OligoceneMiocene age of aridity in the Atacama Desert revealed by exposure dating of erosion-sensitive landforms

Tibor J. Dunai Faculty of Earth and Life Sciences, Vrije Universiteit, De Boelelaan 1085, 1081 HV Amsterdam,
The Netherlands

Gabriel A. Gonzalez Lopez Departamento de Ciencias Geologicas, Facultad de Ingeniera y Ciencias Geologicas,
Universidad Catolica del Norte, Avenida Angamos 0610, Antofagasta, Chile

Joaquim Juez-Larre Faculty of Earth and Life Sciences, Vrije Universiteit, De Boelelaan 1085, 1081 HV Amsterdam,
The Netherlands ABSTRACT The age of onset of hyperaridity in the Atacama Desert, Chile, which is needed to validate geological and climatological concepts, has been heretofore uncertain. Measurement of cosmogenic 21Ne in clasts from erosion-sensitive sediment surfaces in northern Chile shows that these surfaces have been barely affected by erosion since 25 Ma. Surface exposure ages of sediment clasts give replicate values at 25, 20, and 14 Ma and individual values at 37 and 9 Ma. Predominantly hyperarid conditions are required to preserve these oldest continuously exposed surfaces on Earth. Our ndings are compatible with the hypothesis that the onset of aridity in the Atacama Desert could be the reason for, rather than the consequence of, uplift of the high Andes. Keywords: Atacama Desert, Andes, desertication, climate change, erosion, exposure age. INTRODUCTION The Atacama Desert is one of the major hyperarid deserts on Earth. It represents an extreme habitat for life on Earth and serves as an analogue for dry conditions on Mars (McKay et al., 2003). Aridity in the Atacama Desert is primarily caused by the cold water of the Humboldt Current running parallel to the Chilean and southern Peruvian coast, preventing precipitation in the coastal areas (Houston and Hartley, 2003). The aridity is intensied by the pronounced rain-shadow effect of the Andes to the east, which effectively block moisture transfer from the Amazon Basin (Houston and Hartley, 2003). The onset of aridity in the Atacama Desert and changes in its intensity were governed by the onset and uctuations in strength of the proto-Humboldt Currents and the timing and rate of uplift of the Andes (Lamb and Davis, 2003). In turn, the arid climate in the Atacama Desert inuences the rates and patterns of uplift and denudation of the Andes (Lamb and Davis, 2003). It has been suggested that the arid conditions of the Atacama Desert are the cause rather than the result of the uplift of the high Andes (Lamb and Davis, 2003). The driving force would be the climate-controlled sediment starvation in the Peru-Chile trench, causing high shear stress, focusing the plate boundary stresses that support the high Andes (Lamb and Davis, 2003). In order to test the possibility of this causal link, reliable information on the timing of aridication of the Atacama Desert is required. AGE OF ARIDITY IN THE ATACAMA DESERT Our present knowledge of the timing of desertication of the Atacama Desert mostly relies on two lines of evidence. One is the timing of cessation of supergene alteration of orebodies in the Precordillera (Alpers and Brimhall, 1988; Sillitoe and McKee, 1996) (Fig. 1), the other is the nature and timing of changes of sediment input into the Central Depression (Hartley and Chong, 2002). The termination of supergene enrichment of orebodies points to an early regional desiccation, starting ca. 35 Ma and completed by ca. 14 Ma, whereas the sedimentological evidence points to a relatively recent change from semiarid to hyperarid conditions ca. 3 Ma. Both lines of evidence are derived from investigation of areas mostly more than 100 km inland. The rare precipitation in the Atacama, however, comes from the east and reaches the upper regions ( 3500 m) of the western ank of the Precordillera (Houston and Hartley, 2003). Even in the present-day hyperarid conditions in the central Atacama, discharge from the Precordillera into the Central Depression occurs, both occasionally as sheetows on vast alluvial fans and continuously by rivers that have their headwaters deep in the Precordillera (e.g., Ro Loa). No measurable precip itation is observed in the elevated regions of the Coastal Cordillera. Therefore it is possible that previously used records might overestimate past precipitation and underestimate the age of aridity in the Atacama Desert in general and specically in the Coastal Cordillera, the coastal desert proper. Here we assess the age of cessation of erosion, as a consequence of aridication, on erosion-sensitive landforms in the Coastal Cordillera. The large-scale morphostructural units in the study area (Figs. 1 and 2) formed when the coastal region of northernmost Chile and southernmost Peru was near sea level in late Oligoceneearly Miocene time (Noble et al., 1985; Tosdal et al., 1984). In this period, regional erosion formed the rounded-planate summits of the Coastal Cordillera with sediments grading from sand to cobble conglomerate lling the valleys in a coastal plain set-

