June 6, 2008

Andean Influences on the Biogeochemistry and Ecology of the Amazon River

By McClain, Michael E Naiman, Robert J

Although mountains often constitute only a small fraction of river basin area, they can supply the bulk of transported materials and exert strong regulatory controls on the ecological characteristics of river reaches and floodplains downstream. The Amazon River exemplifies this phenomenon. Its muddy waters and its expansive and highly productive white-water floodplains are largely the products of forces originating in distant Andean mountain ranges. The Amazon's character has been shaped by these influences for more than 10 million years, and its present form and host of diverse organisms are adapted to the annual and interannual cycles of Andean inputs. Although the Andes constitute only 13% of the Amazon River basin, they are the predominant source of sediments and mineral nutrients to the river's main stem, and Andean tributaries form productive corridors extending across the vast Amazonian lowlands. Many of the Amazon's most important fish species rely on the productivity of Andean tributaries and main-stem floodplains, and annual fish migrations distribute Andean-dependent energy and nutrient resources to adjacent lower-productivity aquatic systems. Mountain-lowland linkages are threatened, however, by expanding human activities in the Andean Amazon, with consequences that are eventually felt thousands of kilometers away. Keywords: Amazon, Andes, nutrient subsidies, land use, fisheries

The Amazon River exits the Andes mountains more than 4000 kilometers (km) from its estuary, but along its essentially flat and serpentine path through the lowlands of northern Brazil it maintains the character of an Andean river (figure 1). The indelible imprint of this distant mountain range on the main-stem channel of the world's largest river has been noted by naturalists and researchers for more than a century, but the multifaceted nature of Andean influences on the hydrology, biogeochemistry, and ecology of the river system have only come to light during the past two decades. Other fundamental but still obscure linkages remain to be discovered.

Long before scientists took interest in the study of Amazon environments-in fact, long before Europeans "discovered" the river- native peoples of the lowland Amazon recognized the unique characteristics of Andean tributaries. Agriculture thrived on the fertile floodplains of these muddy rivers and gave rise to some of the region's first and most successful chiefdoms (Meggars 1984). Native Amazonians also capitalized on the rich fish stocks of Andean tributaries. Alfred Russel Wallace (1853) was perhaps the first naturalist to write about the white-water, clear-water, and black- water river types of the Amazon basin and to relate the color of tributaries to the nature of their drainage basins (figure 2). Wallace astutely linked the sediment toad of white-water tributaries to erosion in their steep Andean headwaters, and identified clear- water rivers with the crystalline "mountains of Brazil" (the Guyana and Brazilian shields). He knew that black-water rivers emerged from lowland sources, and he correctly attributed their dark coloring to leaching of "decaying leaves, roots, and other vegetable matter" (Wallace 1853). Another naturalist of that time, Henry Bates (1863), marveled at the transport of volcanic pumice in the main-stem Amazon River and correctly assigned its origin to volcanic ranges thousands of kilometers away in the Ecuadorian Andes. He imagined these porous stones as vehicles transporting seeds and insect eggs downstream and thereby dispersing organisms far beyond their original ranges. Over the last 50 years, systematic investigations have further advanced scientists' understanding of the environment and distinct aquatic ecosystems of the lowland Amazon River (summarized in Sioli [1984], Junk [1997], and McClain et al. [2001]).

Steep terrain and young lithologies make the Andes an important source of sediments and solutes to the lower reaches of the Amazon River. The most visible characteristics of the main-stem Amazon and its Andean tributaries are high discharge and heavy loads of suspended and bedload sediment. Associated with this sediment load are abundant organic matter and nutrients. The ramifications of a high particulate load are also far-reaching in their geomorphological, biogeochemical, and ecological effects on the lowland river corridor. Large sediment loads and flooding have created broad floodplains, and associated nutrients support diverse and productive floodplain forests, macrophyte beds, and lakes of seasonal importance to the life cycles of organisms in the rivers and adjoining uplands, hi an important ecological feedback, the products of floodplain primary production eventually return to the main-stem river in floodplain runoff, becoming important energy sources for heterotrophic communities living there (Richey et al. 1990, Melack and Forsberg 2001). Many fish also migrate annually into Andean tributaries from low-fertility black-water and dear- water tributaries to spawn and feed in resource-rich white-water channels and floodplains. Upon their return, migrating fish transport organic matter and nutrients that subsidize the food webs of black-water and clear-water rivers,

