would be reduced considerably in the We constrained Eq.

I to avoid negative regener-

ation values by adjusting gross sedimentation.
16. We gratefully acknowledge the critical discus-
sions and manuscript reviews by T. Kratz, A.
absence of ground-water inflow. In addi- We assume that PBS dissolution in the sediment Beckel, T. Frost, and M. Anderson. Research
tion, if a large fraction of the sedimented traps is negligible. support was provided through the NSF Long-
14. D. R. Lee, Int. J. Speleol. 8, 117 (1976); M. S. Term Ecological Research Program, contract
diatom silica were not regenerated, the McBride, thesis, University of Minnesota DEB 8012313.
diatom crop in Crystal Lake would be (1972); A. L. Tolman, thesis, University of * Present address: Department of Geological Sci-
Wisconsin (1975). ences, Wright State University, Dayton, Ohio
similarly reduced. 15. D. W. Schindler, R. W. Newbury, K. G. Beaty, 45435.
P. Campbell, J. Fish. Res. Board Can. 33, 2531
Ground water clearly is a major factor (1976). 7 August 1984; accepted 17 December 1984
controlling the chemical composition
and biological dynamics of certain lakes.
The role of ground water will be domi-
nant for some constituents and lakes (14)
but may be minor in others (15). Thus, Geologic Youth of Gal.ipagos Islands Confirmed by Marine
we concur with Winter (3) in recognizing Stratigraphy and Paleontology
the need for understanding ground wa-
ter-lake relations when constructing lake Abstract. Six distinctive types of fossiliferous marine deposits occur on the
nutrient and water budgets. Galdpagos Islands that provide evidence for the age of emergence of the islands
JAMES P. HURLEY above sea level and hence a maximum age for the islands' terrestrial biota. These
DAVID E. ARMSTRONG subtidal to supratidal deposits include (i) volcanic tuffs with fossils, (ii) limestones
Water Chemistry Program, and sandstones interbedded with basalt, (iii) terrace deposits, (iv) beach rock, (v)
University of Wisconsin, supratidal talus deposits, and (vi) recently uplifted tidal and subtidal rocks and sand.
Madison 53706 With the exception of (vi), the deposits were previously assigned ages varying from
GALEN J. KENOYER* Miocene to Pleistocene, but all are less than about 2 million years old. This age,
CARL J. BOWSER together with independently determined geologic ages, indicate that the islands
Department of Geology and emergedfrom the sea relatively recently and that all evolution of the islands' unique
Geophysics, University of Wisconsin terrestrial biota occurred within the past 3 to 4 million years.

References and Notes The Galapagos Islands are famous for cene and Pliocene were later described
1. R. G. Wetzel, Limnology (Saunders, Philadel-
their unique terrestrial biota. Species from various sedimentary deposits on
phia, 1975). there evolved in isolation and radiated several islands (7-9). The oldest age in-
2. G. E. Hutchinson, A Treatise on Limnology into a variety of ecological niches. For ferred for the GalApagos is Miocene and
(Wiley, New York, 1957).
3. T. C. Winter, Verh. Int. Verein. Theoret. An- example, 13 species of Darwin's finches is based on fossils and stratigraphy of
gew. Limnol. 20,'438 (1978).
4. J. W. G. Lund, Biol. Rev. (Cambridge) 40, 231 radiated from an apparent single ances- limestone deposits on northeastern Isla
(1961). tral species (1). No fossil evidence has Santa Cruz (10). Radiometric dates (3,
5. P. Kilham, Limnol. Oceanogr. 16, 10 (1971).
6. Samples were collected at five to ten depths at been uncovered that reveals the initial 11-16) and the islands' plate tectonic
the deep-hole station; see Fig. 1 for the collec- history or age of this biota, hence the history (2), however, indicate that the
tion frequency.
7. We determined DRS on 0.4-pm (Nuclepore) earliest time of arrival of any ancestral islands did not begin to emerge from the
filtrates, following the method of J. D. H. Strick- species and the maximum time available sea until the Pliocene, some 3 to 4 million
land and T. R. Parsons [A Practical Handbook
ofSeawaterAnalysis (Fisheries Research Board for radiation on the islands can be estab- years ago.
of Canada, Ottawa, 1972), pp. 65-70]. PBS was lished only from geologic and marine On eight islands we found six types of
analyzed by the method of D. W. Eggiman, F.
