We present Os and Sr isotope ratios and Os, Sr and major/trace element concentrations for river waters, spring waters and rains on the North Island of New Zealand. The Os and Sr data are used to examine whether the NINZ is a significant contributor of unradiogenic Os and Sr to the oceans. Major element chemistry is used to quantify weathering and CO2 consumption rates on the island to investigate relationships between these processes and Os and Sr behavior. Chemical erosion rates and CO2 consumption rates across the island range from 44 to 555 km−2 yr−1 and 95 to 1900 × 103 mol CO2 km−2 yr−1, respectively. Strontium flux for the island range from 177 to 16,100 mol km−2 yr−1 and the rivers have an average flux normalized 87Sr/86Sr ratio of 0.7075. In agreement with the previous studies these findings provide further evidence that weathering of arc terrains contributes a disproportionally large amount of Sr to the oceans and consumes very large amounts of CO2 annually compared to their areal extent. However, the 87Sr/86Sr from the NINZ is not particularly unradiogenic and it is likely not contributing significant amounts of unradiogenic Sr to the oceans. Repeated Os analyses and bottle leaching experiments revealed extensive and variable sample contamination by Os leaching from rigorously precleaned LDPE bottles. An upper bound on the flux of Os from NINZ can nevertheless be assessed and indicates that island arcs cannot provide significant amounts of unradiogenic Os to the oceans.
 The overall objective of this study is to investigate the extent to which chemical weathering on the North Island of New Zealand (NINZ) impacts the Os and Sr isotope composition of the oceans. Specifically, we examine the contribution of mantle-derived (unradiogenic) Os and Sr from the NINZ to the oceans. We also assess CO2 consumption rates among volcanic and sedimentary basins on the NINZ and their relationship to the release of Os and Sr to the oceans.
 A common feature of the marine Os and Sr isotope records over the past 65 Ma is that they display two-component mixing between radiogenic (high 87Sr/86Sr ∼ 0.710 and 187Os/188Os ∼ 1.3) and unradiogenic (low 87Sr/86Sr ∼ 0.7030 and 187Os/188Os ∼ 0.13) end-members and seem to covary at a gross level [Pegram et al., 1992; Peucker-Ehrenbrink et al., 1995; Sharma et al., 1999]. Whereas it is evident that the eroding upper continental crust is the radiogenic end member for both elements [Palmer and Edmond, 1989; Sharma and Wasserburg, 1997; Levasseur et al., 1999], the key issue is whether they share a common unradiogenic end-member. There is, at present, considerable uncertainty about sources of unradiogenic Sr and Os to the oceans [Palmer and Edmond, 1989; Pegram et al., 1992; Peucker-Ehrenbrink, 1996; Sharma et al., 1997; Sharma et al., 2000; Davis et al., 2003; Chen et al., 2006; Sharma et al., 2007]. Seafloor alteration has been suggested as a potential source of unradiogenic Os and Sr [Sharma et al., 2000; Butterfield et al., 2001; Davis et al., 2003; Sharma et al., 2007], and cosmic dust has been investigated as a source of unradiogenic Os [Peucker-Ehrenbrink, 1996; Chen et al., 2005; Sharma et al., 2007]; however, neither has been unequivocally shown to be a significant source. While it is possible that Sr and Os have two different unradiogenic end-members and that their marine isotope records are only partially coupled through continental inputs [Sharma et al., 1999], it is also possible that there is a common unradiogenic end-member for these elements as suggested by their seawater isotope records.
 The rise in 87Sr/86Sr and 187Os/188Os through the Cenozoic has been attributed to an increase in physical and chemical denudation on the continents [e.g., Edmond, 1992; Pegram et al., 1992; Raymo and Ruddiman, 1992; Peucker-Ehrenbrink et al., 1995], consistent with the idea that global cooling over the last 50 Ma has resulted from a decline in atmospheric CO2 through increased silicate weathering on the continents [e.g., Berner et al., 1983; Zachos et al., 2001]. Rates of marine sedimentation during the past 5 Ma are reported to be several times higher than the 35 Ma prior [Hay et al., 1988; Zhang et al., 2001]. A dramatic decrease in atmospheric CO2 would be expected to accompany this increased erosion. However, proxies of atmospheric CO2 do not reflect this. Willenbring and von Blanckenburg  argue that the observed increase in sedimentation during the past 5 Ma is an artifact of observation and measurement biases. The authors use 10Be/9Be to show that chemical weathering fluxes and physical erosion rates have been essentially constant over the past 10 Ma. This would suggest that the synchronous increase in 87Sr/86Sr and 187Os/188Os through at least the last 10 Ma is not due to increased continental weathering and therefore may not be a useful proxy of past silicate weathering rates.
