhocetus is
shown in Fig. 3. Metapodials and phalanges of
the hands and feet are similar to those described
for Ambulocetus natans (7, 36), but the hands of
Rodhocetus are longer (about 165% of radius
length in Rodhocetus, compared with 145% of
radius length in Ambulocetus) and the feet are
even longer (about 279% of radius length and
158% of femur length in Rodhocetus, compared
with 197% of radius length and 121% of femur
length in Ambulocetus). Thewissen and Fish
(37) interpreted Ambulocetus as an otter-like
pelvic paddler, and this is a good model for
Rodhocetus. If the hands and feet were webbed
as inferred here, then Rodhocetus was probably
capable of quadrupedal paddling as well as pel-
vic paddling. The robust tail of Rodhocetus sug-
gests that caudal undulation was important as
well, especially while moving beneath the water
surface. (Note that the length of the tail is not
known.) The forelimbs and hands could not be
extended as broad pectoral flippers, which
would be required to control recoil from undu-
lation or oscillation of a caudal fluke (38);
hence, it is doubtful that Rodhocetus had such a
The previously known astragalus and
cuboid of basilosaurid archaeocetes were too
reduced and specialized to permit comparison
with generalized land mammals (30). Inferenc-
es that astragali of Pakicetidae and Ambuloceti-
tioned because the bones involved are fragmen-
material (13). However, astragali and cuboids
of Artiocetus and Rodhocetus described here
are complete and in both cases were found with
substantial protocetid skeletons (29). These in-
dicate that cetaceans evolved from early artio-
dactyls, which corroborates results of many
immunological, DNA hybridization, and mo-
lecular sequencing studies, and resolves a long-
standing disagreement between paleontologists
and molecular systematists. Although there is a
general resemblance of the teeth of archaeo-
cetes to those of mesonychids, such resem-
blance is sometimes overstated and evidently
represents evolutionary convergence (41).
Close relationship of archaeocetes to ar-
tiodactyls casts new light on the morphology
of primitive artiodactyls (42, 43). Rodhocetus
has a five-fingered mesaxonic hand, and the
entocuneiform bears a facet for a first meta-
Fig. 2. Right astragalus and
clavis, new genus and species,
with virtually complete left
manus (B) and left pes (C) of
new species, GSP-UM 3485
side). All are shown in ante-
oblique hatching were not re-
characteristics in the well-de-
veloped navicular trochlea on
the head of the astragalus,
convex fibular facet on the
calcaneum, and concave as-
tragalar facet paired with a
notched into the cuboid. The
hand is mesaxonic with three
central weight-bearing toes
that evidently bore nail-like
hooves (distal phalanges pre-
served on digits II and IV)
flanked by more gracile later-
al toes that lacked hooves
[distal phalanx preserved on
digit I; see (29)]. The foot is
toes, flanges of bone on the
medial or lateral bases of the middle phalanges (arrows) providing leverage for opening the feet to
maximum breadth during extension, and narrowly pointed ungules (distal phalanges preserved on
digits II and III). Abbreviations: Ast., astragalus; ast. f., astragalar facet; Cal., calcaneum; cal. f.,
calcaneal facet; Cub., cuboid; Cun., cuneiform; fib. f., fibular facet; Lun., lunar; Mc, metacarpal; Mt,
metatarsal; Nav., navicular; nav. tr., navicular trochlea; Pis., pisiform; Sca., scaphoid; tib. tr., tibial
trochlea; Tra., trapezium; Unc., unciform. Trapezoid and magnum are present in carpus but not
separately labeled. Illustration: Bonnie Miljour.
Fig. 3. Composite restoration of the skeleton of a paddling Rodhocetus kasrani (8) to show the
morphology of this primitive protocetid. Limbs are scaled up 16% from R. balochistanensis, with the
scapula restored from A. clavis and the proximal humerus restored from slightly later protocetids.