Figure 1. Geographic setting of study area in Atacama Desert (digital elevation model based on GTOPO30). Black circles with numbers give locations and K-Ar ages of supergene alteration products of orebodies (Alpers and Brimhall, 1988; Bouzari and Clark, 2002; Sillitoe and McKee, 1996). Large dark gray circles indicate sedimentary deposits used for paleoclimatic studies (Hartley and Chong, 2002). Dashed lines give calculated mean annual precipitation based on topographic elevation (Houston and Hartley, 2003).

2005 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or editing@geosociety.org. Geology; April 2005; v. 33; no. 4; p. 321324; doi: 10.1130/G21184.1; 4 gures; Data Repository item 2005053.

321

Figure 2. Enlarged portion of Landsat ETM image (panchromatic band, P002R073 7P20000329) showing sampling sites and geographic features of study area. Major fault scarps and gravitational collapse structures are indicated. WSWENE-trending reverse fault is expression of trench-parallel shortening of Coastal Cordillera (Allmendinger et al., 2005). Ephemeral rivers have deeply incised into Coastal Cordillera. Valleys and fault scarps protect areas around sampling sites AC from runoff from precipitation in Precordillera to east. Quebrada is canyon or valley; ocina indicates (nitrate) plant.

ting (Tosdal et al., 1984). The sources of these sediments are in the Precordillera and/or the Coastal Cordillera. In northern Chile these sediments belong to the Azapa Formation (Worner et al., 2002); their equivalents in Peru belong to the Moquegua Formation (Tosdal et al., 1984; Worner et al., 2002). Regionally, sedimentation ended at the latest ca. 18 Ma (Tosdal et al., 1984), the bulk occurring between 22 and 25 Ma (Mortimer et al., 1974; Tosdal et al., 1984; Worner et al., 2002). Sed imentation in our study area (Figs. 1 and 2) ended shortly after 21.8 0.3 Ma (Mortimer et al., 1974). For the present study we chose depositional surfaces on Azapa sediments that have been effectively protected from runoff from the Precordillera since their deposition (Figs. 2 and 3). Consequently these surfaces have been exclusively affected by local precipitation since deposition of the Azapa Formation. The traces of uvial transport and erosion on these surfaces therefore record pluvial episodes in the coastal desert since 25 Ma. SAMPLES Here we report exposure ages of quartz clasts collected from a sediment surface just inland of Pisagua (Fig. 2). Photographs and descriptions of sampling sites are provided in GSA Data Repository1. The surface is at elevations between 910 m and 1000 m (Fig. 3) and is dissected by the Quebrada de Tiliviche
1GSA Data Repository item 2005053, Table DR1 and Figure DR1, cosmogenic isotope data and images of the study area, is available online at www.geosociety.org/pubs/ft2005.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301 9140, USA.

and the Quebrada de Jazpampa (quebrada means canyon or valley). The deeply incised valleys provide a local low base level for erosion (Figs. 2 and 3). The gradient on the sediment surface (Fig. 3) would be conducive to uvial sediment transport of sand and cobbles of the Azapa Formation, if running water were available. We collected samples at three sites that are indicative of erosion processes. Site A, located between the two quebradas, is protected from hillslope runoff by a 1020-m-deep depression parallel to and at the foot of the NW-SEtrending fault scarp in Azapa sediments (Fig. 2). In the current setting, site A is only affected by runoff from precipitation below the