 This raises questions of whether the increase in the marine Os-isotope and Sr-isotope ratios is due the weathering of material with higher isotope ratio and if weathering of this material is in any way tied to atmospheric CO2 consumption. If, for example, the major source of Sr is weathering of 87Sr-rich calcite-rich deposits or carbonate sequences [e.g., Jacobson and Blum, 2003; Jacobson et al., 2003; Chamberlain et al., 2005], then this isotope system would not be indicative of variations in gross rates of silicate weathering. Similarly, if the dominate source of Os to rivers is the weathering of sedimentary pyrite as suggested by Li et al. , then this system would also be inadequate. If Sr and Os are to be used as proxies of past silicate weathering there must be further investigation into the sources of these elements. This study provides insights into these sources by examining the Sr and Os concentrations and isotopic character of rivers draining many different lithologies across an active arc system.
 Arc terrains have some of the highest rates of chemical weathering and CO2 consumption in the world [Louvat and Allegre, 1998; Dessert et al., 2003; Lyons et al., 2005; Rad et al., 2007; Goldsmith et al., 2008; Gaillardet et al., 2011; Schopka et al., 2011] and it has been postulated that subsurface chemical weathering on island arcs is a source of unradiogenic Sr [Allegre et al., 2010]. Are island arcs are also a significant source of unradiogenic Os? We have investigated this question in the NINZ by analyzing: (a) the rivers and associated hydrothermal springs from active/dormant island arc volcanoes and (b) hydrothermal springs associated with the subduction and dewatering of the Pacific Plate. We have evaluated the role that specific lithologies and spring systems have on the flux of major elements, and Os and Sr isotope compositions of the rivers draining the NINZ and how this relates to the overall rates of CO2 consumption.
2. Study Area
2.1. Tectonic Elements and Petrologic Associations
 An excellent review of the geologic history of New Zealand, including the major lithologies and associated major and minor rock forming mineral of the island is given by Suggate . Briefly, The NINZ straddles an active convergent plate boundary along which the Pacific Plate is subducting obliquely under the Australian plate (Figure 1). The subduction began at ∼23 Ma ago and has created an island arc system consisting of the Hikurangi Trough, the Hikurangi Accretionary Prism (HAP), and the Taupo Volcanic Zone (TVZ). The North Island Shear Belt (NISB) separates the fore arc sediments on the east from the volcanics on the west. It consists of Paleo-Mesozoic metagraywacke that has been cut by zones of dextral strike-slip faulting caused by oblique subduction. The NISB acts as a frontal ridge against which the fore-arc accretionary prism is being actively uplifted and compressed. The HAP consists of sandstone-shale sequences that are intruded at place by a network of methane seep carbonates [Nyman et al., 2010].
 To the west of the NISB are the TVZ and the Taranaki volcanics. The arc volcanics of the TVZ intrude upon a back-arc basin which is estimated to be extending at an average rate of 8–10 mm/yr. The Wanganui and Taranaki Basins (Figure 1) contain several kilometers of mainly Mesozoic marine sediments. While the Wanganui basin lacks arc volcanism, the Taranaki basin is intruded by the Taranaki volcanics.
 The Auckland/Western Waikato and Northland regions are composed broadly of sedimentary deposits with interspersed volcanic units. The sedimentary sequences are variable both in age and type ranging from Jurassic to Quaternary and from sandstones to carbonates. Two parallel belts of late Oliogocene-Early Miocene volcanic deposits occupy the Northlands peninsula. The eastern belt volcanics are andesites and dacites, and the western belt volcanics are typically basalts and basaltic andesites. Both of these belts have arc signatures and provide evidence that the subduction zone has migrated south-southwest over the past 23 Ma to its present location [e.g., Booden et al., 2011].
 Thermal and nonthermal spring systems are present throughout much of the NINZ. Heat fluxes are higher in spring systems west of the NISB [Reyes et al., 2010]. They are generally correlated with extension and recent volcanism and have likely resulted from convective heat transport from shallow magmatic sources. Areas of tectonic compression such as the HAP and the NISB generally experience lower heat fluxes. The chemistries of spring waters vary widely and likely reflect the variable nature of their origin [Giggenbach, 1988; Giggenbach et al., 1993; Reyes et al., 2010]. Reyes et al.  provide a thorough overview of spring water chemistry on the NINZ. Briefly, springs present in the HAP discharge high ionic strength solutions relative to all other spring systems on the island. Spring waters in the accretionary prism are thought to be derived from pore waters associated with accreted and subducted sediments and from dehydration of marine clays [Fehn and Snyder, 2003; Reyes et al., 2010]. The NISB spring systems also discharge fore-arc subducted fluids mixed with meteoric and formation waters. Similarities in the δ18O and δD compositions of spring waters in the HAP and the Northlands have been suggested as evidence for a similar origin, though the fore-arc subducted waters in the Northlands are much older, as subduction was active in this region much earlier [Reyes et al., 2010]. Spring waters in Auckland/Western Waikato and the Taranaki and Wanganui sedimentary basins are heated formation waters that are not subduction related. The thermal waters discharging in the TVZ/Coromandel region are derived from a mixture of heated meteoric waters and subduction related fluids. Cosmogenic 129I ages of springs discharging in the TVZ are younger than those in the fore arc region and have been used to argue for significant inputs from marine sediment pore dewatering [Fehn et al., 2007].