T. Manheim, and P. R. Betzer [J. Sediment.
Petrol. 50, 215 (1980)]. TS was calculated as
paleontologic evidence. To date, geolo- fossiliferous marine deposits: (i) subma-
DRS + PBS. Chlorophyll a was measured ac- gists have suggested that the islands are rine tuff cones, (ii) limestone and sand-
cording to the method of C. J. Lorenzen [Lim- very young, perhaps no older than 3 + stone interbedded with pillow basalts
nol. Oceanogr. 12, 343 (1967)].
8. Data from 57 piezometers around and beneath million years (2, 3), whereas paleontolo- and basalt flows, (iii) subtidal rock and
the lake were input into flow nets for Darcy's gists have suggested a Miocene age (10 sand deposits preserved on terraces, (iv)
law -calculations to determine the monthly
ground-water flux to and from the lake. to 14 million years ago) for some fossil beach rock, (v) supratidal talus debris,
9. Duplicate traps (3:1 ratio of depth to diameter) occurrences and Pliocene to Pleistocene and (vi) recently uplifted subtidal to su-
were deployed at 17 m; see W. D. Gardner [J.
Mar. Res. 38, 41 (1980)] for general design. The ages for others. Thus the age of the initial pratidal rocks and sand. These deposits
relative error between duplicate traps was about
20 percent. Lower accuracy is expected during appearance of the islands above sea level bear significantly on the geologic history
the well-mixed periods (early May to mid-June
and late October to mid-December).
and the time at which they first were of the islands, and types (ii) through (iv)
10. The permanent accumulation rate was based on available for colonization by a terrestrial have been used in previous interpreta-
direct analysis of PBS in depth-sectioned sedi- biota have not been resolved. The re- tions of the islands' age (Fig. 2).
ment cores and a mass sedimentation rate mea-
sured by R. Talbot [thesis, University of Wis- ports of pre-Pleistocene ages based on The fossiliferous tuffs occur as broad
consin-Madison (1981)] from excess '°Pb pro- marine fossils and sedimentary rocks are cones formed during shallow submarine
files. The range reflects the uncertainty in the
area of the depositional zone and the lateral cited in biological arguments concerning eruptions. Six major tuff cones and cone
changes in sediment PBS concentrations.
11. Rainfall was measured at a National Atmospher- the time of colonization and rates of complexes ring Isla Santa Cruz (15) and
ic Deposition Program site located 3 km from
Crystal Lake. The mean DRS content was taken
evolutionary divergence in the fauna (4). are thought to represent an early phase
from unpublished data of S. J. Eisenreich, P. J. We describe here six distinctive types of of Galapagos volcanism (16, 17). The
Emmling, and A. M. Beeton for bulk precipita- fossiliferous marine deposits that con- cones are now eroded and are connected
tion collected in the area.
12. A mixed-lake mass of 100 kg of silica corre- firm the younger ages proposed by geol- to the volcanoes of the main island by
sponds to a concentration of about 25 ,ug liter- '. ogists (Figs. 1 and 2). subaerial basalt flows. We studied the
For comparison, typical concentrations of DRS
in northern temperate lakes are 0.46 to 1.18 mg
liter1 (2).
Marine fossils have long been known tuff cones at Cerro Gallina and Cerro
13. The accuracy of the calculated regeneration flux from the Galapagos Islands. Darwin (5) Colorado (Fig. 1) (18). In the poorly
depends on errors in the other silica fluxes (Eq.
1), estimated to be ahout 30 percent for gross
and Wolf (6) reported marine fossils in sorted and poorly bedded facies near the
sedimentation, 50 percent for ground-water in- volcanic rocks high above sea level, but vents, boulder-sized fragments of shal-
flow, and 20 percent for ATS. Thus, errors in these reports were not investigated sub- low-water coquinas and fossiliferous
regeneration range from about 25 to 30 percent
for high fluxes to >100 percent for low values. sequently. Fossils reported to be Pleisto- limestone are incorporated in the tuff. In
1578 SCIENCE, VOL. 227
the better sorted distal facies, individual ported from terraces at James Bay, Isla extinguished for.at least 1 million years
mollusk shells are relatively common. San Salvador (8). The oldest rocks ex- (3). Thus the terrace deposits in the
The oldest radiometric age obtained to posed on San Salvador were dated at less Galapagos vary widely in age but most
date in the Galapagos Islands is a potas- than 0.77 ± 0.12 million years (13). On are Pleistocene or younger. However,
sium-argon date of 4.8 ± 1.87 million Santa Fe fossiliferous terrace deposits at the islands are tectonically and volcani-
years on the Cerro Colorado tuff at near- different elevations contain a diverse fau- cally active; terrace deposits probably
by South Plazas Island (3, 11). The in- na. The oldest radiometric date for Santa have formed throughout their history.
