EVIDENCE OF LATE MIOCENE ENVIRONMENTAL CHANGE

Stable Isotopes and Paleobotonical Evidence from the Indian Subcontinent

Vertebrate paleontologists working in northern Pakistan with Will Downs seem to have been the first to recognize a significant Mio-Pliocene climate change in that area. Flynn and Jacobs (1982) inferred a change near 7 Ma from forest to grassland from changes in fossil rodents, and Barry et al. (1985, 1995; Barry 1986, 1995) noted the same change from largely browsers before ~7 Ma to grazers afterward. The remarkable fossil record had been the stimulus for applying magnetostratigraphy to continental sediment, and dating of this material in the late 1970s and early 1980s provided an unusually good chronology.

To my knowledge, Quade et al. (1989) presented the first detailed record of apparent climate change in this region (Figure 1). They showed pronounced changes in carbon and oxygen isotopes in pedogenic carbonates from northern Pakistan, just south of the Himalaya. Quade et al. (1989, 1995; Quade and Cerling 1995) showed an abrupt increase of more than 8-12‰ in ¹³C beginning between 7.5 and 8 Ma (with adjustment to Cande and Kent’s (1995) most recent time scale for geomagnetic polarity reversals) and finishing by 6 Ma. Analyses of enamel from fossil teeth showed the same pattern, if with less time resolution (Quade et al. 1992; Stern et al. 1994). Oxygen isotopes, ¹⁸O, from pedogenic carbonates also show a clear change, of 3 ± 1‰, but slightly earlier, between 9.3 and 7 Ma (Quade and Cerling 1995; Quade et al. 1989, 1995), with ages revised by Dettman et al. (2003) who used Cande and Kent’s (1995) time scale. Stern et al. (1997) inferred a similar change in ¹⁸O values from the clay mineral smectite within the paleosol units. Analyses of ¹³C and ¹⁸O values from paleosols in Nepal also showed shifts, if smaller and less well resolved, between 6 and 8 Ma (Harrison et al. 1993; Quade et al. 1995). Moreover, France-Lanord and Derry (1994) found that ¹³C in organic carbon within sediment in the Bay of Bengal also showed this shift at 7 ±1 Ma. The changes in isotopic composition first seen by Quade et al. (1989) seem to characterize much of the northern Indian subcontinent.

Paleobotanical evidence corroborates some kind of environmental change. A change from forest, indicated by fossil leaves, to grassland, shown by fossil pollen, occurred at ~8 to 6.5 Ma in Nepal (Hoorn et al. 2000). Although less precisely dated, Prasad (1993) noted a similar change farther west in the Indian Himalaya. Using both pollen and increasing pedogenic carbonates in the Thakkhola graben of Nepal, Garzione et al. (2003) reported greater aridity since ~11 Ma, and especially near ~7 to 8 Ma.

These various observations both led directly to inferences of climate change and expose their non-unique implications. Quade et al. (1989) suggested that the isotopic shifts toward present-day conditions marked an onset or pronounced strengthening of the Indian monsoon. Later, however, it became clear that the shift in ¹³C occurred elsewhere, if not everywhere at the same time (Cerling et al. 1993, 1997; Wang et al. 1994). Cerling et al. (1993, 1997) argued that the shift in ¹³C implied global ecological change that enabled C4 plants to become dominant in some settings after a long interval of almost entirely C3 plants. As C4 grasses gain an advantage over C3 grasses when climates become seasonably arid, with pronounced dry seasons but not necessarily less rain (e.g., Ehleringer and Monson 1993; Ehleringer et al. 1997), the shift in Pakistan could reflect a change toward more monsoonal conditions. C4 plants also thrive when atmospheric CO₂ becomes low (e.g., Ehleringer and Monson 1993; Ehleringer et al. 1997), but (as discussed below) inferences of global paleo-CO₂ concentrations cast some doubt on this explanation for the emergence of C4 plants. More importantly, the change in ¹⁸O values seems to have occurred earlier than that in ¹³C values, suggesting that whatever caused one might not be responsible for the other.