2.3. Climate and Hydrology
 The climate of the North Island ranges from subtropical in the north of the island to more temperate in the southern part of the island. Mean annual temperatures for the NINZ range between approximately 8 and 18°C. The rainfall is geographically quite variable, ranging from as low as 500 mm/yr in parts of the east coast, to up to 10,000 mm/yr around Mt. Taranaki. The mean annual rainfall for the entire island is around 1300 mm/yr. Average evapotranspiration on the island is approximately 700 mm/yr [McKerchar and Pearson, 1996] or ∼50%. The winter months of June, July and August are generally the wettest months of the year and correspond to the highest flows in most rivers. The summer months of December, January, and February generally exhibit the lowest flow conditions.
3. Methods and Data
3.1. Sample Collection
 We collected samples from the North Island of New Zealand from June to July 2010. We sampled river bed sediments and water from 49 rivers, three precipitation events, and eight thermal and nonthermal spring systems. We chose spring and river sample locations based on basin lithology type, location in the arc system, and whether or not they were gauged by the New Zealand National Institute of Water and Atmospheric Research (NIWA) or one of the regional governmental councils. We also chose to sample rivers and spring systems that were previously analyzed by other researchers [Giggenbach et al., 1993; Lyons et al., 2005; Goldsmith et al., 2008] for comparison.
 Each sampling location (Figure 1) was identified with a Global Positioning System (GPS) point and the Eh, pH, and temperature of the sample were measured with a portable pH meter (Fisher-Scientific Accumet AP62), which was calibrated at least twice a day with pH 4, 7, and 10 buffer solutions. All samples were filtered in the field using a 0.45 μm acetate filter attached to an acid cleaned hand pumping filtration unit. The river and spring samples were filtered using separate filtration units. The filtered samples were split into cation and anion fractions; the former were decanted into 250 mL (LDPE) bottles with polypropylene lids and acidified with 0.54 mL HCl/300 mL of water using concentrated HCl with a resulting pH of ∼1.8. The anion samples were decanted into 50 mL (high-density polyethylene) HDPE bottles, which had previously been soaked and rinsed in deionized water. Rain samples were collected in acid washed bucket which was elevated approximately 2 m from the ground and stored in Teflon bottles. Sediments were collected by hand from either the side of each river or from gravel bars and were stored in plastic bags.
3.2. Analytical Procedures
3.2.1. Major Element Determination
 Major element analysis of water samples was completed within 2 months of returning to Dartmouth College. Prior to analysis, samples were stored in a 4°C refrigerator. Acidified samples were analyzed by inductively coupled plasma-optical emission spectroscopy (ICP-OES) for major and minor cations. Quantification was based on comparison to a three-point standard curve for a standard containing all elements which typically yielded 3–4 orders of magnitude linear range in quantification. Verifications were based on National Institute of Standards and Technology (NIST) 1640a and National Research Council (NRC)-SLRS5 water standards. Concentration values were blank and drift corrected based on recovery of a continuing verification standard run each sixth sample. Uncertainties were typically <5% for most elements, based on analysis of standard materials and repeat sample measurement.
 Anion measurements were made by ion chromatography (IC) and conductivity detection using a Dionex DX500. An ion exchange eluent of HCO3/CO3 was used. Quantification was conducted by comparison of analytes to 6-point linear calibration curves, which afford linearity over 3–4 orders of magnitude. IC verification used NIST traceable calibration standards from Inorganic Ventures (Virginia, USA). Uncertainties were again typically <5% for all elements.
3.2.2. Osmium Concentration and Isotope Ratio Measurements
 We separated Os from river/rain water and analyzed Os isotopes following methods described in detail by Chen and Sharma . The internal precision of measured 187Os/188Os ratios for the samples is better than 0.2% (2σ relative standard deviation (RSD)). Repeated (n = 20) measurements of 1 pg laboratory standard (Max-Planck Institute Os-1 standard) gave an average 187Os/188Os = 0.1067 ± 0.1% (2σ RSD), which is within uncertainty of the established isotope value of 0.1069 for this standard. The blank for the above procedure was estimated by analyzing two samples of “Os free” water and is 1.7 fg with 187Os/188Os ratio of 0.27 [Chen and Sharma, 2009]. In this study, we used the same reagents as in Chen and Sharma  and expected that the procedural blank has remained the same. To confirm this expectation, we analyzed a water sample from East Antarctica (CORSACS2 IS-3) that was measured by Chen and Sharma  who found this sample to have an [Os] of 0.74 ± 0.04 fg g−1 with 187Os/188Os ratio of 0.474 ± 0.007. Our measured Os concentration and isotope ratio, 0.70 fg g−1 and 0.474, respectively, is identical to that measured by Chen and Sharma . A procedural Os blank of 1.7 fg is negligible (<1%) for all samples analyzed in this study.
