The base case
Because of the central significance of DOC to aquatic ecosystems, and indeed, to all ecosystems, several synthesis papers have examined carbon export from land to water and in-channel dynamics (e.g. Hope, Billet & Cresser, 1994; Mulholland, 1997; Findlay & Sinsabaugh, 1999; Sobek et al., 2007; Battin et al., 2008; Tank et al., 2010a). Almost without exception, these papers acknowledge the likelihood of strong human influences on OC dynamics in streams and rivers and emphasise the need to better understand these effects. But there is little subsequent coverage of the topic, in large part because consideration of human influences on OC has been a relatively recent pursuit. We begin by summarizing patterns and controls on DOC that emerge from these synthesis papers. Despite differences in sites, temporal resolution or metrics used, there is substantial agreement regarding the drivers of DOC in streams and rivers. We refer to this consensus view as the ‘base case’ and use it as a starting point for considering how human activities influence lotic DOC dynamics.
The controls on stream and river DOC can be viewed as a function of terrestrial accumulation, transfer to the channel and in-stream processing (Fig. 1). Perhaps with the exception of open-canopy streams, lotic DOC is dominated by terrestrial sources (Aitkenhead-Peterson, McDowell & Neff, 2003; Bertilsson & Jones, 2003), so accumulation of organic matter in the soil environment sets the first constraint on aquatic DOC. The next determinant is the capacity to move OC from terrestrial sources to the channel. Transfer is largely hydrologic, although atmospheric inputs (especially from streamside vegetation) can make important seasonal contributions and dominate particulate OC (POC) inputs (Webster & Meyer, 1997). In-stream leaching of terrestrial POC and gross primary production add DOC, although this latter source is often minor compared to terrestrial loading. Finally, in-stream processing by photooxidation and microbial respiration transform and remove DOC. Collectively, this formula of source, transfer and processing dictates both the quantity and quality of stream DOC loads.
Figure 1. Terrestrial accumulation, transfer and aquatic processing of lotic dissolved organic carbon under natural conditions (the base case). Soil/benthic storage and release includes sorption of dissolved organic carbon to particles, and release through leaching of particulate organic matter. Sorbed dissolved organic carbon may be respired or released by desorption. Both particulate and dissolved organic carbon may be transferred from terrestrial to aquatic systems.
Download figure to PowerPoint
The next step in examining stream DOC regimes is to consider the relative influence of the three base case processes. The major roles of terrestrial accumulation and hydrologic delivery on stream DOC are underscored by a wealth of studies relating landscape and climate attributes to aquatic concentrations or loads. Land cover provides a strong predictor of terrestrial accumulation, while climate variables are typically indicators of hydrologic connection between uplands and the channel. Such investigations have met with good success at local (e.g. Frost et al., 2006; Ågren et al., 2007), regional (e.g. Gorham et al., 1998; Gergel, Turner & Kratz, 1999) and continental to global scales (Mulholland, 1997; Aitkenhead-Peterson & McDowell, 2000), routinely explaining 50–80% of observed variance in DOC among sites. Predictors that frequently appear in such statistical models include wetland cover, topography, precipitation and soil type (reviewed by Hope et al., 1994; Mulholland, 2003), all of which can be related to the basic processes of terrestrial OC accumulation and transfer to the channel.
Once terrestrial DOC is delivered to the aquatic environment, its quantity and quality can be modified by microbial processing, respiration, sedimentation, adsorption/desorption, photobleaching and photooxidation. POC leaching and in situ primary production contribute new DOC, and the latter source differs from terrestrial material in its susceptibility to microbial and photochemical actions. The role of aquatic processing in changing the composition and size of the DOC pool has been a topic of growing interest and debate, with several lines of evidence indicating that the fraction of DOC subject to degradation is usually relatively small (<5–30%). Additionally, much of the biologically available pool is derived from aquatic algal primary production rather than terrigenous sources (e.g. Cole, Likens & Strayer, 1982; del Giorgio & Davis, 2003). More recalcitrant fractions are transported long distances (kilometres) before they are retained or mineralised by biological or physical processes (Worrall, Burt & Adamson, 2006; Kaplan et al., 2008). Thus, the two key constraints on the magnitude of aquatic processing are the overall lability of the DOC pool, and the time available for uptake or transformation of this material – that is, water residence time (Schindler et al., 1992). Given the recalcitrant nature of most DOC in streams (Thurman, 1985) and characteristically brief water residence times of fluvial systems, long DOC transport distances are likely to be the base case norm.
