Silicate weathering is a key process by which CO2 is removed from the atmosphere. It has been proposed that mountain uplift caused an increase in silicate weathering, and led to the long-term Cenozoic cooling trend, although this hypothesis remains controversial. Lithium isotopes are tracers of silicate weathering processes, and may allow this hypothesis to be tested. Recent studies have demonstrated that the Li isotope ratio in seawater increased during the period of Himalayan uplift (starting ca. 45 Ma), but the relationship between uplift and the Li isotope ratio of river waters has not been tested. Here we examine Li isotope ratios in rivers draining catchments with variable uplift rates from South Island, New Zealand. A negative trend between δ7Li and uplift shows that areas of rapid uplift have low δ7Li, whereas flatter floodplain areas have high δ7Li. Combined with U activity ratios, the data suggest that primary silicates are transported to floodplains, where δ7Li and (234U/238U) are driven to high values due to preferential uptake of 6Li by secondary minerals and long fluid-mineral contact times that enrich waters in 234U. In contrast, in mountainous areas, fresh primary mineral surfaces are continuously provided, driving δ7Li and (234U/238U) low. This trend is opposite to that expected if the increase in Cenozoic δ7Li in the oceans is driven directly by mountain uplift. These data suggest that the increase in seawater δ7Li reflects the formation of floodplains and the increased formation of secondary minerals, rather than weathering of mountain belts.
Chemical weathering of silicate rocks is one of two removal processes of carbon from the ocean-atmosphere system (the other being organic carbon burial) and therefore a critical component of long-term climate (Berner, 2003; Kump et al., 2000). Chemical weathering of continental rocks is also one of the main suppliers of material to the oceans, and has significant influence on ocean chemistry. There is an ongoing debate about the factors most significant in controlling chemical weathering rates, with climate—temperature and runoff—(Berner et al., 1983; Gislason et al., 2009; Walker et al., 1981), supply of fresh material (Hilley et al., 2010; Raymo and Ruddiman, 1992; Raymo et al., 1988), or some combination thereof (Li et al., 2014; West et al., 2005) thought to be important in different settings in the modern environment (Jacobson and Blum, 2003). Understanding the controls on weathering is critical to determining the behavior of the long-term carbon cycle. A climate-dominated control would yield a feedback process that could explain how the long-term climate has maintained itself within relatively narrow bands through Earth history, whereas a supply-dominated cycle has been suggested to link Cenozoic cooling to uplift of the Himalayas (Raymo and Ruddiman, 1992).
Marine carbonate strontium isotopes were initially used to examine changes in weathering associated with mountain building, but anomalously radiogenic Himalayan carbonates are thought to dominate the riverine Sr flux (Oliver et al., 2003). Li isotopes are potential alternative tracers of weathering processes, and may be the only tracers available whose behavior is solely dominated by silicate weathering processes. Li isotopes are not fractionated by biological processes or plant growth (Lemarchand et al., 2010) and are not affected by carbonate (low temperature or hydrothermal) weathering (Kisakűrek et al., 2005; Millot et al., 2010; Reyes and Trompetter, 2012). This gives δ7Li a significant advantage over tracers such as Ca isotopes, which tend to be dominated by carbonate dissolution and formation (Moore et al., 2013). The δ7Li of primary silicate rocks defines a narrow range, with an average for continental crust of ∼0‰ ± 2‰ (Burton and Vigier, 2011; Teng et al., 2004), compared to the high variability in rivers draining these rocks (6‰–42‰ (Huh et al., 1998; Kisakűrek et al., 2005; Millot et al., 2010; Pogge von Strandmann et al., 2006, 2010, 2012; Vigier et al., 2009). Fluvial δ7Li is thus effectively independent of primary lithology, and the highly variable δ7Li in rivers is controlled by weathering processes, particularly by the extent of uptake of Li into secondary minerals, which preferentially remove 6Li (Pistiner and Henderson, 2003; Wimpenny et al., 2010). Riverine δ7Li therefore reflects the ratio of primary rock dissolution (driving rivers to low, rock-like δ7Li with high [Li]), relative to secondary mineral formation (driving rivers to high δ7Li and lower [Li]; Pogge von Strandmann et al., 2010). River δ7Li is thus controlled by the fraction of Li dissolved, relative to the fraction incorporated into secondary minerals. The less Li in solution, the more there is in clays, and the higher the δ7Li in solution. This behavior has also been described as weathering congruency: if riverine δ7Li is low (closer to the primary rock value), then less Li is being taken in secondary clays, and weathering is described as congruent (i.e., direct reflection of rock chemistry by water chemistry) (Misra and Froelich, 2012; Pogge von Strandmann et al., 2013). The riverine input to the oceans is combined with the hydrothermal input, and removal by low-temperature clays, to determine the oceanic δ7Li (Misra and Froelich, 2012).