3.2.3. Strontium Concentration and Isotope Ratio Measurements
 Major element, Os and Sr concentrations, 87Sr/86Sr and 187Os/188Os for rivers, precipitation, and spring samples are given in supporting information Table S1.1 The NINZ Rivers are characterized by slightly acidic to slightly alkaline pH (5.97–8.01). Waters from Auckland/Western Waikato display the lowest pH (average 6.45), while waters draining the HAP are the most alkaline (average 7.39). The Piper diagram (Figure 2) reveals that most rivers waters can be described mainly as combination of weathering of carbonate and silicate end members by carbonic acid and to a lesser extent weathering of these lithologies by sulfuric acid derived from sulfide oxidation.
 We used Geochemist's Workbench 9.0 [Bethke and Yeakel, 2009] to estimate saturation states and the stability of various minerals in the rivers. The saturation states of calcite (supporting information Figure S1a) and amorphous SiO2 (supporting information Figure S1b) indicate that only rivers from the HAP with pH values above ∼7.5 are slightly oversaturated with calcite and that all rivers are undersaturated with respect to amorphous silica. Enrichment in Ca-rich and Mg-rich primary minerals has been documented previously in the Taranaki Region [Goldsmith et al., 2008] and Ca-clay and Mg-clay stability diagrams (supporting information Figure S2b) show that the river waters in this study lie mostly in the stability field of kaolinite and Mg-montmorillonite (supporting information Figure S2a). However, several samples from the HAP, Taranaki, Wanganui, and TVZ/Coromandel region are in the stability field of Ca-smectite, and one sample from the HAP lies in the stability field of illite (supporting information Figure S2b).
 Overall, total dissolved solids (TDS) values in the NINZ rivers are higher (23–500 mg/kg) relative to major global rivers (supporting information Table S1), for example the Amazon River basin (6–55 mg/kg [Gaillardet et al., 1999]) and the Congo River basin (28–49 mg/kg [Dupre et al., 1996]). The TDS values appear to be similar to other arc terrains such as Kamchatka (11–407 mg/kg [Dessert et al., 2009]) or the Philippines (74–916 mg/kg [Schopka et al., 2011]). The highest TDS are from the HAP (average = 223 mg/kg, n = 15). The remaining sections of the island range in TDS from 23–248 mg/kg with an average of 80 mg/kg (n = 34).
 The spring temperatures range from 14.7 to 54.4°C with pH values ranging from 4.0 to 8.6. In general, the springs have very distinct chemistries, which appear to reflect their geographic locations on the island and their position relative to the zone of active subduction. As shown in Figure 2, all of the springs are enriched in Na + K relative to Ca and Mg regardless of their location on the island. Most springs, with the exception of Awakeri Spring and the nonthermal spring at Ngawah (NZ-48 and NZ-62, respectively, see supporting information Table S1) from the NISB, are enriched in Cl + SO4 with relatively little or no HCO3. All springs, with the exception of the above two springs, are undersaturated in calcite (supporting information Figure S1a). As expected, all springs are undersaturated in silica (supporting information Figure S1b). In the HAP, the springs are characterized by very high concentrations of Cl− (82,000–440,000 µmol/kg), Mg2+ (160–3500 µmol/kg), Ca2+ (870–71,000 µmol/kg), K+ (220–2300 µmol/kg), and Sr2+ (77–2400 µmol/kg). The estimated TDS for the springs range from 354 to over 25,679 mg/kg. We find, in agreement with Reyes et al. , that springs discharging from the accretionary prism have the highest TDS values, ranging from 5400 to 26,000 mg/kg. Springs discharging in the NISB and in the far eastern edge of the TVZ show the lowest TDS values ranging from 350 to 380 mg/kg. Springs from the Northland are between these two geographic regions in terms of their TDS.
 We measured Sr concentrations and 87Sr/86Sr ratios for 47 of the samples (supporting information Table S1). As expected, samples from the young volcanics draining the TVZ/Coromandel region display the lowest average 87Sr/86Sr values (0.706058) and range from 0.705810 to 0.706467. Their average Sr concentration is 0.3 µmol kg−1. The highest average 87Sr/86Sr values (0.708995) were found within the Permian-Triassic metagraywacke and argillite basins of the NISB. The average Sr concentration of the NISB Rivers is 0.6 µmol kg−1. 87Sr/86Sr isotope ratios of the remaining samples fall in between these two regions. The average Sr concentration of the rivers draining the HAP is about 3.5 µmol kg−1 which is about a factor of 5 higher than that of all other rivers measured. The average flux normalized 87Sr/86Sr ratio of all NINZ Rivers is 0.7075.
