Abstract
Riverine organic matter supports of the order of one-fifth of estuarine metabolism1. Coastal ecosystems are therefore sensitive to alteration of both the quantity and lability of terrigenous dissolved organic matter (DOM) delivered by rivers. The lability of DOM is thought to vary with age, with younger, relatively unaltered organic matter being more easily metabolized by aquatic heterotrophs than older, heavily modified material2, 3, 4. This view is developed exclusively from work in watersheds where terrestrial plant and soil sources dominate streamwater DOM. Here we characterize streamwater DOM from 11 coastal watersheds on the Gulf of Alaska that vary widely in glacier coverage (0–64 per cent). In contrast to non-glacial rivers, we find that the bioavailability of DOM to marine microorganisms is significantly correlated with increasing 14C age. Moreover, the most heavily glaciated watersheds are the source of the oldest (~4 kyr 14C age) and most labile (66 per cent bioavailable) DOM. These glacial watersheds have extreme runoff rates, in part because they are subject to some of the highest rates of glacier volume loss on Earth5. We estimate the cumulative flux of dissolved organic carbon derived from glaciers contributing runoff to the Gulf of Alaska at 0.13 ± 0.01 Tg yr-1 (1 Tg = 1012 g), of which ~0.10 Tg is highly labile. This indicates that glacial runoff is a quantitatively important source of labile reduced carbon to marine ecosystems. Moreover, because glaciers and ice sheets represent the second largest reservoir of water in the global hydrologic system, our findings indicate that climatically driven changes in glacier volume could alter the age, quantity and reactivity of DOM entering coastal oceans.
Biogeochemical cycling in coastal margins near riverine outflows is dominated by the influx of terrestrial organic matter and nutrients. The effect of anthropogenic increases in nutrient export on sensitive systems is well-documented in regions such as the Gulf of Mexico zone of hypoxia6. It is much less clear how climate-induced shifts in the export of terrigenous DOM will affect coastal environments, although the reactivity of this carbon will be key as the extent of its incorporation into marine food webs depends largely on its chemical character7. Riverine DOM is typically dominated by allochthonous material derived from plant detritus, which may be substantially aged and degraded before entering the marine environment, thereby reducing its availability to marine heterotrophs3. Glacial ecosystems are devoid of higher plants but contain abundant microbial communities adapted to extreme temperature, light and nutrient regimes8. Thus, we hypothesize that glacial watersheds export chemically distinct DOM compared to DOM derived from litter and soil organic matter in forested watersheds.
The Gulf of Alaska (GOA) drainage basin contains more than 10% of the mountain glaciers on Earth5, and annual runoff from this region represents the second largest discharge of freshwater to the Pacific Ocean9, 10. Along the GOA, watersheds dominated by glaciers have extremely high water fluxes (commonly >6 m yr-1), such that yields of DOM from these watersheds are substantial11 even with the low concentrations of DOM typical of glacial ecosystems12, 13. Future changes in glacial runoff are markedly larger than those projected for other components of the water cycle, and the continued recession of GOA glaciers with climate warming is expected to shift the timing and magnitude of riverine DOM delivery to downstream coastal ecosystems11.
In this study we characterized the source, bioavailability and magnitude of the DOM flux from glacial watersheds that drain into the GOA. We sampled streamwater DOM from 11 coastal watersheds in three geographic areas along the GOA during peak glacial runoff (Fig. 1). All three study areas are located in coastal temperate rainforest biomes characterized by maritime climates, with mean annual sea level temperatures ranging from 2.5 to 4.5 °C. The 11 watersheds are largely free of human disturbance, and land cover is dominated by coniferous forest in the lower elevations of all of the watersheds. Glacier coverage varied markedly (0–64%) within the watersheds owing to differences in elevation and distance from the ocean. Here we use glacier cover as a relative estimate of the proportion of streamwater derived from glacial runoff in each watershed. All riverine water samples were collected above tidal influence within 6 km of tidewater. In addition to measuring bulk dissolved organic carbon (DOC), we used fluorescence spectroscopy, the vascular plant biomarker lignin, and carbon isotopic analyses (δ13C and Δ14C) to evaluate the chemical character, source and age of riverine DOM exported from the 11 watersheds.
Figure 1: Map of the Gulf of Alaska drainage area.
![Figure 1 : Map of the Gulf of Alaska drainage area. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.com](https://webarchive.library.unt.edu/web/20170201184834im_/http://www.nature.com/nature/journal/v462/n7276/images/nature08580-f1.0.jpg)
a, The region draining into the GOA covers 420,500 km2 within the United States and Canada and contains 75,300 km2 of glaciers (shown in dark grey). b–d, Eleven watersheds (named in bold font) around the GOA were sampled within 6 km of their estuaries in three locations: Girdwood (b), Cordova (c) and Juneau (d). The watershed areas ranged from 37 to 430 km2 (mean, 201; s.e., 41).
High resolution image and legend (109K)Download Power Point slide (490K)Slides may be downloaded for educational use, according to the terms described in Nature Publishing Group's licensing policy.