To examine seasonal variations in rainfall, Dettman et al. (2001) examined ¹⁸O variation across freshwater bivalve shells from Nepal. In monsoon climates, such shells should record pronounced seasonal variations in ¹⁸O values throughout the period of their growth, with wet-season values much more negative than dry-season values. The range of ¹⁸O values, however, shows no obvious changes since 10.7 Ma that might suggest stronger wet and dry seasonality since that period. Thus, Dettman et al. (2001) concluded that insofar as the shells recorded seasonal variations in precipitation associated with the monsoon, the monsoon had changed little since 10.7 Ma. In fact, if their data did indicate a change at ~8 Ma, the change would be toward less seasonality, and hence perhaps a weaker monsoon in a more arid climate. Four of five of the pre-7.5 Ma wet season ¹⁸ O values presented by Dettman et al. (2001) are more negative (-9.5‰) than the six post-7.5 Ma wet season values (-6.5‰) . They might suggest that the wet season was wetter prior to 7.5 Ma than after that time, which is consistent with aridification at ~7-8 Ma, but not with an increased seasonal (monsoonal) rainfall at that time. Thus, stable isotopes from the Indian subcontinent do not provide a firm footing for the inference of a stronger monsoon since 6-8 Ma than before that time.

In summary, most of the observations can be interpreted as a shift toward more arid conditions at 6-8 Ma. Moreover, the various factors that can contribute to a shift in ¹⁸O values, which is one of the more precisely dated and clearer signals, do not allow them to resolve what climate change actually occurred or, more precisely, whether or not the Indian monsoon changed significantly.

Winds over the Arabian Sea

Results from the Ocean Drilling Project’s Leg 116 off the southwest coast of Arabia gave impetus to the inference that the Indian monsoon strengthened near 8 Ma (Figure 1). Seasonally changing winds, from the southwest in summer and from the northeast in winter, define the monsoon, and one manifestation of these winds is an upwelling of cold, nutrient-rich, deep water, especially during summer. The steady wind causes a transfer of the upper waters to the right of the wind direction (Ekman transport), and with the Arabian coast on the northwest side, cold water must upwell to replace the water that has been transported southeastward during summer. In the present-day ocean, one foraminifer, Globigerina bulloides, dominates plankton in the northwest Arabian Sea (Curry et al. 1992; Kroon 1988; Prell and Curry 1981). Although most G. bulloides plankton live at high latitudes, this foraminifer reproduces abundantly during summer monsoons, and then virtually disappears between the summer and winter monsoons. G. bulloides currently comprises roughly half of the planktonic foraminifera in the northwest Arabian Sea, and it has done so since ~8 Ma (An et al. 2001; Kroon et al. 1991; Prell et al. 1992). It evolved before 14 Ma, but in the northwest Arabian Sea, it contributed only a few percent of the planktonic foraminifera until 8 Ma (Figure 1). Thus, the obvious inference, drawn by An et al. (2001), Kroon et al. (1991), and Prell et al. (1992) and exploited by others, is that the monsoon strengthened at ~8 Ma. Prell et al. (1992) supported this inference further using approximately simultaneous, qualitative changes in abundances of radiolaria that also seem to thrive during upwelling.

Sedimentation in the Bay of Bengal

Roughly 40% of the sediment derived from erosion of eastern Asia accumulates in the Bay of Bengal (Métivier et al. 1999). As a result of its exceptional thickness, the sediment volume can be constrained well (e.g., Curray 1994), but studying most of it is impossible. Ocean Drilling Project cores on the distal edge, however, recovered sediment spanning the last 16 Myr. This record revealed two surprises, at least to those of us who infer a strengthening of the monsoon from the increase in G. bulloides at ~8 Ma. At ~7 Ma, both grain sizes and sediment accumulation rates decreased at the distal edge of the Bengal Fan (Figure 1; France-Lanord et al. 1993). Most would expect a stronger monsoon to convert rivers into torrents capable of carrying abundant large grains toward the plains of northern India and onward to the Bengal fan. Dating of this material is less precise than that in the Arabian Sea, and the decreases in both accumulation rates and grain sizes at 7 Ma could have begun at 8 Ma. Similarly, increases in accumulation rates and grain sizes at ~1 Ma might have begun earlier.

Aalto (1999; personal commun. 1999) suggested that the logic of associating increased grain sizes and accumulation rates with stronger monsoons may overlook an important internal shift within the fluvial dispersal system. The change from forests to grasslands at ~7-8 Ma could have altered the geomorphic regime and the locus of storage for fine sediment. Streams through forests are commonly wide, dynamic, and anastomosing, but those through grasslands are confined to deep, narrow, single-threaded channels until their discharge exceeds a sharply defined bank-full level. The anastomosing forested channels are inefficient at transporting gravel but efficient at transporting suspended fine sediment. Conversely, deep grassland channels are efficient at sluicing gravel to the delta, but floodwater conveyed over bank diverges across wide, hydraulically rough grasslands, and sands and silts are trapped. This mechanism is corroborated by many observations that the size of sediment accumulated switched from coarse channel deposits to fine over bank deposits in the Siwalik foreland basins (with a minimal change in total volumetric rates) at the same time substantially more coarse sediment began accumulating in the delta. Because of the large volume of fine sediment now stored in the foreland floodplains, the transition to grassland channels might have decreased the discharge of sands and silts to the Bay of Bengal and distal reaches of the Bengal fan. If Aalto’s hypothesis is correct, the decrease in accumulation rates and grain sizes at ~7 Ma at the distal edge of the fan need not be surprising.