 The average 87Sr/86Sr ratios of springs in the HAP and NISB are 0.708238 (± 0.00054, N = 4) and 0.706982 (N = 1) and within the range found in marine carbonates and younger silicates. As noted above, the measured Sr concentrations of the springs in the HAP are the highest and those in the NISB the lowest. The 87Sr/86Sr ratios of springs in the current and ancient arc regions of the TVZ and the Northlands are 0.707905 (N = 1) and 0.706302 (± 0.000014, N = 2), respectively. These isotope ratios are much higher than those reported for subsurface waters such as hot springs in the Philippines (87Sr/86Sr ∼ 0.704) [Schopka et al., 2011] and groundwater wells in the Antilles (87Sr/86Sr ∼ 0.705) [Rad et al., 2006]. These higher 87Sr/86Sr likely are an indication of interaction between the spring waters and older sedimentary and carbonate lithologies, while the studies of Schopka et al.  and Rad et al.  sampled springs draining younger volcanic units.
 The average Os concentrations of the rivers, hydrothermal springs, and rain are 15 pg kg−1 (±15, N = 43), 6 pg kg−1 (± 5, N = 6), and 1.3 pg kg−1 (±0.3, N = 2), respectively (supporting information Table S1). The average 187Os/188Os ratios of the rivers, hydrothermal springs, and rain are 0.52 (± 0.13, N = 43), 0.44 (± 0.09, N = 6), and 0.3 (±0.03, N = 2), respectively. The river [Os] appear to be within the limits of those measured elsewhere (e.g., major global river [Os] = 5–24 pg kg−1 [Levasseur et al., 1999], Rio Orinoco [Os] = 3–84 pg kg−1 [Chen et al., 2006], and Iceland [Os] = 1–20 pg kg−1 [Gannoun et al., 2006]). As expected, the 187Os/188Os ratios in the rivers are also much lower than those measured in continental settings. Similarly, the Os data for the rain samples appear to be in line with those measured by our laboratory [see Chen et al., 2009].
 A close inspection of our data, including replicate measurements described in detail in supporting information Text 2 and Figure S3, however, indicates that the [Os] and 187Os/188Os ratios of samples stored in trace element clean acid-washed LDPE bottles may have been compromised by Os leaching out from bottle walls! These include all but the rain samples, which were stored in Teflon bottles. The LDPE bottles were specially obtained for this project from Dynlab Corporation as they were found to be clean for a majority of trace elements. They were cleaned in the following manner: (1) water + soap, (2) 20% distilled HNO3 (1 month), (3) concentrated doubled-distilled HCl (overnight). Sample bottles were only opened when acid was added in New Zealand, during sampling and in the clean lab at Dartmouth. Osmium contamination by LDPE and HDPE plastic sample bottles has recently been documented by Sharma et al. , and is described in detail therein. Our work further supports the contamination issues highlighted in that paper. With regards to this study, we believe that the large blank from the sample bottles and the effect it has on the Os isotope composition and concentration of the waters precludes us from making any quantitative evaluations of relationships among Os, chemical weathering and various lithological parameters. However, the data can be used to provide an upper bound on Os flux from the NINZ.
 The aim of this study is to quantify chemical weathering taking place in different regions on the NINZ and to determine how this relates to the lithology of the river basins, CO2 consumption rates and to fluxes of Os and Sr. In the following, we first calculate bulk chemical erosion and CO2 consumption rates and fluxes of Sr and Os for the regions defined in this study. We additionally provide chemical erosion rates and CO2 consumption rates reported for other parts of the world to place our data in the context of previously observed weathering rates and their associated CO2 consumption.
5.1. Chemical Weathering and CO2 Consumption
 Over the past decade it has been realized that over a million year time scale the atmospheric CO2 consumption from carbonic acid weathering of silicates may be offset by sulfuric acid weathering of carbonates that releases CO2 into the atmosphere [e.g., Spence and Telmer, 2005; Calmels et al., 2007]. Thus to assess chemical weathering and CO2 consumption using riverine chemistry the mass balance equations need to incorporate sulfuric acid weathering. We quantified chemical weathering rates following Spence and Telmer  and Dessert et al. . This method is described in detail in the supporting information Text 1. Briefly, we first obtained an inventory of riverine ions produced from low temperature weathering reactions by removing rain and spring inputs from river water. Spring corrections were made to rivers in which there is a spring system identified and characterized by Reyes et al.  and the corrections were made using the average spring composition for each region given in supporting information Table S2. The percentage of spring-rain input to the rivers is <17% in all regions and therefore the spring-rains contribution is small but significant. As sulfuric acid and carbonic acid weathering of carbonates and carbonic acid weathering of silicates all produce bicarbonate ions (=dissolved inorganic carbon, DIC), we used method from Spence and Telmer  to assess the relative proportions of DIC from these sources in each of the NINZ regions defined in Figure 1. A key assumption of this method is that the ratio of sulfuric acid consumed by silicate and carbonate weathering is identical to that of carbonic acid consumed by silicate and carbonate weathering.
