Organic carbon yields from small, mountainous rivers, New Zealand



[1] Small mountainous rivers (SMR) have recently been recognized as important contributors of organic carbon to the marine environment. We present dissolved and particulate organic carbon fluxes from previously unexamined rivers of the southwestern region of New Zealand's North Island and additional flux data from the Southern Alps, South Island, New Zealand. The dissolved fluxes are relatively low and show little variation. The particulate fluxes are less than our previous estimates but still high compared to other non-SMR watersheds.

1. Introduction

[2] Organic carbon transport from the continents to the ocean is a major process in the global carbon cycle [Ludwig et al., 1996; Raymond and Bauer, 2001]. Estimates of this flux vary greatly in part due to our lack of quantitative estimates from much of the world [McKee, 2003]. However, it is now clear that small mountainous rivers may be a much more important source of organic carbon to the ocean than previously thought [Kao and Liu, 1996, 1997; Lyons et al., 2002; Blair et al., 2003; Gomez et al., 2003; Komada et al., 2004]. The total organic carbon (TOC) flux from Oceania's SMR alone has been estimated to be 21–38% of the total oceanic input [Schlünz and Schneider, 2000] but is poorly constrained. TOC flux from rivers to the marine environments is estimated to be ∼55% in the dissolved form (DOC) [Ludwig et al., 1996]. TOC flux may be greatly affected in the future by changes in climate [Tranvik and Jansson, 2002]. Thus it is important that better, more quantitative fluxes be determined. Few data exist on DOC concentrations and hence the fluxes of DOC from SMR. Herein we present DOC data from SMR in New Zealand. We use organic carbon data from sediments collected at the same time as the DOC samples to estimate particulate organic carbon (POC) yields from these watersheds. These data add additional information on the flux of TOC from SMR.

2. Site Description

[3] DOC and POC samples were collected from rivers in the Taranaki-Wanganui andesite/rhyolite volcanic region of western North Island and from the rapidly uplifting, tectonically active Southern Alps of western South Island (Figure 1). In the Southern Alps, the denudation rate is approximately equal to the uplift rate, thereby maintaining relief of 2–4 km [Hovius et al., 1997]. High denudation rates require tremendous amounts of material be transported out of the watersheds by stream flow [Hicks and Griffiths, 1992]. Samples were also collected from two watersheds in the greywacke terrain of the eastern South Island, the Waimakariri and the Waiau, whose watersheds receive less rainfall than their west coast neighbors.

Figure 1.

Map of New Zealand showing locations sampled for this study. Inset maps of southwestern North Island and central South Island show the rivers sampled for this study. Dots represent downstream sampling location for which yields were determined (Table 1). Locations of all sampling areas are given in Table 1 of the auxiliary material for this article.

[4] In North Island watersheds 40–100% of the indigenous vegetation has been lost since human occupation. On South Island, especially in the higher elevations, there has been little to no loss of the indigenous vegetation since human occupation [Leathwick et al., 2003]. Watersheds draining to the east on South Island have lost as much as 100% of their indigenous vegetation, especially near the coast [Leathwick et al., 2003]. Indigenous vegetation has been replaced there by agriculture, non-native vegetation or scrub and tussock. The nature of the soils in these regions reflects the parent material, the age of the soil and the climate [Leathwick et al., 2003].

3. Methods

[5] In March 2004 water and sediment samples were collected by hand from rivers in New Zealand (Figure 1). Only one sample was taken per sampling location. Water samples were collected in polyethylene bottles that had been cleaned by rinsing 5 times with 18.2 MΩ deionized (DI) water, soaked for 1 week in DI water, rinsed 5 times with DI water, then refilled with DI water and stored until use.

[6] Samples for DOC analysis were taken from the polyethylene bottles and vacuum filtered directly into 20 ml amber glass vials through Millipore™ 0.7 μm nominal pore size glass fiber filters. Filters, vials, and filter funnel were combusted before use at 450°C for 3 hours and rinsed with 10% v/v HCl in DI water. After filtration, samples were acidified to pH 2 with concentrated HCl and stored at 4°C. A Shimadzu 5050A Total Organic Carbon analyzer equipped with a high sensitivity catalyst was used for DOC determinations. The high sensitivity catalyst allowed for quantification to approximately 0.1 mg L−1 of DOC. Samples were sparged with hydrocarbon-free air for up to 7 minutes to ensure all purgeable organic carbon and inorganic carbon were removed. The remaining non-purgeable organic carbon in the sample was calculated using 3–4 injections per sample. The relative standard deviation (RSD) of multiple injections of individual samples ranged from <1% to 3.5%.

[7] Sediment samples were collected in plastic ziplock bags or in polypropylene jars with snap caps. Within 4 weeks of collection, samples were dried overnight at 105°C and then combusted for 2 hours at 550°C to determine their organic carbon content as 33% of the loss on ignition [Lyons et al., 2002; Hunt, 1981]. RSD of multiple analyses ranged <1% to 18.2%, but most were <10% for South Island samples. For North Island samples, RSD of multiple analyses ranged from 1.4% to 46%, but most were <20%.