1000 m isoline (Fig. 3). Due to the slightly convex form of the area around site A (Fig. 3), it is well protected from runoff, erosion, and/or deposition of material from higher areas. Sites B and C are located at the axis of a wide topographic low. Hence most runoff from the surface to the south of the Quebrada de Jazpampa will ow across these two sites. Therefore, in contrast to site A, sites B and C are very sensitive to erosion and/or deposition by runoff from higher areas (Fig. 2). Site B is at the bottom of one of the rst in a series of steep-walled salt-karst depressions, mostly 2 m deep, that occur between site B and the Quebrada de Jazpampa. These depressions act as a perfect sediment trap for debris carried by runoff from higher areas (Fig. 3). The salt karst is formed in old evaporites that were deposited in a salina-mudat setting, when Azapa sediment surfaces were still at or close to sea level (Tosdal et al., 1984). The quartz clasts collected at this site were lying on salt (mostly halite) in the karst pit. The clasts are probably not residual material from dissolved evaporites, but were carried by runoff from higher up in the catchment onto the evaporites (Fig. 3). The rock clasts found in the karst pits are indistinguishable from clasts found on surfaces adjacent to the karst pits. Site C is on a at, undulating (decimeter scale) surface. All runoff that will eventually end up in the karst pits around site B must cross the area around site C (Fig. 3). Site C can be affected by runoff erosion and is the location of (temporary) deposition of clasts from higher up in the catchment. In addition to the sediment surface described here, we collected quartz clasts in the Quebrada de Jazpampa (site D) and on an alluvial fan (site E) situated on tectonically

Figure 3. Elevation contour plot based on Shuttle Radar Topography Mission (SRTM) 90 m data showing hydrographic situation of sites A, B, and C. Below 900 m only 50 m contours are shown; above 900 m, 10 m contours are provided. White areas in steep section in NW corner indicate missing data. Sites B and C are on axis of wide topographic low draining to Quebrada de Jazpampa. Runoff of all precipitation in this catchment will cross these sites.

322

GEOLOGY, April 2005

Figure 4. Cumulative probability plot of clast exposure ages. Individual ages of clasts are also indicated, as is age of amalgamated sample from site A. Minimum and maximum deposition ages indicated are independent age constraints for onset and cessation of sedimentation of Azapa sediments in study area (Mortimer et al., 1974; Tosdal et al., 1984, 1981; Worner et al., 2002).

rates, since the sediments sampled at 950 m were deposited near sea level (Noble et al., 1985; Tosdal et al., 1984). If the uplift was decelerating or accelerating, and was not constant as we assume for our age calculation, the corresponding exposure ages would decrease or increase, respectively. For example, ages would change by 10% if the uplift rate decreased/increased fourfold after half the exposure time. OLIGOCENEMIOCENE ONSET OF ARIDITY IN THE ATACAMA DESERT The majority of clasts sampled on the geomorphologically old sediment surface have exposure ages that are indistinguishable from the sediment deposition age, 2225 Ma, or are slightly ( 10%) younger. The age of the amalgamated sample from the best-protected surface (site A) is 23 Ma. The concordant exposure and deposition ages (Fig. 4) leave little chance for erosive modication of this sediment surface since their deposition in the early Miocene. The sediment surface sampled is by far the oldest continuously exposed geomorphologic surface on Earth, being about twice as old as ancient surfaces in Antarctica (e.g., Schafer et al., 1999; van der Wateren et al., 1999, and references therein). The events that led to the deposition of the younger clasts at site B and C did not significantly erode this region, otherwise the older clasts present at site C would have been removed. It is likely that the younger clasts come from the higher areas surrounding sites B and C (Figs. 2 and 3). We interpret the ages of the younger clasts as evidence for pluvial episodes ca. 20 Ma, ca. 14 Ma, and ca. 9 Ma. The runoff that drained through the trough axis at sites B and C (Fig. 3) was able to locally dissolve 2 m of evaporites to form the salt karst at site B and farther north. Given the 25 m.y. period available to dissolve the salt, we conclude that only marginal precipitation has occurred since that time. Generally the climate must have resembled the present-day hyperarid climate for most of the past 25 m.y. The proposal of a semiarid climate in the central Atacama Desert until 3 Ma (Hartley and Chong, 2002) is clearly incompatible with our ndings. Our ndings are, however, in agreement with an estimate of 21 Ma for the end of supergene weathering in the Coastal Cordillera, as may be inferred from the only orebody dated in a climatic setting equivalent to that of our study area (Fig. 1; Sillitoe and McKee, 1996), i.e., the coastal desert proper. Based on the large-scale causes for the regional climate, we assume that the roughly coast- and/or orogen-parallel zoning in precipitation (Fig. 1) also existed in the past. Thus, noting the absence of conditions conducive to a special microclimate in our study area, we infer that the long-term prev-