 We used NIWA's Water Resource Explorer (http://wrenz.niwa.co.nz/webmodel/) to estimate mean annual discharge data. This web-based geographic information system (GIS) hydrologic modeling tool uses long term gauge data to calculate mean annual discharges for gauged rivers and also provides catchment areas for all sampled river basins. Bulk chemical weathering yields were then calculated by multiplying the rain corrected TDS measurements by mean annual discharge and then dividing by the catchment area. CO2 consumption rates (ΦCO2) were estimated by multiplying the DIC derived from carbonic acid weathering of silicates with the mean annual discharge and dividing it by the catchment area. As described in supporting information Text 1, the estimated uncertainty in ΦCO2 is of the order of 56%. We find that the DIC derived from silicate weathering comprises >50% of the total DIC flux in all regions except for in the HAP and the TVZ/Coromandel regions (Figure 3). In the HAP and the TVZ/Coromandel regions, the majority of bicarbonate is being produced from weathering of carbonates via carbonic and sulfuric acids, respectively. Carbonate-derived DIC accounts for between 24 and 42% of the total DIC flux on the NINZ.
 Sulfuric acid weathering does not appear to be a major contributor to the DIC budget except for in the TVZ/Coromandel region where it accounts for 38%. This possibly provides further insight into why the weathering calculation method we adopted from Spence and Telmer  yielded inconsistent results for all but one river in the TVZ/Coromandel region and no consistent results in the Northlands. When we examine the spring data from Reyes et al. , we indeed find that these two regions have spring systems with the highest average SO4 concentrations and so it is possible that the spring influence invalidates the assumption that carbonic and sulfuric acid weathering are occurring in equal proportions.
 We estimate ΦCO2 in Taranaki range from 210–946 × 103 mol CO2 yr−1 km−2 (Table 1). In comparison, Goldsmith et al.  reported ΦCO2 in the Taranaki region between 217 and 2926 × 103 mol km−2 yr−1. As Goldsmith et al.  sampled the same rivers at a different time during the year, the discrepancy between the estimated ΦCO2 may be explained by comparing hydrological data from the two sampling periods. We find that we sampled rivers in the Taranaki region at significantly higher flow rates (∼475–1000% higher) than when sampled by Goldsmith et al. . Sampling at higher flow rates generally introduces dilution effects and may explain the significantly lower ΦCO2 we observed in Taranaki. In the NISB, we find slightly more concordance between our weathering and CO2 consumption rates and what has been previously reported [Lyons et al., 2005] (Table 1). The observed discrepancies among the data sets reinforce the need for more long term weathering studies that would capture the inherent variability in the riverine systems.
Table 1. Chemical Erosion, CO2 Consumption Rates, and Sr and Os Flux Calculationsa
 On average the greatest consumption of CO2 in the NINZ occurs in the HAP. The average ΦCO2 in Taranaki is also high but is slightly lower than in the HAP. The average ΦCO2 on the NINZ is 557 × 103 mol km−2 yr−1. It is comparable to CO2 consumption estimated for the Andes (=378 × 103 mol km−2 yr−1) [Edmond and Huh, 1997] and the Deccan Traps (=50–760 × 103 mol km−2 yr−1) [Das et al., 2005] but much lower than that observed for volcanic arc systems lying in tropical settings. For example, the ΦCO2 for: (a) Java arc is 6410 × 103 mol km−2 yr−1 [Dessert et al., 2003], (b) Philippines Islands is 3580 × 103 mol km−2 yr−1 [Schopka et al., 2011], and (c) Lesser Antilles is 300–3500 ×103 mol km−2 yr−1 [Rad et al., 2006; Goldsmith et al., 2010; Gaillardet et al., 2011]. While lower compared to tropical islands, the ΦCO2 values for the NINZ are still much higher than those reported for the continental drainages (e.g., Congo-Zaire in Table 1). Gaillardet et al.  estimated that silicate rock weathering on volcanic islands and arcs constitutes more than 30% of the continental ΦCO2. Unfortunately, so far only a few systematic studies have been undertaken to better assess ΦCO2 from island arcs. A recent study by Schopka et al.  points out that tropical arcs occupy about 1% of land area but may consume up to 10% of global annual CO2. This observation has been further substantiated in the Lesser Antilles by Goldsmith et al.  and Gaillardet et al. ; the latter study points out the importance of relief in impacting the regional rainfall pattern, which in turn, impacts chemical weathering and CO2 consumption. Due to the large variations in the ΦCO2 between this work and other studies on the NINZ it is evident that a number of year-round and systematic studies of key island arcs need to be undertaken to refine the figure for CO2 consumption via island arc weathering.