Alternatively, the sediment deposited at the distal edge of the fan might not be representative of what rivers bring to the head of the fan. Currently, only one narrow channel transports all sediment entering the head of the fan across it (e.g., Curray et al. 2003), and the evolution and migration of that channel could bias records of accumulation 2000 km from the mouth of the river that delivered the material. In fact, as J. Quade (personal commun., 2004) pointed out, if coarser material were preferentially deposited in the Ganga Basin between 7 and 1 Ma, we might expect to see a higher deposition rate there during that interval, but Burbank et al. (1993) report the opposite. Thus, the observations of decreased grain sizes and accumulation rates at the distal edge of the Bengal fan suggest some kind of environmental change at ~7 Ma, but what change occurred remains unresolved.

Upwelling in the South China Sea

East Asia undergoes seasonal climate changes that are not simultaneous with the Indian monsoon, but that nevertheless share patterns typical of monsoons. In summer, moisture from the western Pacific and South China Sea enters South China. Depending on the strength of this circulation, moist air penetrates to North China or stops farther south. In winter, winds from the northwest transport cold dry air into central China. Like the Indian Monsoon, winds reverse seasonally in phase with reversals in meridional temperature gradients between Tibet and equatorial regions (e.g., He et al. 1987; Hsu and Liu 2003).

G. bulloides does not thrive in the South China Sea, but Wang et al. (2003a) suggested that the relative abundance of Neogloboquadrina dutertrei can provide a measure of monsoon strength in this region. N. dutertrei lives above the thermocline but below the mixed layer. Hence, the depth of the thermocline affects its productivity; when too deep, light cannot reach the organisms on which N. dutertrei feeds, but when shallow, it thrives. Wang et al. (2003a) showed that the percentage of N. dutertrei increased at ~7.6 Ma, from which they inferred a strengthening of monsoonal winds that caused a shoaling of the thermocline. A greater increase in the percentage of N. dutertrei at 3-4 Ma (Figure 1) presumably reflects yet stronger winds associated with an ice-age world.

Loess Deposition in China and Aeolian Sediment in the Pacific Ocean

In spring, winds over the deserts of Mongolia and western China sweep up dust and spread it over eastern China, if not much farther east. This process has a long history, for loess deposition had begun just west of the Loess Plateau, in Qinan (Figure 1), by 22 Ma (Guo et al. 2002). Moreover, isotopic fingerprinting of aeolian sediment in the North Pacific indicates a constant source, the Gobi Desert region, since at least 12 Ma (Pettke et al. 2000). Loess accumulation increased at ~7-8 Ma; this can be seen in peaks of accumulation rates both at Qinan and in the North Pacific (Rea et al. 1998), and by the onset of loess deposition at other sites: Lingtai, 7.05 Ma (Ding et al. 1999), Xifeng, 7.2 Ma (Sun et al. 1998a, 1998b), and Jiaxian, 8.35 Ma (Qiang et al. 2001). At these three latter sites, the basal loess overlies much older rock of a different type. What this increase in aeolian sediment transport and deposition implies for climate change remains open, but obviously some kind of climate change must have occurred.

Palynological and Isotopoic Evidence of Aridification of the Linxia Basin

Fossil pollen spectra from the Linxia Basin show a number of changes near 8.5 Ma, with perhaps the most important being a decrease in conifer pollen (Figure 1) and a concurrent increase in grass pollen (not shown in Figure 1; Ma et al. 1998). These changes suggest that the region became more arid at this time (e.g., An et al. 2001). Moreover, less precisely dated pollen from the Qaidam Basin, west of the Linxia Basin, also suggest aridification in late Miocene time (Wang et al. 1999).

Dettman et al. (2003) reported a relatively small change in ¹⁸O values, from -12‰ to -9‰ from the Linxia Basin near 12 Ma, which they associate with a shift in atmospheric circulation and a more arid climate. Superimposed on this baseline shift are brief intervals with much less negative ¹⁸O values with the least negative reaching -2‰ to  -3‰ between ~9.6 and 8 Ma. These values, from lacustrine sediment, suggest that lakes were closed and that evaporation was high in this period, consistent with marked aridification. These observations, which show a larger environmental change at ~12 Ma than at 8 Ma do not offer much support for a change at 8 Ma, but they do permit one at that time.

To What Extent do Paleoenvironmental Data Imply Synchronous Environmental Change Near 8MA?