Terminal vertebrae of the tail (hatched) are conjectural, and the tail was almost certainly longer
than shown here. Forelimbs were probably folded against the body during rapid swimming by pelvic
paddling at the sea surface and during rapid swimming by pelvic paddling and caudal undulation
when submerged. On land, Rodhocetus supported itself on hoofed digits II, III, and IV of the hands
and on the plantar surfaces of the feet, and probably progressed somewhat like a modern eared seal
or sea lion. Illustration: Doug Boyer.
R E P O R T S
www.sciencemag.org SCIENCEVOL 29321 SEPTEMBER 2001
on November 20, 2007
tarsal (Mt-I) retained from a formerly five- Download full-text
toed foot. These primitive characteristics are
found in the oldest known artiodactyls like
early Eocene Diacodexis (42, 44), but also in
later anthracotheriid artiodactyls such as late
Eocene-Oligocene Bothriodon from Europe
(“Hyopotamus”) (45), and Elomeryx and pos-
sibly Heptacodon from North America (46,
47). If hippopotamids are derived from an-
thracotheres (48), then it appears plausible
that hippopotami may be the closest living
relatives of whales.
References and Notes
1. R. E. Fordyce, L. G. Barnes, Annu. Rev. Earth Planet.
Sci. 22, 419 (1994).
2. G. G. Simpson, Bull. Am. Mus. Nat. Hist. 85, 1 (1945).
3. A. Boyden, D. Gemeroy, Zoologica 35, 145 (1950).
4. L. M. Van Valen, Bull. Am. Mus. Nat. Hist. 132, 1
5. F. S. Szalay, Evolution 23, 703 (1969).
6. P. D. Gingerich, N. A. Wells, D. E. Russell, S. M. I. Shah,
Science 220, 403 (1983).
7. J. G. M. Thewissen, S. T. Hussain, M. Arif, Science 263,
8. P. D. Gingerich, S. M. Raza, M. Arif, M. Anwar, X. Zhou,
Nature 368, 844 (1994).
9. J. G. M. Thewissen, J. Mammal. Evol. 2, 157 (1994).
10. M. C. McKenna, S. K. Bell, Classification of Mammals
Above the Species Level (Columbia Univ. Press, New
York, 1997), pp. 366–368.
11. W. P. Luckett, N. Hong, J. Mammal. Evol. 5, 127
12. S. Bajpai, P. D. Gingerich, Proc. Natl. Acad. Sci. U.S.A.
95, 15464 (1998).
13. M. A. O’Leary, J. H. Geisler, Syst. Biol. 49, 455 (1999).
14. Z. Luo, P. D. Gingerich, Univ. Mich. Pap. Paleontol. 31,
15. D. Graur, D. G. Higgins, Mol. Biol. Evol. 11, 357
16. C. Montgelard, F. M. Catzeflis, E. Douzery, Mol. Biol.
Evol. 14, 550 (1997).
17. F. G. R. Liu, M. M. Miyamoto, Syst. Biol. 48, 54 (1999).
18. J. K. Lum et al., Mol. Biol. Evol. 17, 1417 (2000).
19. V. M. Sarich, in Mammal Phylogeny: Placentals, F. S.
Szalay, M. J. Novacek, M. C. McKenna, Eds. (Springer,
New York, 1993), pp. 103–114.
20. D. M. Irwin, U. Arnason, J. Mammal. Evol. 2, 37 (1994).
21. J. E. Gatesy, C. Hayashi, M. A. Cronin, P. Arctander,
Mol. Biol. Evol. 13, 954 (1996).
22. M. Shimamura et al., Nature 388, 666 (1997).
23. B. Ursing, U. Arnason, Proc. R. Soc. London Ser. B 265,
24. M. Nikaido, A. P. Rooney, N. Okada, Proc. Natl. Acad.
Sci. U.S.A. 96, 10261 (1999).
25. J. E. Gatesy, M. C. Milinkovitch, V. G. Waddell, M. J.
Stanhope, Syst. Biol. 48, 6 (1999).