Effects of human land use
Human activities have a range of consequences on streams and catchments, but routinely involve changes in plant cover, catchment hydrology, soil attributes and nutrient inputs, especially in agricultural areas (Allan, 2004; Millenium Ecosystem Assessment [MEA], 2005). Agricultural extent and crop selection represent strong forcings on river DOC quantity and lability. From a global assessment based on data from 1992 (Leff, Ramankutty & Foley, 2004), major crop types ranked in order of decreasing coverage were wheat, maize, rice, barley, soybeans, pulses and cotton (Table 1). C : N ratios vary widely among these crop types, and in many cases, ratios are noticeably lower or higher than for native vegetation. Other basic differences in organic composition include the amount and form of lignins and tannins (Kögel-Knabner, 2002) and overall chemical diversity of crop sources relative to species-rich native plant communities. These are obvious and well-known contrasts in the material that is the primary input to the soil carbon pool, and then eventually, the aquatic pool.
Given the basic formula of terrestrial accumulation, hydrologic delivery to the channel and aquatic processing, it is reasonable to assume that human land uses strongly affect DOC loads in streams. Yet, while some studies have been able to detect a clear signal of land use on stream DOC, others have not. Agriculture (and other human land uses) has been associated with increased, decreased and undetectable changes in DOC (Table 2). As will be discussed below, these mixed responses are perhaps not surprising given the diversity of farming practices (and other land uses) and their affiliated effects on terrestrial and aquatic carbon cycling.
While the direction and magnitude of the agricultural signal on fluvial DOC may be ambiguous, divergence in the composition of the DOC pool between undisturbed and human-impacted catchments is emerging as a consistent observation across disparate locations and land uses. Changes include shifts from high- to low-molecular-weight DOC, increased redox state, reduced aromaticity and in general, increased lability. These differences have been attributed to altered terrestrial sources as well as greater in situ DOC production (Cronan, Piampiano & Patterson, 1999; Wilson & Xenopoulos, 2008; Petrone, Richards & Grierson, 2009).
In the following section, we return to the accumulation-transfer-processing framework to consider how human land use – particularly agricultural activities – affects terrestrial pools of OC (accumulation), the connections between land and water (transfer), and the production and fate of DOC in streams and rivers (processing) (Fig. 2). The intent of this overview is to highlight the range of changes that can influence DOC quantity and quality, in either opposing or reinforcing directions.
Figure 2. Major categories of anthropogenic influence on lotic dissolved organic carbon. Plowing and other forms of soil disturbance such as planting or animal stocking disrupt soil structure, increase susceptibility to erosion and influence turnover of soil organic carbon pools. Harvest practices include factors such as crop selection in farm lands or tree type in silviculture, and timing and method of harvest for crops or timber. Hydrologic modification includes changes in surface-groundwater connectivity and the timing and magnitude of runoff resulting from attributes such as extent of impervious cover, drainage ditches or soil disturbance in the basin or riparian zone. Engineering and diversion includes factors such as the density and size of dams and reservoirs, and characteristics of flood control or water supply infrastructure.
Download figure to PowerPoint
The base case highlights the significance of terrestrial OC stocks on stream DOC, and it is clear that human activities have substantial effects on terrestrial carbon pools. Agriculture, especially cropping, is associated with reduced terrestrial OC storage (Ogle, Breidt & Paustian, 2005; McLauchlan, 2006), so it is reasonable to hypothesise that soil OC (SOC) losses should have clear consequences for aquatic DOC (Sickman et al., 2010). Yet, investigating the aquatic consequences of this seemingly simple, directional change in terrestrial OC highlights a diverse array of processes that may cause increases, decreases or no net change in stream OC loads in agricultural areas.
Converting natural lands to row crop agriculture causes substantial loss of OC stored in soils because of increased erosion and decomposition rates associated with physical disruption and aeration by tilling or lowering of the water table (Guo & Gifford, 2002; Jarecki & Lal, 2003). In the U.S., much of the best and most productive agricultural land occurs in low-lying, low-relief regions that were once dominated by prairies and wetlands habitats that are characterised by substantial below-ground carbon stocks (Smith & Johnson, 2003; Bridgham et al., 2006). For example, Ohio, Indiana, Illinois, Iowa and Missouri have lost over 85% of their historic wetlands (Dahl, 1990) and 85–99% of the native prairie (Sampson & Knopf, 1994) due predominantly to agricultural conversion. Similarly, vast peatland areas were drained to increase farming and forestry production in the U.K. and elsewhere in northern Europe, causing substantial C losses from these environments (Holden, Chapman & Labadz, 2004; Armstrong et al., 2010).