downfaulted sediments of the surface sampled in sites AC (Fig. 2). The Quebrada de Jazpampa is the only overow of the Pampa Tamarugal (Fig. 1), and records spilling events from this large salt pan that serves as a local base level for all discharge from the Precordillera for the next 200 km to the south. The 435 m elevation of site E on the coastal cliff is in the elevation range of 3001000 m, where rain is occasionally observed along the coastal cliff, e.g., in major El Nino events. Consequently sites D and E record the last geomorphologically relevant precipitation in the Precordillera and on the coastal cliff, respectively. EXPOSURE AGES Exposure ages were determined from the concentration of in situproduced cosmogenic 21Ne in quartz clasts. Details on the samples, experimental procedure, age calculation, and the isotope data are provided in Table DR1 (see footnote one). The exposure ages of the clasts from the sediment surface are generally very old. From sites A, B, and C, 4 individual clasts were analyzed per site along with 1 amalgamated sample (Repka et al., 1997) of 24 clasts from site A. The majority of the ages are older than 19 Ma (n 9) with clusters at 20 Ma (n 3) and 25 Ma (n 5) (Fig. 4). One clast yielded an exposure age of ca. 37 Ma. Few clasts are younger: one age is ca. 9 Ma, and two identical ages are ca. 14 Ma. The site best protected from runoff erosion (A) yielded no clasts younger than 19 Ma and contained the oldest clast. The amalgamated sample gives a mean age of 23.3 0.2 Ma ( 1 ) for site A. The samples of the other two sites (B and C) contain the three younger clasts, together with ve clasts of ages indistinguishable from those of site A. The samples from the riverbed of the Quebrada de Jazpampa (D) and the alluvial fan (E) both give ages ca. 120 ka. When dating sedimentary deposits accumulation of cosmogenic nuclides at the source area or during transport has to be considered.

At sites A, B, and C it was impractical to obtain shielded samples to correct for this preexposure (Repka et al., 1997). We use the samples from site D to assess preexposure of quartz clasts in the Azapa Formation; the sediments in the riverbed are recycled Azapa sediments that were cannibalized from the valley anks that are cut into the Azapa Formation. The two amalgamated samples from this site indicate that average clasts from the Azapa Formation have no signicant preexposure. The 21Ne concentration found in these samples is in agreement with a single-stage exposure for 120 k.y. at the surface and at 90 cm depth, respectively. This nding does not exclude the possibility that individual clasts have signicant preexposure, as suggested by the oldest clast age of ca. 37 Ma. However, based on the rather tight clustering of the remaining clastsca. 14, ca. 20, and ca. 25 Mawe judge that it is unlikely that these clasts had a signicant, necessarily random, preexposure exceeding 1 Ma. The source regions of the quartz clasts analyzed are in magmatic rocks and medium to high-grade metamorphic rocks. Any preexposure would have occurred during erosion of these rocks and not during previous sedimentary cycles, as could be the case if sedimentary rocks were the source of the clasts. In order to mimic an exposure age of 1 Ma by preexposure, erosion rates in the higher source areas in the Precordillera (1500 m elevation at 25 Ma; Lamb and Davis, 2003) would have been 1 m/m.y. Denudation rates during the pluvial episode in the emerging Precordillera that led to the deposition of the Azapa Formation were, however, probably much higher (Lamb and Davis, 2003; Tosdal et al., 1984). The sediments forming the depositional surface investigated in this study were deposited at the end of this pluvial episode. By this time most relicts of older landscape surfaces probably had vanished by erosion. We use an average uplift rate of 40 m/m.y. to calculate time-integrated 21Ne production