26. R. G. Kleineidam, G. Pesole, H. J. Breukelman, J. J. Bein-
tema, R. A. Kastelein, J. Mol. Evol. 48, 360 (1999).
27. U. Arnason, A. Gullberg, S. Gretarsdottir, B. Ursing, A.
Janke, J. Mol. Evol. 50, 569 (2000).
28. B. Schaeffer, Am. Mus. Novit. 1356, 1 (1947).
29. Web figs. 1 through 7, Web table 1, and supplemental
text are available at Science Online at www.
30. P. D. Gingerich, B. H. Smith, E. L. Simons, Science 249,
31. W. A. Berggren, D. V. Kent, C. C. Swisher, M.-P. Aubry,
in Geochronology, Time Scales and Global Strati-
graphic Correlations: A Unified Temporal Framework
for an Historical Geology, W. A. Berggren, D. V. Kent,
M.-P. Aubry, J. A. Hardenbol, Eds. (Society of Econom-
ic Paleontologists and Mineralogists, Tulsa, 1995),
special vol. 54, pp. 129–212.
32. J. A. Hardenbol et al., in Mesozoic and Cenozoic
Sequence Stratigraphy of European Basins, P.-C. d.
Graciansky, J. A. Hardenbol, T. Jacquin, P. R. Vail, Eds.
(SEPM Society for Sedimentary Geology, Special Pub-
lication 60, 1998), pp. 3–13 (eight charts).
33. Order Cetacea, Suborder Archaeoceti, Family Protoceti-
dae. Artiocetus clavis, new genus and species. Etymol-
ogy: artios, entire or even-numbered, and ketos, Gr.,
whale; clavis, L., key or clavicle; reflecting possession of
shared characteristics of Artiodactyla and Cetacea, and
alluding to both the key intermediacy of this taxon and
retention of a rudimentary clavicle in the shoulder
girdle. Holotype: GSP-UM 3458 (Geological Survey of
tually complete skull with much of the axial skeleton;
parts of the shoulder girdle and forelimb including a
rudimentary clavicle, scapula, distal radius and ulna; and
parts of pelvic girdle and hind limb including an ilium,
distal femur with a patella, and complete astragalus and
cuboid. Eruption of all permanent teeth and fusion of
most epiphyses shows that the specimen was fully
mature. Type locality: Kunvit, Kohlu District, eastern
Balochistan Province, Pakistan (30°05?44?N latitude,
cetid archaeocete (estimated weight, 420 kg) with a
skull distinctive in having anteriorly positioned nares, a
relatively narrow frontal shield, and a relatively broad
cranial base (29). Femoral shaft is roughly cylindrical.
Astragalus differs from that in contemporary Rodhoce-
tus balochistanensis in being smaller and relatively low-
er, with smaller ectal and fibular facets. Description:
See supplementary material (29).
34. P. D. Gingerich, M. Arif, W. C. Clyde, Contrib. Mus.
Paleont. Univ. Mich. 29, 291 (1995).
35. Order Cetacea, Suborder Archaeoceti, Family Protoceti-
dae. Genus Rodhocetus Gingerich et al. (8). Rodhocetus
chistanensis, referring to provenence of type specimen.
Holotype: GSP-UM 3485: braincase of skull with much
of axial skeleton; parts of forelimb including distal hu-
merus, radius and ulna, virtually complete carpus and
acetabular rim of pelvis, femur, patellae, tibia, virtually
complete tarsus and pes. Fusion of most epiphyses
shows that the specimen was fully mature. Type local-
ity: Kunvit, Kohlu District, eastern Balochistan Province,
Pakistan (30°05?20?N latitude, 69°47?39?E longitude).
450 kg compared with 590 kg in R. kasrani; see (49)].