[8] Reconnaissance-level estimates of the annual average DOC and POC fluxes were computed as follows. For DOC, the average concentration of the DOC samples was combined with the annual average water discharge. For POC, the fraction of organic carbon in the sediment sample was multiplied by the average annual suspended sediment loads. Errors in the suspended sediment yields were based on error of the estimate from sediment rating curves as in the work by Hicks et al. [2000].

4. Results and Discussion

[9] DOC and POC data (POC is the fraction of the total sediment that is organic carbon) are plotted in Figure 2. The DOC data fall into the range <0.1–4.0 mg C L−1, with the exception of a small tributary of the Whanganui River where the DOC value was rather high, 7.6 mg L−1. DOC for South Island streams draining the Southern Alps was much lower (range <0.1–1.35 mg L−1; mean 0.38 mg L−1; median 0.6 mg L−1) than those for southwestern North Island (range 0.8–7.6 mg L−1; mean 2.39 mg L−1; median 1.9 mg L−1). With the exception of one sample from a small tributary of the Whanganui River (the Makirikiri, in an agricultural area), all POC values were <1.0%. In general data from each island form a group, with the North Island sites having similar POC concentrations to the South Island sites but having higher DOC concentrations. Two-tailed t-test with a null hypothesis of no difference in mean POC for North and South Island samples retains the null hypothesis of no difference in mean POC (α = 0.05, p = 0.166). For the DOC data, the t-test rejects the null hypothesis of no difference in mean DOC for the two islands (α = 0.05, p = 0.00012). The South Island sites with the highest DOC values are those from the Grey (1.75 mg L−1) and Buller Rivers (1.35 mg L−1). It is unclear whether these higher DOC values reflect landuse, lithology or sediment yield. These two rivers have greater anthropogenic activity at higher elevations, flow through varied lithology and have lower erosion rates than rivers to the south.

Figure 2.

Dissolved organic carbon of stream water and organic carbon content of fine-grained sediment from New Zealand streams sampled in March 2004. Circles represent samples from North Island and squares represent samples from South Island. Error bars represent the relative standard deviation of 3–5 replicate analyses.

[10] Are these DOC fluxes representative? Duncan [1992] classified New Zealand rivers as having slightly variable flow when the coefficient of variation (CV) of the flow is between 0.85–1.25. Of the rivers we investigated, only the Haast, Hokitika/Cropp, Otaki and Waitotara have CVs of flow slightly higher than this, reflecting higher baseflows and more frequent flood events. Winter rainfall events increase flow in North Island rivers, while South Island rivers demonstrate a more even distribution of flow [Duncan, 1992]. In high elevation, low-order streams in the Southern Alps, “carbon flushing” on the rising limb of the hydrograph and “dilution” on the falling limb have been documented [McGlynn and McDonnell, 2003], leading to 3–4-fold differences in DOC concentrations. Our samples were collected at the lowest end of the drainages in order to investigate the carbon fluxes from the entire watershed. Southern Alps rivers were sampled on the falling limb of a hydrograph and North Island rivers during baseflow, but approximately 6 weeks after a 100–250 year flood event, and may represent a lower limit of DOC flux. However, the yearly flux calculated by Moore [1989] for two undisturbed catchments in the headwaters of the Grey River, determined over a 4-month period, was 6.8 tons km−1 yr−1, similar to our value for the entire Grey system of 5.2 tons km–2 yr−1.

[11] DOC concentrations, especially from rivers draining the Southern Alps, are low by terrestrial water standards [Thurman, 1985]. Median DOC in these New Zealand streams is approximately only twice that of marine-derived rainfall [Willey et al., 2000]. High elevation portions of these watersheds have very high rainfall rates, thin soils and rapid flow through, as evidenced by DOC values as low as <0.1 mg L−1 in the Cropp River. The Cropp watershed lies at the highest elevation and is the smallest watershed investigated. It receives an average of 12,000 mm of precipitation annually and it has mean soil residence time of as little as 100–200 years [Basher et al., 1988]. The soil organic carbon concentrations in the Cropp drainage basin decrease rapidly from between 1–7% in the top few cm to values approaching zero by 10–15 cm depth [Tonkin and Basher, 2001].

[12] We have calculated the organic carbon yields from the largest of these New Zealand watersheds using available flow and sediment flux data (Table 1). The DOC yields vary by an order of magnitude, but, in general, are similar to those yields published for other larger watersheds (Table 2). However, New Zealand DOC yields are lower than from tropical high-standing islands such as Puerto Rico [McDowell and Asbury, 1994] and Papua New Guinea [Burns et al., 2001] and from small mountainous watersheds in the Himalaya [Sharma and Rai, 2004], but within the range of larger, more temperate to sub-polar rivers [Strieg et al., 2004; Telang et al., 1991; Martin and Probst, 1991; Hope et al., 1997; Clair et al., 1994; Callahan et al., 2004] (Table 2).