Streamwater DOC concentrations ranged from 0.6 to 2.2 mg C l-1 and were negatively correlated with watershed glacier coverage (Table 1; r2 = 0.63, P = 0.003). This is consistent with higher inputs of soluble carbon from forest soils in the watersheds with lower glacier coverage13. Fluorescence spectroscopy and δ13C-DOC measurements both indicated that increased inputs of glacial meltwater were associated with an increase in the proportion of DOM derived from microbial sources. DOM fluorescence associated with proteinaceous material was positively correlated with watershed glacier coverage (Table 1; r2 = 0.82, P = 0.001), indicating that DOM in heavily glaciated watersheds has a high protein content and relatively low concentrations of aromatic organic compounds that typically dominate streamwater DOM14. The δ13C values for streamwater DOC ranged from -25.6‰ to -22.0‰ and similarly demonstrated significant enrichment with increasing watershed glacier coverage (Table 1; r2 = 0.53, P = 0.01). The δ13C-DOC values in the heavily (>40%) glacial watersheds were all enriched compared to the range for C3 plant-derived carbon (about -25 to -30‰)3, 15, which is consistent with DOC derived from autochthonous material in freshwater and marine ecosystems16, 17.
Unlike δ13C, which is a bulk property of DOM, lignin phenols are a unique biomarker for vascular plants and thus terrigenous plant material in aquatic ecosystems17, 18. Across the 11 watersheds, carbon-normalized lignin yields decreased dramatically with increasing watershed glacier area (Table 1; r2 = 0.91, P < 0.001). This finding indicates that the bulk of the DOM in glacially dominated rivers is not directly derived from higher plants. The presence of abundant, active microbial communities in supra19-, sub8- and pro20-glacial environments, combined with limited spectroscopic observations12 and the low C:N values11 of DOM in glacial streams, have previously led to the hypothesis that DOM in glacial runoff has a microbial source. Our findings provide direct corroboration of this hypothesis through three independent lines of evidence, indicating that increasing glacier contributions to streamflow shift riverine DOM sources from vascular plant material toward microbial biomass.
In order to better understand the fate and impact of glacier-derived microbial DOM on coastal ecosystems, we conducted short-term (2 week) bioavailability incubations of riverine DOC with inoculums of microorganisms from downstream near-shore marine ecosystems. DOM derived from microbially based ecosystems such as Antarctic dry lakes that lack inputs from higher plants is thought to be highly labile because of low C:N ratios and low aromatic carbon content21. We found that the lability of riverine DOC increased dramatically with glacier coverage (Table 1), and that the bioavailability of DOC was highly correlated with Δ14C-DOC values across the 11 watersheds (Fig. 2). In the most heavily glaciated watershed, Sheridan River, 66% of the riverine DOC was readily degraded by marine microbes despite having a Δ14C value of -386‰ (~3,900 yr Δ14C age). Heterotrophic microbes in both sub-glacial22 and pro-glacial20 environments have been shown to subsist on aged carbon overrun by ice during periods of glacier advance. It is additionally possible that CO2 respired from glacially sequestered carbon may support microbial primary production in glacial ecosystems. Along the GOA, the last major cycle of glacier retreat and re-advance occurred during the Hypsithermal warm period between 7,000 and 2,500 yr bp23. Our data indicate that microorganisms in GOA glacial ecosystems can liberate carbon fixed during the Hypsithermal as DOC in a form that is highly bioavailable to aquatic heterotrophs. Riverine DOC in the less glacial watersheds was substantially less bioavailable despite having modern, enriched Δ14C values consistent with inputs of recently fixed carbon from temperate forest soils. These findings contradict the prevalent view that the labile fraction of riverine DOC is dominated by young, lightly degraded material2, 3, 4. Moreover, the fact that glaciers can release ancient DOC has important implications for the interpretation of DOC age in rivers4 and coastal oceans24 affected by glacial runoff.
Figure 2: Relationship between Δ14C-DOC and bioavailable DOC for the 11 rivers sampled.
![Figure 2 : Relationship between |[Dgr]|14C-DOC and bioavailable DOC for the 11 rivers sampled. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.com](https://webarchive.library.unt.edu/web/20170201184834im_/http://www.nature.com/nature/journal/v462/n7276/images/nature08580-f2.0.jpg)
The percentage of riverine DOC that was readily bioavailable to marine microorganisms increased significantly as Δ14C-DOC values became more depleted (y = -7.2x + 108.4; r2 = 0.79; P = 0.001). Values of Δ14C are expressed as the deviation in per mil (‰) from the 14C activity of nineteenth century wood. Errors (±1 s.d.) associated with Δ14C AMS analyses averaged 2.0‰ (±20 yr for radiocarbon age) and error bars are smaller than the symbols. Error bars on bioavailable DOC are 1 s.e.
High resolution image and legend (75K)Download Power Point slide (456K)Slides may be downloaded for educational use, according to the terms described in Nature Publishing Group's licensing policy.