SOC losses following agricultural conversion often continue for decades (McLauchlan, 2006), but the period of initial mobilization of DOC from mineral soil appears to be shorter, lasting anywhere from <2 to 10 years (Chantigny, 2003). Further, at least some of the eroded SOC may be redistributed and buried in other terrestrial locations, never reaching the aquatic environment (Van Oost et al., 2007). Nonetheless, the net long-term effect of agricultural conversion is a smaller terrestrial OC pool relative to native conditions, meaning that the potential supply of OC available for aquatic loading is reduced. Thus, it is often assumed that the historic loss of wetlands and SOC has resulted in lower contemporary DOC loads and concentrations in many agricultural streams (e.g. Royer & David, 2005; Dalzell, Filley & Harbor, 2007).
Declines in SOC pools associated with land conversion are accompanied by other agricultural practices that can complicate land use-stream DOC relationships. Modern changes in farming practices such as reduced plowing depth or no-till agriculture have been adopted to slow or even reverse soil and SOC losses (Smith, 2004; Ogle et al., 2005). Amendments of crop residues, organic fertilizers and manure disposal also add to the SOC pool. Because these additions are not fully integrated into the soil structure, they may be easily mobilized and cause both short-term and more sustained increases in stream DOC concentrations (Jardé, Gruau & Mansuy-Huault, 2007; Royer et al., 2007; Molinero & Burke, 2009). Thus, cases of undetectable changes in aquatic DOC because of land-use conversion may simply reflect a balance between losing one carbon source (wetlands or SOC) but gaining another (agricultural amendments). Initial evidence for such a cancellation effect is provided by similar concentrations of dissolved organic nitrogen in streams draining human-dominated (agriculture + urban) and undisturbed, wetland-rich catchments in Wisconsin, U.S.A. (Stanley & Maxted, 2008).
Despite variability in the magnitude and direction of agricultural effects on the quantity of stream DOC, these activities appear to be consistent in altering the composition of these pools. This is not surprising, given wholesale changes in the source of the terrestrial OC pool from native vegetation to crops and organic fertilizers. Shifts in chemical composition have also been described in areas subject to forestry (Amiotte-Suchet et al., 2007) and most conspicuously, in urban areas (Baker & Spencer, 2004; Aitkenhead-Peterson et al., 2009). Other novel additions to the terrestrial OC pool are synthetic compounds that include biocides, antibiotics and growth hormones along with residues of genetically modified crops that are now a part of modern farming practices. In all cases, these new terrestrial sources are now routinely detectable in agricultural streams (e.g. Pedersen, Soliman & Suffet, 2005; Jardéet al., 2007; Tank et al., 2010b), and possible consequences of these additions are discussed below.
Finally, conversion of riparian areas to agricultural land use may have larger than expected consequences, given riparian involvement in all facets of the stream DOC regime (as a terrestrial source, terrestrial-aquatic transfer and affecting in-stream processing and production). As a source area, DOC export from the riparian zone can be a major input to streams, especially during periods of high flow in high-relief regions (e.g. McGlynn & McDonnell, 2003; Bishop et al., 2004). Leaf litter can also represent a seasonally significant source of DOC in some forested headwater streams (e.g. McDowell & Fisher, 1976). In cases where buffer strips are not in agricultural production, the riparian plant community can still be substantially different from its original (native) composition, often dominated by invasive species (Tickner et al., 2001). These novel assemblages often differ in rates of litter production (Ellis, Crawford & Molles, 1998) and can affect loading of bioavailable DOC to streams (Wiegner & Tubal, 2010). Overall, we expect the quantity, form and timing of DOC transfer to aquatic ecosystems to change significantly following removal or modification of riparian habitats within a catchment.
Hydrologic modification is a hallmark of agricultural land use and includes altered rates of evapotranspiration and infiltration, installation of drains and ditches to remove excess water from soils or construction of storage ponds and irrigation systems to provide water to crops (Fig. 2; Scanlon et al., 2007; Gordon, Peterson & Bennett, 2008). Thus, flow paths that connect land to water have been re-shuffled or wholly reorganized in areas dominated by agriculture. Over the past 300 years, areas converted to pasture and rain-fed croplands have experienced large increases in discharge because of reduced terrestrial evapotranspiration (Scanlon et al., 2007). Irrigation-supported agriculture, which is rapidly expanding in global extent, has opposing effects on stream flow, routinely resulting in moderate to extreme declines in discharge (Döll, Fiedler & Zhang, 2009). Inevitably, major changes in how water moves from land to water will affect the strength, timing and type of connections that transport terrestrial OC to streams.