Figure 1. Nine major rivers flow from the Andes to form fertile corridors across the lowland Amazon (shown in bold). The Ucayali, Maranon, and Napo rivers drain southern Ecuador and northern and central Peru, converging to form the mainstem Amazonas, which becomes the Solimoes River where it crosses into Brazil The Caqueta River flows from Colombia, becomes the Japura upon entering Brazil, and merges with the main stem at about 65[degrees]W (west). The Putumayo River flows from Colombia and Ecuador to become the Ica in Brazil The Madeira River collects the Andean tributaries flowing from southern Peru and Bolivia and traverses thousands of kilometers of lowland Amazon rainforest before merging with the main-stem Amazon at about 59[degrees]W.

Many fundamental aspects of the geomorphology, biogeochemistry, and ecology of the main-stem Amazon are therefore linked to the magnitude and variability of water and materials supplied from the Andes. In feet, the dominant downstream trend in biogeochemical and trophic characteristics of the main-stem Amazon and its large Andean tributaries is the progressive dilution of Andean contributions by lowland tributary inputs (Devol and Hedges 2001 ). Even though we are beginning to understand the dynamics of Andean-derived materials in the mainstem Amazon River corridor, and the degree to which lowland ecosystems depend on upstream inputs, we still know little about the nature and variability of processes that mobilize these materials from the Andes and modify them during downstream transport and storage in the extensive floodplains.

Figure 2. The main rivers of the Amazon have long been classified according to the color of their waters, which also reflects their source, (a) The Ica (Putumayo) River is a characteristic white- water river colored by the high loads of sediments transported from the Andes, (b) The Negro River is the largest of the black-water rivers, tinted by high levels of dissolved organic matter leached from low-tying areas of sandy soils, (c) The Rio Tapajos is the most notable of the dear-water rivers carrying low levels of sediments and organic matter from the crystalline Guyana and Brazilian shields. Photographs: Margi Moss (http://brasildasaguas.com.br).

In this article, we briefly introduce the geomorphology and ecological zones of Andean headwater regions of the Amazon, as these are poorly known even among scientists specializing in Amazon ecology. We then examine the multifaceted ways in which the main- stem Amazon River is influenced by-and depends on-Andean inputs. We conclude by exploring frontiers in research linking Andean and lowland parts of the Amazon, considering the possible impacts of increasing human-related development and climate change in the Andean Amazon.

The Andean Amazon

The Andes mountains rise steeply along the western margin of the Amazon basin and stand 3000 meters above sea level (masl) in elevation over much of their length (figure 1). Approximately half of the Andean Amazon lies at elevations between 500 and 2000 masl, while most of the remainder is between 2000 and 4000 masl; about 16% is above 4000 masl (table 1). The highest point in the Amazon basin is the Nevado de Huascaran in the Cordillera Blanca of Peru, at 6768 masl, but several other peaks extend above 6000 masl. Active volcanoes are prominent features of the Ecuadorian and Bolivian Andes. The eastern Cordillera of the Altiplano, a high-elevation endorheic basin containing Lake Titicaca, forms on one of the widest sections of the Andes, spanning nearly 300 km near the lake.

Characterization of the precipitation, soils, and vegetation of the Andean Amazon is fundamental to understanding Andean influences on the lower Amazon River (figure 3). Precipitation is greatest on the lower and mid slopes of the Cordillera (500 to 3000 masl) because of orographic controls on air masses coming from the east. The wettest parts of the basin lie in the eastern cordillera of Colombia and near the Peru-Bolivia border, where annual precipitation may exceed 4 meters (figure 3a). The most abundant soil order in the Andean Amazon is inceptisol (61%), a young, mineral-rich soil that occurs at mid elevations. More developed but less fertile ultisols occupy 16% of the region and occur mostly at lower elevations in Peru. Mollisols, or grassland soils, are the third most abundant soil order, covering 6% of the region, primarily near the Peru-Ecuador border and at higher elevations in southern Peru. Exposed rock is common at very high elevations (greater than 4000 masl) in southern Peru. Table 1. Elevation ranges of the Andean Amazon.