A history of the Li isotope ratio of Cenozoic seawater has been assessed and shows increasing δ7Li values from ca. 40 Ma to the present (Hathorne and James, 2006; Misra and Froelich, 2012). The Misra and Froelich (2012) study interpreted this increase as due to Himalayan uplift, increased denudation, and more incongruent weathering in the mountain belt (i.e., an increasing amount of clay formation). A problem with this interpretation, however, is that rivers draining the high Himalayas have δ7Li values lower than the global average (Kisakűrek et al., 2005), implying that Himalayan weathering is more congruent, so that Himalayan uplift should have driven seawater δ7Li to lower values rather than higher.
In this study we examine Li isotope ratios in rivers from New Zealand terrains for which uplift rates have been determined in order to assess the effect of uplift on riverine Li isotope ratios, and therefore the effect of orogeny on riverine δ7Li.
Rivers were sampled from multiple catchments around South Island, New Zealand, divided between the east and west of the island (Fig. DR1 in the GSA Data Repository1). These samples were collected and analyzed for (234U/238U) (parentheses indicate activity ratio), and were used previously to define the interaction between weathering, erosion, and U activity ratios (Robinson et al., 2004). In general, the individual catchment areas are small, so rivers flow through a narrow range of rainfall and uplift environments. Robinson et al. (2004) developed a hydrologically accurate digital elevation model that allows estimates of the average rainfall and uplift rate for each catchment by using digitized rainfall and uplift maps. The western coast has significantly more rainfall than the east (an average of 8000 mm/yr compared to 1600 mm/yr), and a higher uplift rate (5.8 ± 1.5 mm/yr compared to 1.9 ± 0.8 mm/yr). Overall there is a positive correlation (r2 = 0.61) between uplift and rainfall, and uplift/rainfall ratios are higher in the east. Hydrothermal springs were sampled at Hanmer Springs to assess the effects of hydrothermal processes, although these springs do not drain into any of the studied rivers. Weathering lithologies have a relatively uniform bulk lithology, dominantly comprising Mesozoic graywackes and schists (Rattenbury et al., 2006; Jacobson et al., 2003), with relatively low groundwater contributions (Mongillo and Clelland, 1984). For descriptions of analytical methods, see the Data Repository.
Element concentrations are within the range shown by other studies of South Island rivers (Jacobson et al., 2003). Molar Ca/Na (11.8 ± 5.5) covary with Mg/Na ratios (0.36 ± 0.16), as expected for rivers draining the continental crust. Li concentrations vary between 35 and 540 nmol/L, within the range of rivers draining similar terrains in the Mackenzie Basin, Canada (Millot et al., 2010). Li isotope ratios (δ7Li) vary widely between 7.6‰ and 34.7‰ (Table DR1 in the Data Repository; Fig. 1A). In general, rivers from the west of the island have lower δ7Li than the eastern rivers, and overall there is a negative trend between [Li] and δ7Li. Lithium concentrations show positive covariations with uplift rates, while δ7Li is negatively correlated to uplift (Fig. 1B; r2 = 0.67, significant >99%, Spearman-Rank correlation), as well as less significantly to rainfall (r2 = 0.44).