Currently, studies that specifically examine terrestrial-aquatic linkages and DOC inputs to streams in agricultural systems are limited. The best-studied examples we are aware of focus on tile-drained crop systems that are common throughout the Midwestern U.S. and many agricultural regions worldwide. Tile drain sites contain networks of buried drainage pipes that collect soil water, lower the water table and quickly route water to the channel. The results are flashier hydrographs and increased annual water export from recipient streams (Skaggs, Brevé & Gilliam, 1994; Blann et al., 2009). Floods in these systems are responsible for the majority of annual DOC export because of both increased discharge and increased DOC concentration during high flows (Dalzell, Filley & Harbor, 2005; Royer & David, 2005; Ruark, Brouder & Turco, 2009). Hence, as with water, DOC is rapidly routed from field to channel as a result of artificial drainage. Flood-dominance of inputs causes substantial intra-annual variance in stream water DOC concentrations (Stedmon et al., 2006; Dalzell et al., 2007). This represents a distinct departure from historical or undisturbed conditions, given that annual variance in stream DOC tends to be low and inputs of floodwater dilute, rather than enrich, the stream DOC pool in areas where wetlands persist (e.g. Hinton, Schiff & English, 1997; Gorham et al., 1998).
While extensive ditching and draining are widespread in low-topography mesic environments such as the U.S. Midwest, this represents just one of many hydrologic modifications in agricultural areas. Irrigation represents another equally heavy-handed and widespread modification, with 40% of the world’s food production coming from irrigated agriculture (Siebert et al., 2005). These water additions can increase SOC stocks in farm fields (Denef et al., 2008; Blanco-Canqui et al., 2010) as well as DOC concentrations in drainage water (Hernes et al., 2008; King et al., 2009). However, as with tile drain systems, studies investigating effects of irrigation on stream DOC are surprisingly scarce. In short, there is a substantial knowledge gap regarding the consequences of agricultural (and more broadly, anthropogenic) modification of flow paths that connect terrestrial and aquatic environments for inputs of DOC to streams and rivers. Yet, it is clear from the studies that do exist that this re-plumbing of catchments alters the timing, magnitude, amounts and composition of aquatic DOC delivery to streams and rivers.
Studies of organic matter processing in human-dominated lotic systems are sparse, as this topic is only now beginning to receive serious research attention. Understanding DOC processing in rivers is complicated by the diversity of molecular forms and the range of physical, chemical and biological factors that affect DOC production and removal from the aquatic pool. In this section, we consider three drivers affecting DOC processing: nutrient enrichment, changes in irradiance and altered sediment inputs. Each can be strongly modified by land-use practices such as farming and urbanization (Carpenter et al., 1998; Julian, Stanley & Doyle, 2008a; Hoffman et al., 2010) and also has known influences on stream DOC dynamics. The relative importance of these drivers is ultimately constrained by DOC quality and water residence time, which are also strongly affected by human activities (Fig. 2).
Human activities have caused a pervasive increase in the nitrogen and phosphorus content of surface waters (Carpenter et al., 1998; Smith & Schindler, 2009). Nutrient enrichment has long been known to lead to eutrophication, and greater autochthonous production should translate to greater inputs of relatively labile DOC (Bertilsson & Jones, 2003; Hilton et al., 2006). This prediction has been tested in nutrient-rich agricultural streams in Indiana (U.S.A.) during summer months when dense filamentous green algal mats develop (Royer & David, 2005; Warrner et al., 2009). As expected, DOC concentrations did in fact increase; however, there was no commensurate increase in DOC lability. Given that microbial respiration and organic matter degradation can also be enhanced by nutrient enrichment (Howarth & Fisher, 1976; Benstead et al., 2009), labile fractions might have been rapidly consumed, resulting in no detectable change in the composition of the bulk DOC pool. This example notwithstanding, the influence of nutrient enrichment on primary production versus respiration and the overall DOC balance in human-dominated streams represents yet another substantive knowledge gap.