cluded limestone clasts there are litho- Fe is 2.85 ± 0.06 million years (3). One ter- Fossils from beach rock have also
logically and faunally identical to fossilif- race deposit there underlies a lava flow, been reported as being Pleistocene (8).
erous limestones interbedded with re- and volcanism on this island has been Beach rock, dipping seaward and ce-
verse-polarity basalts, radiometrically
dated at near 2 million years (16), that
crop out within a few hundred meters of
the tuff cone. At Cerro Gallina we col- 1. Pinta
lected 20 species of mollusks occurring Galapagos Islands
as individual shells in the tuff from sea
level to the top of the eroded cone, and 3
1. Marchena
1~~~~~.
Genovesa
we collected a similar but less diverse
1 Tuff cones
fauna from the Cerro Colorado cone
(18). The specimens are inferred to have -o + ± 2 Limestones,
~~~~~~~~~~~~~~sandstones
been alive or recently dead at the time of 1-3 1. 3 Terrace deposits
eruption because they are infilled with San Salvador 4 Beach rock
tuff. All the species are alive today in 4~.
-~~~~~ ~~5 Talus deposit
shallow waters of the eastern Pacific 2 6 Recent uplift
Ocean. The large standard deviation in
the tuff date, the possible lithologic cor- 1. Rabida2
1. Fernandina 1 ,2
relation at Cerro Colorado, and the mod- 1. Santa Cruz
em aspect of the tuff-cone faunas suggest <
I. Isabela
. ~~1. Pinz6n
.Pnzn3 SnaF San Crist6bal
3
that the cones are younger than nearby
submarine basalts and limestones. Santa Fe
Fossiliferous marine limestone and 3
tuffaceous sandstone beds are exposed -1 +
in sea cliffs on Isla Baltra and just north 4
of Cerro Colorado on Isla Santa Cruz.
The species are from a variety of inter- Santa Maria 1.
tidal and shallow subtidal associations. &Z Espanola
The proportion of extinct mollusk spe-
cies from Isla Baltra led Hertlein (9) to Fig. 1. Map of the Galapagos Islands showing fossil collection localities and types of deposits.
assign a Pliocene age to these deposits. It Cerro Colorado tuff cone is in the northeastern part of Isla Santa Cruz and offshore South Plazas
has also been suggested that the rocks Island and Cerro Gallina cone is in the southwestern part.
near Cerro Colorado were later Miocene
(10). Radiometric age dates and paleo- Previous age Revised age
magnetism of the associated volcanic assignments assignments
rocks, however, confirm that the inter- Urvina Bay uplift
bedded marine deposits are approxi- Recent Urvina Bay uplift Recent W Talus deposits
mately 2 million years old (3, 11, 15, 16). z Beach rock
Marine terrace deposits on the Galapa- --10,000 years--
m James Bay terraces
gos occur mostly between 5 and 10 m 0.72 mya Villamil terrace
above sea level and range from pockets Talus deposits
of fossiliferous sand on lava platforms Pleistocene
Beach rock
James Bay terraces Pleistocene 4
Santa Fe terraces
and between basalt boulders to stratified Villamil terrace 4
sand up to 3 m thick that may be cement- : Tuff cones???
ed at surface exposures. The fossils in -1.6 mya--
these deposits are generally well pre- Interbedded limestones
served and abundant; species diversity is on Baltra and at
commonly high. On Islas Isabela, San Pliocene Limestones on
2.47 mya - _ Cerro Colorado
~~Baltra
Salvador, and Santa Fd these deposits Pliocene
have molluscan faunas of more than 50 4
species, and nearly all of these are living Limestones and tuff
today. A terrace deposit just north of Miocene cone at
Villamil, Isla Isabela, reported previous- Cerro Colorado ---5 mya--
ly as being Pleistocene (7, 8), is sur- Miocene
rounded by a younger lava flow. The Fig. 2. Previous age assignments for fossil deposits and revised age estimates based on this
radiometric age for Isla Isabela is 0.09 study together with radiometric age dates and paleomagnetic stratigraphy obtained on rocks
± 0.04 million years (3, 11), and the island from the Galapagos. The Pliocene-Pleistocene boundary is now placed at 1.6 million years ago
is still active. Pleistocene fossils were re- (mya) by others (19).