Figure 3. (a) Areas of higher precipitation are focused on the lower slopes of the Andes>> with maximal registered precipitation in the headwaters of the Madre de Dios River in southwest Peru and the Napo River of central Ecuador, (h) Montane forests dominate the land cover between 500 and 3000 meters above sea level and transition into natural high-elevation grasslands above. Compiled from Shuttle Radar Topography Mission 90-meter data and Global Land Cover 2000 data (CJRC 2000).

The major vegetative cover types in the Andean Amazon-mapped using Advanced Very High Resolution Radiometer satellite imagery (Eva et al. 1998)-are submontane (700 to 2000 masl) and montane (2000 to 3700 masl) forests, which together constitute approximately 42% of the region (figure 3b, table 2). Montane herbaceous vegetation interspersed with shrubland and agriculture is also widespread, covering nearly a quarter of the region. As of 2000, at least 40% of the region had been converted to human uses or fragmented by these uses (JRC 2000). The most intense human alteration has historically been at high elevations (> 3000 masl), where high levels of alteration continue today, but change is increasingly concentrated at mid and lower elevations as colonization continues and roads spread across the region (Mena et al. 2006).

The modern Amazon River is born in numerous Andean springs, but cartographers locate the most distant source of the river at 5300 masl on the northern slope of Nevado Mismi. From this stream, the Carhuasanta, the main stem of the Amazon, changes names at least nine times: from Carhuasanta to Lloqueta, Hornillos, Apurimac, Ene, Tambo, Ucayali, Amazonas, Solimoes, and finally Amazon below the confluence of the Solimoes and Negro rivers. The entire north-south length of the Andean Amazon basin is drained by eight major rivers- the Caqueta, Putumayo, Napo, Maranon, Ucayali, Madre de Dios, Beni, and Mamore (figure 1).

Table 2. Land cover of the Andean Amazon basin.

Andean influences on the load of the Amazon main stem

The main-stem Amazon River integrates the flow of subbasins containing distinct combinations of geology, soils, and vegetation. There are four major Andean tributaries to the main-stem Amazon River: the Solimoes, Ica, Japura, and Madeira (figure 1 ). (Andean tributaries to the main stem are denned as those with headwaters above 500 masl in the Andes mountains, assuming that the western limit of the main-stem Amazon River is set as the Brazil-Colombia border.) Where they intersect with the main stem, the combined mean annual flow of these white-water tributaries is approximately 90,000 cubic meters per second: roughly half of the main-stem Amazon River's mean annual discharge, or five times the flow of the Mississippi River (Dunne et al. 1998).

Figure 4. The disproportionate loads of sediments carried by the main Andean tributaries are evident when comparing the inflows of (a) water and (b) sediments to the main-stem Amazon river from its major tributaries. Inputs at the top of each diagram represent the contributions of the Amazonas/Sotimoes River flowing from Peru. Data were compiled by R. H. Meadefrom water-discharge data listed by Carvalho and da Cunha (1998) and from the sediment-discharge data of Dunne and colleagues (1998).

The Andes cover only about 13% of the Amazon basin upstream of Obidos, and Andean tributaries may flow through hundreds to thousands of kilometers of lowlands (below 500 masl} before connecting with the main stem. Yet most measurements of "Andean" contributions to the main-stem Amazon have been made at the main- stem confluences of the four Andean tributaries. Clearly these rivers have accumulated water, participates, and solutes from the lowlands before reaching the main stem, and therefore one must be careful to consider what part of these loads actually derived from the Andes rather than from the lowlands. In the case of water, we noted that the combined flow of the Andean tributaries amounts to approximately half of mainstem flow, but the volume of water actually originating in the Andes is probably roughly proportional to the areal coverage of the Andes. Although annual precipitation on the lower slopes of the Andes exceeds the Amazon average, higher valleys of the Andes are more arid, and thus the average precipitation for the entire range is not likely to be greatly different from precipitation for the basin as a whole. But while Andean contributions of water to the main-stem Amazon may be proportional to area, contributions of sediments and solutes are disproportionately greater. Moreover, energy and nutrients carried from the Andes by the river appear to largely drive main-stem productivity, both directly and indirectly through biophysical feedbacks with the massive lowland floodplain.