Anterior thoracic vertebrae average 13% smaller in
linear dimensions (meaning R. kasrani is 16% larger),
whereas the femur is slightly longer. Femoral shaft has
a more distinct third trochanter but is otherwise simi-
larly diamond-shaped in cross-section. Astragalus dif-
larger and relatively higher, with larger ectal and fibular
facets. Description: See supplementary material (29).
36. J. G. M. Thewissen, S. I. Madar, S. T. Hussain, Cour.
Forschungsinst. Senckenb. Frankfurt 191, 1 (1996).
37. J. G. M. Thewissen, F. E. Fish, Paleobiology 23, 482
38. P. W. Webb, Am. Zool. 28, 709 (1988).
39. J. G. M. Thewissen, S. I. Madar, S. T. Hussain, Nature
395, 452 (1998).
40. J. G. M. Thewissen, S. I. Madar, Syst. Biol. 48, 21
41. G. J. P. Naylor, D. C. Adams, Syst. Biol. 50, 444 (2001).
42. K. D. Rose, Science 216, 621 (1982).
43. K. D. Rose, Proc. Natl. Acad. Sci. U.S.A. 93, 1705
44. J. G. M. Thewissen, S. T. Hussain, Anat. Histol. Em-
bryol. Zentralbl. Veterinarmed. C 19, 37 (1990).
45. W. Kowalevsky, Philos. Trans. R. Soc. London 163, 19
46. W. B. Scott, J. Acad. Nat. Sci. Philadelphia 9, 461
47. W. B. Scott, Trans. Am. Philos. Soc. 28, 363 (1940).
48. E. H. Colbert, Am. Mus. Novit. 799, 1 (1935).
49. P. D. Gingerich, in The Emergence of Whales: Evolu-
tionary Patterns in the Origin of Cetacea, J. G. M.
Thewissen, Ed. (Plenum, New York, 1998), pp. 423–
50. We thank Hasan Ghaur, A. Latif Khan, S. Ghazanfar
Abbas, and Imran Khan, Geological Survey of Pakistan,
Quetta, for encouragement and logistical support in the
field. Specimens were prepared by W. J. Sanders, J.
Groenke, and D. Erickson at the University of Michigan.
P. Myers provided access to comparative collections of
mammals in the University of Michigan Museum of
Zoology and J. G. M. Thewissen provided casts of pre-
Fordyce, K. D. Rose, W. J. Sanders, B. H. Smith, J. G. M.
Thewissen, and M. D. Uhen read and improved the
manuscript. Field and laboratory research was support-
ed by NGS (5072-93) and NSF (EAR-9714923).
28 June 2001; accepted 15 August 2001
Rapid Diversification of a
Species-Rich Genus of
Neotropical Rain Forest Trees
James E. Richardson,1,2R. Toby Pennington,1*
Terence D. Pennington,3Peter M. Hollingsworth1
Species richness in the tropics has been attributed to the gradual accumulation
of species over a long geological period in stable equatorial climates or, con-
Pleistocene climates. DNA sequence data are consistent with recent diversi-
fication in Inga, a species-rich neotropical tree genus. We estimate that spe-
ciation was concentrated in the past 10 million years, with many species arising
as recently as 2 million years ago. This coincides with the more recent major
uplifts of the Andes, the bridging of the Isthmus of Panama, and Quaternary
glacial cycles. Inga may be representative of other species-rich neotropical
genera with rapid growth and reproduction, which contribute substantially to
species numbers in the world’s most diverse flora.
The neotropical flora comprises about 90,000
plant species—37% of the world’s total and
more than the floras of tropical Africa (35,000
spp.) and Asia (40,000 spp.) combined (1).
Most of these species are found in rain forests,
which have higher plant species diversity than
any other habitat on the planet. How this diver-
sity arose is unexplained (2, 3). Early theories
(the “museum model”) suggested that a stable
tropical climate allowed species to accumulate
over time, with low rates of extinction in the
absence of major environmental perturbations
R E P O R T S
21 SEPTEMBER 2001 VOL 293SCIENCE www.sciencemag.org
on November 20, 2007