5.2. Strontium Fluxes
 The overall Sr flux of for the entire island is 1411 mol Sr yr−1 km−2. The estimated Sr fluxes for individual regions are directly related to the total chemical erosion (Table 1). In concurrence with the high concentrations found for nearly all other elements in HAP, we find that the annual flux of Sr in this region is extremely high relative to the rest of the island (Figure 4). We also find that flux of Sr in the accretionary prism is much higher than most other arc systems reported in the literature (Figure 4 and Table 1). This is curious as the prevalence of volcanic lithologies is used to explain the high dissolved fluxes observed in arc systems. The Sr fluxes calculated for other sedimentary regions of the NINZ are similar to those reported for other arc terrains. Of significance here is the observation that the lithological units composing the accretionary prism are not drastically different compared to the lithologies of other sedimentary basins we sampled and main difference appears to be the spring systems discharging in the accretionary prism. The extremely high concentrations of Sr in the springs in this region lead us to believe that the large flux of Sr that we calculate is due in part to the influence of springs.
 The Sr fluxes of the volcanic arc regions of the island appear to be fairly typical when compared to other island arc systems (Table 1). The volcanic region of Taranaki lies between Sr fluxes observed for the island arc of the Lesser Antilles and the volcanic island of Reunion, while the arc volcanics in the TVZ/Coromandel region fall slightly below these islands. The remaining regions of the island exhibit variability in their Sr fluxes but all fall within the range of data reported for other arc systems and nonarc related volcanic regions. Most importantly, our data suggest that hydrothermal spring fluids in fore arcs may boost the overall flux of Sr from these regions. This is an unexpected result and would imply that the sedimentary fore arc region in arc systems may be equally if not more important to volcanics in supplying Sr to the oceans.
5.3. Osmium Fluxes
 As documented in supporting information Text 2, we are not able to obtain an accurate assessment of the Os concentration and isotope composition of NINZ riverine flow due to Os leaching from storage bottles. As inferred in Sharma et al.  Os-rich metal nuggets incorporated in bottle walls may get exposed when the bottles are being precleaned with acids especially with HNO3, which tends to oxidize the plastic. We recommend that for future Os studies water samples need to be stored in Teflon precleaned using the recipe presented in Sharma et al.  and Peucker Ehrenbrink et al. .
 Polyethylene bottles have been used regularly in studies of aqueous Os geochemistry. If the contamination problems we encountered are always associated with these materials it may be necessary to reconsider previous reports of [Os] and 187Os/188Os in waters. For example, Li et al.  observed a good correlation between SO42−/Si and Os/Si ratios for large rivers worldwide and argued that this indicates sediment weathering exerts a dominant control on continental Os fluxes, since SO42− primarily comes from weathering of sedimentary rocks. Interestingly, we find a similar correlation (Figure 5) with the rivers on the NINZ despite evidence of extensive and variable sample contamination from LDPE bottles. Li et al.  used the slope of the linear regression shown in Figure 5 to quantify the amounts of Os coming from sediment and silicate weathering. The correlation among SO42−/Si and Os/Si may indicate that sediment weathering is a large contributor of Os, as suggested, however the slope of this correlation could be affected by sample contamination. A regression fit to Rio Orinoco River samples from Chen et al.  has a much steeper slope (Figure 5) suggesting that these samples could also be impacted. The St. Lawrence River has an anomalously high Os/Si suggesting possible contamination. However, of all the rivers included in Figure 5, the St. Lawrence is the only sample stored in Pyrex glass. Since Pyrex glass, in general, has low Os blank [Sharma et al., 2012] it is likely the anomalously high Os/Si ratio of this river results from anthropogenic effects as inferred by Li et al. . With the caveat that Os concentrations in NINZ waters may have been variably impacted with contamination, we calculate the largest fluxes of Os for the HAP (avg = 160 µmol Os km−2 yr−1, n = 15). The remaining regions of the island have average Os fluxes ranging from 21 to 97 µmol Os km−2 yr−1. These estimates provide an absolute upper limit to the amount of Os coming from the NINZ and appear similar to those estimated for other large basins (Table 1).
5.4. Does NINZ Provide Unradiogenic Sr and Os to the Oceans?
 The average flux weighted 87Sr/86Sr ratio of the NINZ Rivers is 0.7075. One of the surprising outcomes of this study is that the Sr isotope signatures of the rivers draining the NINZ are not particularly unradiogenic and likely represent a mixture of weathering of older sedimentary and volcanic lithologies with some less radiogenic inputs coming from more recent arc volcanics (Figure 6). Flux weighted 87Sr/86Sr ratios in the currently volcanic regions of TVZ and Taranaki are 0.7061 and 0.7065, respectively. The highest flux weighted 87Sr/86Sr ratio is from the rivers draining the NISB (=0.7090). The Sr isotope mixing diagram (Figure 6) shows that (a) the Sr in HAP springs is likely from the dissolution of marine carbonates that are present in copious amounts in this area [Nyman et al., 2010] and (b) the springs may provide bulk of Sr to the HAP rivers. The springs from other parts of the NINZ are also relatively radiogenic. If the springs sampled in this study are representative of the average chemistry of subsurface fluids draining the NINZ, the subsurface flux of water from NINZ cannot be a source unradiogenic Sr to the oceans.