The impact of labile glacially derived DOM on heterotrophic production in GOA marine ecosystems depends on the magnitude of DOM fluxes from glaciers. A GIS (geographic information system) model of annual precipitation on the 75,300 km2 of glaciers that drain to the GOA was used in combination with previous estimates of distributed glacier volume loss across the region5 to calculate the annual water flux from GOA glaciers. We estimate specific discharge from GOA glaciers at 5.4 ± 0.6 m yr-1 (Table 2), which yields a total annual runoff of 410 ± 40 km3 yr-1 that is comparable to the annual discharge of the Mississippi River9. Based on previous measurements of DOC concentrations in glacial runoff12, 13, 14, we suggest that direct runoff from GOA glaciers produces a conservative DOC flux of 0.13 ± 0.01 Tg yr-1 to downstream ecosystems (Table 2). Furthermore, the DOC bioavailability value from the most heavily glaciated catchment (Sheridan; Table 1) suggests that ~0.1 Tg of the annual DOC derived from GOA glaciers is readily bioavailable. Because glacial streamwater turbidities are high and riverine transit times from glaciers to their estuaries are short across broad regions of the GOA, a substantial portion of this labile DOC is probably delivered to marine heterotrophic communities without biological or photochemical alteration. The biogeochemical significance of inputs of labile DOM discharged from glaciers is magnified by the relatively small volume of the GOA. On a per volume basis, the flux of labile glacier-derived DOM to the shelf waters of the GOA is approximately 25–50% of the total labile DOM input to the polar surface waters of the Arctic Ocean25, 26, which has the highest per volume inputs of DOM of any ocean on Earth27.
The quantitative importance of glacial ecosystems as sources of riverine DOM has been greatly underappreciated, as evidenced by the dearth of studies on DOM dynamics in glacial rivers12, 13. However, the annual area-weighted flux of DOC that we report for GOA glaciers (1,650 kg km-2 yr-1; Table 2) is higher than the average per area flux of DOC from the watersheds of the five largest rivers draining to the Arctic Ocean (1,600 kg km-2 yr-1)4. Thus, we conclude that glacial ecosystems, which cover approximately 15,000,000 km2 or ~10% of the Earth’s surface, have the potential for substantial DOM export in a form that is highly available to aquatic heterotrophs. Along the GOA, glacier wastage accounts for a substantial portion (~20%, Table 2) of the total riverine flux of water and DOM. As a result, fluxes of glacier-derived DOM to the GOA are likely to increase in the near term as ice thinning rates28 and discharge in glacial rivers29 continue to rise.
Globally, the rate of glacier ice loss from terrestrial storage is increasing, with important implications for sea level30 and the physical oceanography of coastal oceans10. Our findings indicate that future changes in discharge from glacial rivers may have previously unrecognized impacts on coastal biogeochemistry. Because bacterial production in coastal river plumes can be largely supported by terrigenous DOM31, our results suggest that changes in the magnitude and timing of glacial runoff to the ocean could alter carbon availability and heterotrophic productivity in marine ecosystems. These changes could be particularly pronounced in regions that currently support commercially important fisheries such as the GOA and the North Atlantic, the latter influenced by a 41% increase in Greenland ice sheet discharge between 1961–1990 and 1998–200332.
Extending our estimates of glacier-derived DOM export to other regions of the world will require sampling and characterization of DOM in additional glacial rivers. Moreover, the amount and quality of organic matter within and beneath other coastal glaciers may differ from that of the coastal temperate rainforests of Alaska. Even so, our results demonstrate that glacial ecosystems are quantitatively important and highly dynamic sources of reduced carbon to freshwater and marine ecosystems that receive their runoff. The existence of such large and highly labile carbon fluxes from GOA glaciers underscores the uncertainties in carbon budgets of coastal regions, which play critical roles in moderating the flux of reactive species to the ocean.
Methods Summary
Stream samples from 11 coastal watersheds on the Gulf of Alaska were collected over a three day period in late July 2008 and were filtered within one day of collection. Water samples were immediately analysed for bulk DOC concentration and DOM fluorescence. Protein fluorescence components were quantified from fluorescence excitation-emission matrices using the multivariate modelling technique parallel factor analysis (PARAFAC), which decomposes the fluorescence spectra of DOM into independent components. Streamwater DOM was also analysed for δ13C-DOC. Lyophilized DOM samples from each site were analysed for lignin phenols and Δ14C-DOC. To determine the bioavailability of DOC from the study streams, a laboratory incubation was conducted using near-shore marine water from the GOA as a microbial inoculum.
Fluxes of water from glaciers along the GOA were estimated using a combination of: (1) runoff derived from PRISM (parameter-elevation regression on independent slopes model) precipitation models that were calibrated with regional streamflow data and (2) runoff derived from the estimates of glacier wastage in ref. 5. Estimates of glacial runoff (1) and glacier wastage (2) were constrained using the regional glacier area from ref. 5 that drains to the GOA. Fluxes of DOC delivered from glaciers to downstream aquatic ecosystems were calculated using a conservative estimate of the average melt season concentration of DOC in glacial runoff in conjunction with the modelled water fluxes from glaciers along the GOA.
Full methods accompany this paper.