Insolation to streams often increases in association with agricultural land conversion because of the removal of woody riparian vegetation (Julian et al., 2008b) and can change the quality and form of DOC by several mechanisms. However, the outcome of altered irradiance is likely to be difficult to predict because of confounding and offsetting processes. For example, benthic light availability can actually be lower in open-canopy agricultural streams because land-use conversion may increase the input of light-absorbing sediment (Julian et al., 2008a). Further, more solar radiation may increase photosynthetic activity and associated production of labile DOC, or conversely might reduce DOC stocks and/or change its quality via photobleaching and photooxidation (Bertilsson et al., 1999; Köhler et al., 2002). The effect of photodegradation on DOC quality also varies between algal and terrestrial carbon sources. Tranvik & Bertilsson (2001) found that humic DOC is predominantly degraded into more labile forms when exposed to UV, whereas more labile, algal-derived DOC becomes more recalcitrant over time. Clearly, changes in irradiance can influence aquatic DOC directly and indirectly through multiple pathways, but it remains to be determined as to how these various mechanisms actually do play out as a result of land-use change.
In addition to modifying benthic light availability, alterations to river sediment regimes through tillage (Tiessen et al., 2010) and grazing (Suren & Riis, 2010) have likely influenced DOC loads and processing in rivers by providing additional sources of, and sorption sites for DOC. Much of the sediment load contributed from cultivated areas is fine-grained (Walling & Amos, 1999), and this material can be highly effective in DOC adsorption. Sorption can occur irreversibly, creating a DOC sink (McKnight et al., 2002), or reversibly, representing a potential future source to both microbes and the water column (Riggsbee et al., 2008). Thus, as with light, predicting consequences of altered sediment regimes for DOC is far from straightforward, as the capacity exists for both increases and decreases in quantity and quality.
As noted in the discussion of the base case, water residence time plays a key role in constraining the degree of aquatic DOC processing, regardless of mechanism. Widespread re-engineering of river channels has altered the water residence time of river networks and thus changed the time available for different processes to influence the amount or form of DOC in fluvial systems. Most conspicuously, reservoir construction has increased the water residence time of runoff, and thereby increased the proportion of DOC loads metabolized by inland aquatic systems (Cole et al., 2007). The mean age of global continental runoff at river mouth has been extended by an average of 31–58 days, with a greater than twofold increase for North America, Europe, Asia, Africa and Australia/Oceana (Vörösmarty et al., 1997). Conversely, many un-impounded stream and river reaches have reduced water residence time because of the construction of canals and levees, and elimination of wetlands or floodplains which would otherwise slow water movement. Anthropogenic decreases in residence times are particularly pronounced in urban settings, where impervious surfaces result in rapid downstream routing of water by preventing infiltration into soils and ground water (Paul & Meyer, 2001). Similar hydrologic short-circuiting also occurs in agricultural areas with tile drains. Overall, streams are probably responsible for a low percentage of overall DOC uptake within surface water networks because of slow processing rates relative to water residence time (Köhler et al., 2002; Kaplan et al., 2008). However, uptake rates for specific simple dissolved organic compounds, including acetate (Johnson, Tank & Arango, 2009), urea and glutamic acid (Brookshire et al., 2005), are comparable to rates for inorganic nutrients (Ensign & Doyle, 2006). Given the shift towards more labile forms of DOC in agricultural streams, such high uptake rates may become more common.
We have focused the above discussion on a few specific DOC-processing mechanisms that are affected by human activities, but several additional factors can also influence DOC dynamics and are undoubtedly important in different situations. For example, carbon mineralisation is temperature-dependent (Gudasz et al., 2010), and altered thermal regimes are widespread among aquatic systems. Thermal pollution of rivers caused by warming of irrigation or urban runoff is a major environmental problem that can impact multiple trophic levels (Gibbons & Sharitz, 1974; McCullough, 1999). The consequences of climate-driven warming on carbon cycling are now receiving substantial attention, but more acute localised warming resulting from heated discharges and land-use change have rarely been considered in terms of effects on DOC metabolism. Other environmental drivers influencing DOC that are amplified by human land use include shifting redox conditions (e.g. in reservoirs; Bellanger et al., 2004) and increased salinization of soils and surface waters (Green, Machin & Cresser, 2008), among others.
This overview of terrestrial accumulation, transfer and aquatic processing underscores opportunities for wholesale changes at all points along the continuum from OC production to its delivery and consumption or export in the aquatic environment. Further, any one process may have opposing effects in different settings or at different times of the year. Certainly, this highlights substantial uncertainty, but also important opportunities for continued investigation. Land-use changes and human perturbation are clearly altering native DOC regimes in ways we are only now beginning to recognize. And undoubtedly, future changes in land use and management will reveal new influences on aquatic DOC cycling.