Lithium Isotopes and Uplift
The observed trend between δ7Li and [Li] (shown as 1/Li in Fig. 1A) is typical for the Li system in rivers (Pogge von Strandmann et al., 2010) and relates to the congruency of weathering: Li isotopes in rivers are controlled by the ratio of primary mineral dissolution to secondary mineral formation (Kisakűrek et al., 2005; Pogge von Strandmann et al., 2006, 2012). The range of global rivers extends to lower concentrations and higher δ7Li than the New Zealand rivers, but the latter follow this trend well (Fig. 1A), suggesting that the results of this study can be extrapolated more generally. The similarity between different rivers suggests that, globally, the fractionation caused by clay formation remains similar, allowing behavior approaching mixing.
The primary significance of these New Zealand data is the negative relationship between uplift rate and δ7Li (Fig. 1B). This correlation implies that higher uplift rates rapidly provide fresh primary material, resulting in more dissolution of primary rock material relative to secondary mineral formation. The weaker relationship between δ7Li and rainfall suggests that the hydrological cycle is a less important control on silicate weathering. These data show that, in areas of steep relief where there is continuous supply of fresh rock by uplift and rapid runoff, secondary mineral formation is relatively inhibited. In contrast, in the flatter eastern catchments of South Island, uplift and runoff are lower, waters become more supersaturated, and secondary minerals precipitate, driving δ7Li to higher values. These data therefore indicate that δ7Li values are linked to orogenic processes, showing that in mountainous terrains with high uplift rates, chemical weathering processes are relatively congruent. This observation is consistent with those of high Himalayan rivers, which have relatively low δ7Li values (Kisakűrek et al., 2005).
Uranium Isotopes and Weathering Regimes
The weathering processes can be further elucidated by comparing (234U/238U) and δ7Li data for these streams, with the two systems providing complementary information. Uranium activity ratios are controlled by the ratio of physical erosion to mineral dissolution (Henderson, 2002; Andersen et al., 2009; Chabaux et al., 2003; Pogge von Strandmann et al., 2006, 2010, 2011; Robinson et al., 2004). Relatively high physical erosion rates increase mineral surface area, promoting α recoil of 234U and the leaching of 234U from recoil-damaged lattice sites, and driving riverine (234U/238U) to high values. In contrast, relatively high dissolution rates will drive river water (234U/238U) toward the secular equilibrium value of 1, the value of the bulk silicate rock. Unlike δ7Li, (234U/238U) is not expected to be affected by formation of secondary minerals. These isotope systems are therefore both driven to low values by dissolution, but to high values by different processes, i.e., physical erosion or residence time for (234U/238U), and secondary mineral formation for δ7Li. Therefore coupled use of Li and U isotopes in weathering studies can yield complementary information on dissolution versus clay formation versus erosion (Pogge von Strandmann et al., 2006, 2010).
Rivers from the west coast have (234U/238U) close to 1, and variable but low δ7Li. This implies high dissolution rates and variable and low secondary mineral formation (Fig. 2). The implication is that the high uplift rates in these mountainous catchments result in rapid dissolution of the host rocks, but, due to swift removal of material, rivers rarely reach oversaturation with regard to secondary minerals, which therefore do not form. In contrast, the eastern rivers have higher δ7Li and (234U/238U), indicating both an increase in secondary mineral formation and an increase in grain surface area, from physically eroded material transported to the floodplain. Thus eroded grains settle on the flat topography, where the waters dissolving them become oversaturated, decreasing the dissolution rate and leading to precipitation of secondary minerals, thus increasing the δ7Li.