GEOLOGY, April 2005

323

alence of hyperarid climatic conditions we nd in our study area is probably valid for most of the hyperarid portion of the presentday Atacama Desert. The oldest exposure age of ca. 37 Ma, obtained from a single clast, gives evidence of the existence of remnants of old surfaces in the source region of the Azapa sediments at the time of deposition. This is in line with the fact that regionally the oldest supergene weathering ages of ca. 3435 Ma were obtained in an orebody to the east of our study area (Cerro Colorado; Bouzari and Clark, 2002; Sillitoe and McKee, 1996), providing evidence that very old landforms were present in the source region of the Azapa sediments. The actual exposure age of the sample is probably somewhat younger than calculated for the sampling elevation, as production rates of cosmogenic nuclides increase with altitude, i.e., the production rate during exposure of this sample in the source region was higher. A slowly eroding surface ( 0.1 m/m.y.) that was at 1500 m ca. 25 Ma could be a model source for the clast with the exceptionally high exposure age. Such low erosion rates usually only occur in desert environments (van der Wateren and Dunai, 2001), and indicate that the region had an arid climate prior to the deposition of the Azapa sediments. Potential source areas in the Precordillera were at 1500 m ca. 25 Ma (Lamb and Davis, 2003). OROGRAPHIC RAIN SHADOW VS. GLOBAL CLIMATE CHANGE AS DRIVING FORCE FOR ENVIRONMENTAL CHANGE IN THE ATACAMA DESERT Most ages for the ancient pluvial phases in the Precordillera and Coastal Cordillera, as inferred from the present and previous studies (Alpers and Brimhall, 1988; Sillitoe and McKee, 1996; Tosdal et al., 1984) as ca. 9, ca. 14, ca. 20, and ca. 25 Ma, broadly coincide with periods of global cooling in the middle Miocene and cool climates across the Oligocene-Miocene boundary (Zachos et al., 2001a, 2001b). Termination of a potential earlier pluvial period ca. 34 Ma recorded in the Precordillera (Bouzari and Clark, 2002; Sillitoe and McKee, 1996) coincides with global cooling shortly after the Eocene-Oligocene boundary (Zachos et al., 2001a). These global climate changes were the result of opening or closing of oceanic pathways and orbital forcing (Houston and Hartley, 2003; Zachos et al., 2001a, 2001b), and inuenced the availability and transport of humidity from the Amazon Basin to the Atacama Desert (Houston and Hartley, 2003). This trans-Andean transfer has been the main source of humidity in the Atacama Desert

since the establishment of a proto-Humboldt Current in conjunction with the opening of the Tasmania-Antarctic passage ca. 33 Ma (Zachos et al., 2001a). The climatic connection of these two regions is illustrated by ages of inferred wetter periods in the Amazon Basin ca. 1217, ca. 20, ca. 24, and 3335 Ma (Vasconcelos et al., 1994) that correspond to the pluvial phases in the Atacama Desert identied in the present and previous studies (Alpers and Brimhall, 1988; Sillitoe and McKee, 1996; Tosdal et al., 1984). The dominantly hyperarid conditions we infer for the Coastal Cordillera since ca. 25 Ma, and prevailing arid conditions since ca. 34 Ma, are equivalent to the postulated early aridity (Lamb and Davis, 2003) that is required for the hypothesis that the onset of aridity in the Atacama Desert is the cause (Lamb and Davis, 2003), rather than the result of the uplift of the high Andes. The ensuing positive feedback between increasing altitude of the Andes and increasing rain shadow could create and maintain hyperarid conditions in the Atacama Desert. Only exceptional global climatic disturbances have occasionally permitted humidity transfer across the Andes into the driest regions of this coastal desert since ca. 25 Ma.
ACKNOWLEDGMENTS
Reviews by R. Anderson, P. Bierman, and F. Stuart helped to improve this manuscript; Juez-Larre was sup ported by a Netherlands Organization for Scientic Research (NWO) grant to Dunai. We thank R. van Elsas for density separations, and B. van der Wagt for U and Th determinations.

REFERENCES CITED
Allmendinger, R.W., Gonzalez, G., Yu, J., Hoke, G., and Isacks, B., 2005, Trench-parallel shortening in the northern Chilean forearc: Tectonic and climatic implications: Geological Society of America Bulletin (in press). Alpers, C.N., and Brimhall, G.H., 1988, Middle Miocene climatic change in the Atacama Desert, northern Chile: Evidence from supergene mineralization at La Escondia: Geological Society of America Bulletin, v. 100, p. 16401656, doi: 10.1130/0016 7606(1988)1002.3.CO;2. Bouzari, F., and Clark, A.H., 2002, Anatomy, evolution, and metallogenic signicance of the supergene orebody of the Cerro Colorado porphyry copper deposit, I region, northern Chile: Economic Geology, v. 97, p. 17011740. Hartley, A.J., and Chong, G., 2002, Late Pliocene age for the Atacama Desert: Implications for the desertication of western South America: Geology, v. 30, p. 4346, doi: 10.1130/0091-7613(2002) 0302.0.CO;2. Houston, J., and Hartley, A.J., 2003, The Central Andean west-slope rainshadow and its potential contribution to the origin of hyper-aridity in the Atacama Desert: International Journal of Climatology, v. 23, p. 14531464, doi: 10.1002/ joc.938. Lamb, S., and Davis, P., 2003, Cenozoic climate change as a possible cause for the rise of the Andes: Nature, v. 425, p. 792797, doi: 10.1038/ nature02049.