29 MARCH 1985 1579
mented in various degrees by CaCO3, at least 3 million years ago (14). The 8. L. 0. Hertlein and A. M. Strong, Proc. Calif.
crops out at low-tide to mid-high-tide marine paleontological record is thus 9. Acad. Sci. 23, 367 (1939).
L. G. Hertlein, ibid. 39, 25 (1972).
levels on many beaches in the archipela- reconciled with the geologic evidence, 10. J. W. Durham,R.Pac. Discovery 18, 3 (1965);
and A. McBirney, in The Encyclope-
go. The faunas from these rocks include and together they indicate that all adap- dia of World Regional Geology, part 1, Western
the same species found on adjacent mod- tive radiation in the terrestrial biota of Hemisphere, R. W. Fairbridge, Ed. (Dowden,
Hutchinson & Ross, Stroudsburg, Pa., 1975),
ern beaches. Because beach rock can the GalApagos Islands occurred within pp. 285-290; J. W. Durham, Veliger 21, 369
form very rapidly in tropical and sub- the past 3 to 4 million years. (1979).
11. A. Cox and B. Dalrymple, Nature (London) 209,
tropical regions and because these rocks CAROLE S. HICKMAN 776 (1966).
in the Galapagos are associated with Department ofPaleontology, University 12. K. Bailey, Science 192, 465 (1976).
13. F. J. Swanson, H. W. Baitis, J. Lexa, J. Dy-
modem sea level stands, contain modem of California, Berkeley 94720 mond, Bull. Geol. Soc. Am. 85, 1803 (1974).
14. M. L. Hall, Science 221, 545 (1983).
faunas, and dip seaward, we infer that JERE H. LiPPs 15. C. S. Bow, thesis, University of Oregon, Eu-
these deposits are younger than Pleisto- Department of Geology, gene (1979).
16. A. R. McBirney and H. Williams, Geol. Soc.
cene. University of California, Davis 95616 Am. Mem. No. 197(1969).
On Isla Rabida, Hertlein and Strong 17. H. Williams, in The Galdpagos, R. I. Bowman,
Referenoes and Notes Ed. (Univ. of California Press, Berkeley, 1966),
(8) reported marine mammal bone and pp. 65-70.
two species of mollusks from scoria- 1. P.Steadman,R. Grant, Am. Sci. 69, 653 (1981); D. W. 18. W. D. Pitt et al., Proc. Calif. Acad. Sci., in
Trans. San Diego Soc. Nat. Hist. 19, press.
ceous talus at the base of steep cliffs 279 (1982). 19. L. Tauxe, N. D. Opdyke, G. Pasini, C. Elmi,
Hey, Bull. Geol. Soc. Am. 88, 1404 (1977). Nature (London) 304, 125 (1983); B. Backman,
above high-tide level. They concluded 2.3. R. A. Cox, in Patterns of Evolution in Galdpagos N. J. Shackleton, T. Tauxe, ibid., p. 156.
from the species present that the age was Organisms, R. I. Bowman and A. E. Leviton, 20. This report is based on results obtained on an
Eds. (American Association for the Advance- expedition organized b W. D. Pitt; other partic-
Pleistocene, although they listed the two ment of Science, Washington, D.C., 1983), pp. ipants were L. Pitt andM. J. James. W. D. Pitt
mollusk species as ranging into the Re- 4. J.11-23. and M. J. James identified the fossil mollusks
S. Wyles and V. M. Sarich, ibid., pp. 177-186. from Cerro Gallina and Cerro Colorado. R. Hey
cent., Our examination of the shore of 5. C. R. Darwin, Geological Observations on Vol- read a draft of- the report. We thank them, F.