 Flux normalized 187Os/188Os ratios from the arc and from the accretionary prism are ≥0.43 and ≥0.68, respectively. Overall, we find waters draining the NINZ to have an 187Os/188Os ratio of ≥0.51. This observation suggests that although the arc is contributing less radiogenic Os relative to the sedimentary fore arc region both regions have 187Os/188Os ratios that are much higher than the mantle (=0.13). Overall the NINZ does not appear to be a source of unradiogenic Os even when the samples from the volcanically active Taranaki region display 187Os/188Os ratios in the vicinity of about 0.2 (supporting information Table S1).
 From the above observations it is evident that the NINZ does not provide relatively unradiogenic Sr and Os isotopes to the oceans. Is NINZ an atypical island arc in this regard? The answer appears to be no. For example, the 87Sr/86Sr ratios of the rivers draining a part of the Philippines volcanic arc vary from 0.7034 to 0.7064 [Schopka et al., 2011]. Data compiled from the Georoc database (georoc.mpch-mainz.gwdg.de) show that a majority of island arc lavas are quite radiogenic (supporting information Table S4). This is mainly due to interaction of mantle-derived magmas with ancient rocks (with radiogenic Sr and Os) prior to eruption. Also, the lavas have extremely Re/Os ratios leading them to rapidly grow radiogenic 187Os. Data compiled from the Georoc database also show that the median 187Re/188Os ratio of island arc volcanics is 361. The 187Os/188Os ratio of recently erupted lava with 187Re/188Os ratio of 361 would grow by 10% in less than 3 Ma! The above observations are likely to be valid for all active island arcs that are built on an old edifice and/or have an ancient fore-arc sedimentary prism and/or have magmas intruding ancient sedimentary regions.
 We carried out an extensive sampling campaign of rivers and springs over much of the North Island of New Zealand. We sought to characterize chemical weathering processes, gain insight into the influences of spring systems on riverine chemistry, to assess the impact that the NINZ may be having on the Os and Sr budgets of the ocean and to determine whether these isotope systems can act as good proxies of CO2 consumption by silicate weathering. The following principal conclusions can be drawn using our data:
 1. The CO2 consumption rates on the NINZ are high relative to many shield/basement river systems and are lower than the high consumption rates observed for arc systems in the tropics [e.g., Dessert et al., 2003; Rad et al., 2006; Goldsmith et al., 2010; Schopka et al., 2011; Gaillardet et al., 2011]. On average, the greatest consumption of CO2 in the NINZ occurs in the Hikurangi accretionary prism, however.
 2. We observed that the largest flow of relatively radiogenic Sr is from the Hikurangi accretionary prism. Spring systems sampled in this region display extremely high concentrations of Sr, and we believe the large Sr fluxes in the surface waters are a reflection of this. The remaining regions of the island have Sr fluxes that lie between those observed for oceanic and continental arc systems [e.g., Louvat and Allegre, 1997; Edmond and Huh, 1997; Rad et al., 2006; Schopka et al., 2011].
 3. The NINZ is not a source of unradiogenic Sr to the oceans. The average flux weighted 87Sr/86Sr ratio for the NINZ rivers (86Sr/87Sr = 0.7075) is similar to that observed for Phanerozoic marine carbonates and likely reflecting contributions from weathering of older volcanic and sedimentary lithologies that are present throughout much of the island. We find similar 87Sr/86Sr values for the spring systems on the NINZ, indicating that subsurface flows from the island to the ocean may not be drastically different than what we observe for surface waters. This runs contrary to recent suggestions that subsurface flows in island arcs provide large fluxes of unradiogenic Sr.
 4. We observed variable amounts of unradiogenic Os leaching from trace element clean acid leached low density polyethylene bottles that were used for sampling. Experiments revealed both the concentration and the isotopic composition of waters stored in these bottles are compromised. The slope obtained from the regression between SO42−/Si and Os/Si using data from previous studies may be erroneous and utilizing it to assess Os fluxes may give incorrect results. Our data from the Hikurangi accretionary prism indicate that sediment fills in the fore arc basins may provide a disproportionately larger flux of Os compared to their area. The isotopic character of these waters has been lowered by sample contamination. Even with this lowered 187Os/188Os the NINZ does not appear to be a large source of unradiogenic Os and suggests that arc systems may not provide the missing source of mantle-derived Os to the oceans.
 We wish to acknowledge Charles Lee and Steve Cameron at the University of Waikato for allowing us to ship clean acid to their laboratory and for use of their fume hoods. We thank Joshua Landis for his help in major element analysis and his intellectual input throughout the project. We are grateful to Jérôme Gaillardet and Jody Spence for clarifying some of the weathering calculation used in the text. We also wish to thank Lou Derry and Sunil Singh and an anonymous reviewer for constructive comments which greatly increased the quality of this manuscript. This paper constitutes bulk of TBs MS Thesis at Dartmouth. Funding for the field work at NINZ came through a graduate student research grant to TB by the Geological Society of America.