If one treats phenomena occurring within the interval between ~9 and ~6 Ma as simultaneous, then nearly all of the changes discussed above occurred simultaneously. Uncertainties in dating most of the marine records are less than ~1 Myr, except perhaps that for sediment accumulation on the Bengal Fan, for which the uncertainty is surely less than 2 Myr. In particular, magnetostratigraphy of terrestrial material gained much of its credibility from applications to the Siwalik series in Pakistan, and the magnetostratigraphic records from the loess in China are among the most impressive. The least accurately dated change discussed above is that of pollen in the Linxia Basin. Even ignoring it, there seems little doubt that environmental changes occurred near 8 Ma in the region surrounding Tibet.

Ignoring suggested changes at 11 or 12 Ma (e.g., DeCelles et al. 1998; Dettman et al. 2003), a closer look at timing of the others requires that changes within the period between 9 and 6 Ma not be simultaneous. Changes in ¹⁸O occurred ~1-2 Myr before those of ¹³C, and because the same samples were used for analyses of both isotopic systems, this difference seems to be required.

Most agree that the change in ¹³C implies a change from a dominance of plants that use the C3 pathway for photosynthesis to a rise in importance of plants, grasses in nearly all cases, that exploit the C4 photosynthetic pathway. Environmental changes that could facilitate a shift from C3 to C4 plants include (1) a switch to more seasonally concentrated precipitation, perhaps associated with greater aridity, but not necessarily so, and (2) a reduced partial pressure of CO₂ (e.g., Cerling et al. 1993, 1997; Ehleringer and Monson 1993; Ehleringer et al. 1997). Estimates of paleo-pCO₂ do not indicate much change since ~10 Ma (Demicco et al. 2003; Pagani et al. 1999; Pearson and Palmer 2000; Royer et al. 2001; Van der Burgh et al. 1993), and thus ascribing the emergence of C4 plants at 6-7 Ma to changes in pCO₂ requires ignoring this evidence. As noted above, variations of ¹⁸O across freshwater bivalve shells, which measure amplitudes of seasonal differences in precipitation, show no indication of change near 8 Ma in Nepal and do not support the inference of increased seasonality (Dettman et al. 2001). Yet other observations, such as the shift from browsers to grazers (Barry 1986, 1995; Barry et al. 1985, 1995; Flynn and Jacobs 1982) and the shift from forest to grassland macrofossils and pollen (Garzione et al. 2003; Hoorn et al. 2000; Ma et al. 1998; Prasad 1993), do corroborate a change consistent with some combination of greater aridity and (if one ignores the evidence of Dettman et al. (2001) more seasonally concentrated precipitation. Thus, the association of the abrupt increase in C4 plants with such climate changes seems sensible.

The increase in ¹⁸O values (wet season) near 8 Ma almost surely reflects a change in the water precipitated on the northern Indian subcontinent. Distinguishing the extent to which the increase, however, indicates a shift in the source of water, the degree to which ¹⁸O was rained out either where ¹⁸O was deposited or en route via the tendency for ¹⁸O to become more negative with increasing rainfall (the “amount” and the “continental effects” of Dansgaard 1964), or the degree of evaporation of surface water (which is important because measurements exploit the carbonate record) remains open to debate. Accordingly, associating this increase with a strengthening of the monsoon must be considered speculative. In fact, the shift to monsoon rainfall should lead to a depletion of ¹⁸O in atmospheric water vapor that precipitates as far inland as northern Pakistan, not the increase in ¹⁸O that has been observed. The simplest explanation for the increase in ¹⁸O values is a change to a more arid environment, which is puzzling because it preceded the increases in ¹³C values, which suggest the same change. Moreover, the evidence most suggestive of a strengthening of the monsoons, the increases in, especially, G. bulloides in the Arabian Sea and in N. dutertrei in the South China Sea, seem to have occurred before the change in ¹³C, but apparently concurrently with that in ¹⁸O.

These observations suggest a crude simultaneity of environmental change near 6-9 Ma throughout the region surrounding Tibet. Moreover, they can be tentatively associated with aridification at least north and northeast of Tibet, perhaps with more seasonally concentrated precipitation on the Indian subcontinent, and with stronger winds over the Arabian and South China Seas. Yet, the apparent difference in timing between changes within the period 9 to 6 Ma makes deducing cause and effect among possible physical processes speculative, if not premature. It is worth recalling that the features that we associate with the monsoons – seasonal winds of constant direction, seasonally concentrated rains, and very heavy rains – owe their origins to different aspects of the ocean-atmosphere system (e.g., Webster et al. 2002). Perhaps not all developed at the same time.