Consequences for Silicate Weathering Reconstructions
The conclusion that higher uplift drives more congruent weathering, and hence riverine δ7Li to lower values, is consistent with the observation that high Himalayan river δ7Li is almost ubiquitously lower than the global mean (Kisakűrek et al., 2005). The weathering of mountain belts should therefore have driven seawater δ7Li lower during the Cenozoic, rather than toward the higher values observed (Misra and Froelich, 2012). In contrast, it is the tectonically stable areas such as floodplains, associated with and supplied by high-relief tectonically active terrains, that exhibit incongruent weathering and high clay formation (West et al., 2002; Jacobson and Blum, 2003; Moore et al., 2013), and drive riverine δ7Li high. The few river δ7Li data from the Himalayan floodplain show higher δ7Li relative to rivers of the high Himalayas (Huh et al., 1998), supporting this conclusion. Therefore, while Misra and Froelich (2012) are likely correct that orogeny is responsible for the increase in seawater δ7Li (Wanner et al., 2014), the locus of the dominant Li isotope fractionation is the floodplain and foreland surrounding the mountains, rather than the mountains, possibly coupled to a shift in the oceanic Li sink (Li and West, 2014). This conclusion agrees with modeling of the δ7Li record, which suggests a significant increase in retention of Li by clays during the Cenozoic (Li and West, 2014; Wanner et al., 2014). It is possible that this increase in clay retention is linked to Himalayan and Tibetan Plateau and/or Andean uplift (e.g., Hoorn et al., 2010).
For a given denudation rate, congruent weathering provides more cations to the ocean than incongruent weathering (where a proportion of cations is retained in clay minerals). Cenozoic uplift increased the surface area available for weathering and might be expected to lead to an increase in the dissolution of silicates and drawdown of CO2. The formation of significant floodplains associated with this mountain building would, however, have led to retention of a higher fraction of the released cations on the continents, thus limiting the effectiveness of uplift in driving CO2 removal. By recording the extent of this cation retention, and thereby the overall congruency of weathering, lithium isotopes are a record of the efficiency of continental weathering in driving CO2 removal, rather than the overall amount of CO2 removal.
The use of Li isotopes to assess past cation retention is predicated on the assumption that Li is retained in clays in a manner similar to that of cations such as Ca or Mg. Such behavior is seen in the well-studied example of Iceland, with the mobility of Li, Ca, and Mg being almost identical, indicating similar clay retention (Gislason et al., 1996; Pogge von Strandmann et al., 2006; Hindshaw et al., 2013). Iceland data also show that Li and Ca isotopes correlate with one another in rivers where Ca is being removed into clays (Hindshaw et al., 2013), further supporting the use of Li isotopes to assess the retention of cations during clay formation. On a global scale, if weathering of uplifted areas led to Cenozoic cooling, then the increase in primary dissolution caused by mountain building must have outweighed the greater retention of cations on the continents recorded by changes in seawater Li isotopes.
Rivers from South Island, New Zealand, show a strong negative correlation between the uplift rate of their catchments and their Li isotope ratio. This implies that when uplift rates are high, fresh primary material is continuously supplied for dissolution, leading to highly congruent weathering with a relative absence of secondary mineral formation, and δ7Li that reflects the original rock. In contrast, when uplift rates are lower, in tectonically stable areas, formation of secondary minerals preferentially enriches rivers in 7Li. The correlation we observe between δ7Li and uplift rates is in the opposite sense to that required if uplift of the Himalayas directly caused the increase in seawater δ7Li during the Cenozoic. The results instead indicate that it may have been the formation of large floodplains associated with uplift that explains the increase in seawater δ7Li. By recording the congruency of weathering, lithium isotopes may provide a record not of the overall rate of dissolution on the continents, but of the efficiency of this dissolution in driving uptake of CO2.
Analyses and Pogge von Strandmann were funded by Natural Environment Research Council fellowship NE/I020571/1. We thank Andrew Jacobson and two anonymous reviewers for their comments.