McKay, C.P., Friedmann, E.I., Gomez-Silva, B., Carceres-Villanueva, L., Andersen, D.T., and Landheim, R., 2003, Temperature and moisture conditions for life in the extreme arid region of the Atacama Desert: Four years of observation including the El Nino of 19971998: Astro biology, v. 3, p. 393406, doi: 10.1089/ 153110703769016460. Mortimer, C., Ferrar, T.E., and Saric, N., 1974, K-Ar ages from Tertiary lavas of the northernmost Chilean Andes: Geologische Rundschau, v. 63, p. 484490. Noble, D.C., Sebrier, M., Megard, F., and McKee, E.H., 1985, Demonstration of two pulses of Paleogene deformation in the Andes of Peru: Earth and Planetary Science Letters, v. 73, p. 345349, doi: 10.1016/ 0012-821X(85)90082-2. Repka, J.L., Anderson, R.S., and Finkel, R.C., 1997, Cosmogenic dating of uvial terraces, Fremont River, Utah: Earth and Planetary Science Letters, v. 152, p. 5973, doi: 10.1016/S0012821X(97)00149-0. Schafer, J.M., Ivy-Ochs, S., Wieler, R., Leya, I., Baur, H., Denton, G.H., and Schluchter, C., 1999, Cos mogenic noble gas studies in the oldest landscape on Earth: Surface exposure ages of the Dry Valleys, Antarctica: Earth and Planetary Science Letters, v. 167, p. 215226, doi: 10.1016/S0012821X(99)00029-1. Sillitoe, R.H., and McKee, E.H., 1996, Age of supergene oxidation and enrichment in the Chilean porphyry copper province: Economic Geology, v. 91, p. 164179. Tosdal, R.M., Clark, A.H., and Ferrar, E., 1984, Cenozoic polyphase landscape and tectonic evolution of the Cordillera Occidental, southernmost Peru: Geological Society of America Bulletin, v. 95, p. 13181332. Van der Wateren, F.M., and Dunai, T.J., 2001, Late Neogene passive margin denudation history: Cosmogenic isotope measurements from the Central Namib Desert: Global and Planetary Change, v. 30, p. 271307, doi: 10.1016/S09218181(01)00104-7. Van der Wateren, F.M., Dunai, T.J., Van Balen, R.T., Klas, W., Verbers, A.L.L.M., Passchier, S., and Herpers, U., 1999, Contrasting Neogene denudation histories of different structural regions in the Transantarctic Mountains rift ank constrained by cosmogenic isotope measurements: Global and Planetary Change, v. 23, p. 145172, doi: 10.1016/S0921-8181(99)00055-7. Vasconcelos, P.M., Renne, P.R., Brimhall, G.H., and Becker, T.A., 1994, Direct dating of weathering phenomena by 40Ar/39Ar and K-Ar analysis of supergene K-Mn oxides: Geochimica et Cosmochimica Acta, v. 58, p. 16351665, doi: 10.1016/ 0016-7037(94)90565-7. Worner, G., Uhlig, D., Kohler, I., and Seyfried, H., 2002, Evolution of the West Andean Escarpment at 18 S (N. Chile) during the last 25 Ma: Uplift, erosion and collapse through time: Tectonophysics, v. 345, p. 183198, doi: 10.1016/S00401951(01)00212-8. Zachos, J., Pagani, M., Sloan, L., Thomas, E., and Billups, K., 2001a, Trends, rhythms and aberrations in global climate 65 Ma to present: Science, v. 292, p. 686693, doi: 10.1126/science. 1059412. Zachos, J.C., Shackleton, N.J., Ravenaugh, J.S., Palike, H., and Flower, B.P., 2001b, Climate response to orbital forcing across the Oligocene-Miocene boundary: Science, v. 292, p. 274278, doi: 10.1126/science.1058288. Manuscript received 13 September 2004 Revised manuscript received 17 December 2004 Manuscript accepted 17 December 2004 Printed in USA

324

GEOLOGY, April 2005