Isla Rabida revealed that storm-tossed 6. T. canic Islands (Smith, Elder, London, 1846). Koster, A. DeRoi, J. DeRoi, and Ecuadorian
Wolf, Ges. Erd-Kunde Berlin Verh. 22, 246 military officials for assistance and courtesies.
shells and bones of sea lions and birds (1895). This is contribution 367 from the Charles Dar-
win Research Foundation.
W. H. Dall, Geol. Mag. 61, 428 (1924);
are now being rapidly buried supratidally 7. and W. H. Ochsner, Proc. Calif. Acad. Sci. 17,
by debris falling from the cliffs. We con- 89 (1928). 3 December 1984; accepted 15 January 1985
clude that these deposits at outcrop are
also very young, perhaps no more than a
few hundred years old.
In 1954 at Urvina Bay, Isla Isabela,
the shore was suddenly uplifted 4.5 m Blood-Brain Barrier: Endogenous Modulation by
shortly before the eruption of nearby Adrenal-Cortical Function
Volcan Alcedo (16). The uplift exposed
supratidal to subtidal bay and rocky Abstract. The blood-brain barrier restricts the passage of molecules from the
shore communities. Now only well-skel- blood to the brain. The permeability of the barrier to iodine-125-labeled bovine
etonized forms remain, such as calcare- serum albumin was examined in rats that had undergone adrenalectomy, adrenal
ous algae, mollusks, echinoderms, and demedullation, and corticosterone replacement. Adrenalectomy, but not adrenal
barnacles. Uplifted biotas may thus be demedullation, increased the permeability of brain tissue to the isotopically labeled
quite recent and do not necessarily imply macromolecule; corticosterone replacement reversed this effect. These results
Pleistocene or older ages, as commonly indicate that the blood-brain barrier may be hormonally regulated; that is, the
assumed in the past. The relative impor- pituitary-adrenal axis may physiologically modulate the permeability of the brain
tance and scale of localized uplift in microvasculature to macromolecules.
forming the present islands have been
debated by geologists (16, 17). These The blood-brain barrier (BBB) re- more, dexamethasone reduces the dis-
deposits and terrace deposits confirm the stricts the passage of polar compounds ruption of the BBB produced by drug-
importance of localized uplift in shaping and macromolecules from the blood into induced acute hypertension (9), repeated
the islands, but they do not indicate the brain interstitium (1, 2). This barrier convulsive seizure activity (10), or hy-
similarity of ages. arises from several morphologic charac- pertonic perfusion of the brain (11).
Shallow-water, tidal, and supratidal teristics of the brain vasculature; name- These observations prompted our specu-
marine fossil assemblages in the Galapa- ly, the tight junctions between adjacent lation that the pituitary-adrenal axis may
gos Islands are more abundant and di- endothelial cells and the virtual absence take part in the endogenous regulation of
verse than previously recognized. They of fenestrations and cytoplasmic pinocy- BBB integrity and that adrenal glucocor-
occur in at least six geologic settings. In totic vesicles within endothelial cells (1- ticoids may specifically influence the
contrast to early paleontologic studies, 3). Anatomic and physiologic evidence permeability characteristics of the undis-
our observations and conclusions (Fig. indicates that cerebral microvascular rupted brain microvasculature. There-
2) corroborate independent geologic evi- permeability may be responsive to both fore, we investigated the effects of adre-
dence (2, 3) that the western Galapagos neural and humoral influences (4, 5) and nalectomy, selective adrenal demedulla-
Islands emerged from the sea less than 2 has led to speculation on the roles of tion, and corticosterone replacement on
or 3 million years ago. The easternmost these systems in maintaining homeosta- the permeability of the BBB to '25I-
islands may be about 1 million years sis within the central nervous system labeled bovine serum albumin (BSA;
older on the basis of plate tectonic age through actions on the BBB (5, 6). 69,000 molecular weight) in conscious,
estimates (2, 3), although these estimates Dexamethasone and other synthetic freely moving rats. We now report that
do not suggest when the islands ap- glucocorticoids have been widely used total adrenalectomy, but not selective
peared above sea level. Radiometric for the clinical treatment of brain edema adrenal demedullation, significantly in-
dates on subaerial lava flows indicate (7) and have been shown to alter the flux creases the permeability of 125I-labeled
that Isla Espafiola stood above sea level of water across the BBB (8). Further- BSA into the brain and that corticoster-
1580 SCIENCE, VOL. 227