Table 1. Organic Carbon Yields From New Zealand Streams
StreamWatershed Area Upstream of Sampling Site (km2)POC Yield(tons km−2 yr−1)aDOC Yield(tons km−2 yr−1)DOC/TOCPrimary Geology
  • a

    Error in POC yield estimate based on error of estimate from the sediment rating curve, assuming negligible error in water discharge determination, as in the work by Hicks et al. [2000].

  • b

    — means sample not collected.

  • c

    Data in parentheses from Lyons et al. [2002].

Waitara11222.9 ± 26%3.252%Quaternary lahar and sandstone/mudstone
Waitotara10982.3 ± 100%bMarine sandstone/mudstone
Whanganui67853.8 ± 15%2.439%Marine sandstone/mudstone
Whangaehu19442.6 ± 35%1.638%Marine sandstone/mudstone
Rangitikei35411.3 ± 32%Marine sandstone/mudstone
Waiau162617.4 ± 21%3.115%Greywacke and Quaternary alluvium
Waimakariri31057.5 ± 30%0.577%Greywacke and Quaternary alluvium
Buller63791.4 ± 34%2.967%Mixed; igneous and alluvial
Grey38311.6 ± 28%5.276%Schist and Quaternary alluvium
Cropp1281 ± 12% (52)c1.11%Alpine schist
Hokitika36344 ± 25% (43)c2.15%Alpine schist
Haast1020(168 ± 29%)cAlpine schist
Table 2. Organic Carbon Yields From Other Locations
LocationPOC Yield (tons km−2 yr−1)DOC Yield (tons km−2 yr−1)DOC/TOC
British rivers (median)d3.2
Canadian Atlantice2.9
St. Lawrenceb0.271.3585
Puerto Rico forested montanef0.4–5.22.6–11
Sepik, Papua New Guineag2–5.84.9–15.571–73
Himalaya, Sikkimh286770
Pearl, Chinai2.95

[13] POC yields for some Southern Alps watersheds are quite high, and compare favorably to the yields determined from previous work on the Hokitika River [Lyons et al., 2002]. The Cropp is a high elevation portion of the Hokitika watershed, and the differences between the present and earlier work (36%) may represent seasonal differences or even interannual variability. The Buller and Grey Rivers have much lower POC yields than rivers to the south and reflect, in part, the lower sediment yields from the Buller and Grey watersheds [Hicks et al., 1996]. Lower sediment yields are due in part to the different lithologies. The Cropp, Hokitika and Haast watersheds are dominated by the fissile alpine schist. In addition, the Buller and Grey watersheds are farther from the Alpine Fault, have lower uplift rates than those experienced by the southern set of rivers, and have less rainfall overall. The rivers draining the eastern slope of the Southern Alps, the Waimakariri and the Waiau, have intermediate POC yields (Table 1), while rivers from the North Island have yields similar to the Buller and Grey watersheds. Much more of the total organic carbon yields of the North Island rivers and from the Buller and the Grey Rivers are of DOC. The portion of the total OC yield attributable to DOC from these rivers falls close to what is argued to be the global average of ∼55% [Ludwig et al., 1996]. Yet in a few of these watersheds, the DOC dominates the yield. DOC yields of these New Zealand streams are similar to those of larger rivers (Table 2) but DOC constitutes only a small portion of the total organic carbon yield due to the far greater POC yields of many of these New Zealand streams, particularly those draining alpine schist and greywacke of the Southern Alps.

[14] DOC yields range over an order of magnitude in these systems, with highest yields from the Grey River watershed and lowest from the Waimakariri watershed (Table 1). There is no apparent geographic trend in the yields. There are similar values from North Island rivers and Southern Alps rivers draining both to the east and the west of the main divide. Aitkenhead and McDowell [2000] have summarized all the available riverine DOC fluxes and yields and demonstrated that the yield could be predicted by the C:N ratio of the soil in each watershed. The C:N ratio is in turn related to biome type within the watershed. The higher rainfall regions of the western Southern Alps are dominated by mixed coniferous/hardwood forests, while the areas east of the main divide contain primarily tussock grassland, introduced pasture and remnants of scrubland and beech forest [Tonkin and Basher, 2001].

5. Summary

[15] We have presented total organic carbon data for several small mountainous rivers in New Zealand. DOC yields from these systems are similar to those reported in other watersheds, no matter the size, throughout the world. The data suggest that the yield may vary with biome type within the watershed (i.e., vegetation and landuse). The watersheds with the highest sediment yields also have high POC yields, as we have previously reported. Although recent work on other SMR systems in the tropics indicates high POC fluxes to the marine environment [Jennerjahn et al., 2004], this present work shows that not all the New Zealand river systems have high POC yields and that in some of these rivers the organic carbon flux is greatly influenced by the DOC yields. As demonstrated by Lyons et al. [2002], POC yield in New Zealand scales to the size of the watershed and hence the total sediment yield. These new data from the North Island, when incorporated with our knowledge of POC yields from previous investigations, lower the estimate of overall POC flux from New Zealand. The DOC flux appears more influenced by vegetation, landuse and probably soil development.


[16] This work was supported by NSF EAR 0309755. We thank Zdravka Karanovic for help with sample collection and two anonymous reviewers whose suggestions greatly improved the manuscript.