Inorganic sediments and solutes

Four decades ago, Ronald J. Gibbs wrote that "the Andean mountainous environment controls the geochemistry of the Amazon River" (Gibbs 1967). He had sampled the Amazon main stem and 16 of its major tributaries and had compared total particulate and solute concentration data for the wet and dry seasons against nine environmental parameters. On the basis of strong correlations with the environmental parameter "mean relief," Gibbs concluded that the Andes were the source of 82% of the total suspended solids exported by the Amazon River. The importance of Andean sources of suspended sediment to the main-stem Amazon River was reaffirmed by the subsequent work of Robert Meade and others, who concluded that between 90% and 95% of the suspended sediment load of the main stem derived from the Andean tributaries (figure 4; Meade 1984, Meade et al. 1985).

Returning to the question of how much of the water and suspended particles carried by the Amazon River originate from the Andes mountains, we speculated that less than a quarter of the water originates in the Andes but that most suspended sediments could originate in mountain areas. Loads of suspended and bed sediments measured along the entire length of the Madeira River, from its Andean headwaters to its confluence with the main stem, show a sharp decrease in sediment load (as much as 60%) at the base of the Andes, a decrease in the mean diameter of suspended particles in the piedmont region, and a progressive decrease in the mean diameter of bed sediments (Guyot et al. 1999)-all indicators of a declining energetic capacity to transport materials. These characteristics indicate that Andean rivers supply more than enough sediment to account for the total load of sediments in the lowland sections of the Andean tributaries. Conclusive evidence of an Andean source is found in the mineralogical and isotopic composition of the suspended sediments. The mineral composition of sediments in the main-stem Amazon correlates well with that of the Ucayali and Maranon rivers in the Andes (Gibbs 1967). Measurements of neodymium, strontium, and lead isotopic ratios reaffirm that Andean sources account for an overwhelming proportion of the main-stem sediment load (Allegre et al. 1996).

Andean-derived suspended sediments bring a large flux of minerals into the main-stem Amazon River, but they also bring other elements and materials. Andean tributaries deliver an order of magnitude more particulate nitrogen (1170 megagrams [Mg] per year) and phosphorus (806 Mg per year) to the main stem than their lowland counterparts (119 and 43 Mg per year, respectively; Richey and Victoria 1993). Most particulate nitrogen is likely to be organic, whereas phosphorus is mainly phosphate strongly adsorbed to iron and aluminum oxide surfaces (Berner and Rao 1994). The availability of this phosphorus to main-stem organisms is not known, but significant amounts of phosphorus are released from Amazon sediments upon entering the estuary and may be available to organisms on the floodplains (Melack and Forsberg 2001). The question of whether particulate nitrogen and phosphorus actually derive from the Andes or from some intermediate river section is tied to the origin of the fractions with which they are associated. The tendency of phosphate to adsorb to mineral surfaces links this nutrient to the Andean sources of the mineral sediment, but the organic association of nitrogen is tied to that of the particulate organic fraction, which is less well understood.

Two features of the Andes enhance their importance to the solute geochemistry of the Amazon River and to its ecological characteristics. First, the Andes contain the only significant deposits of evaporites and carbonates in the Amazon basin (StaUard and Edmond 1983). High fluxes of Ca^sup 2+^ (calcium), Mg2+ (magnesium), HCO^sub 3^^sup -^ (bicarbonate), and SO^sub 4^^sup 2-^ (sulfate) ions occur in rivers draining carbonate deposits, and high fluxes of Na+ (sodium) and Cl- (chloride) ions occur in rivers draining evaporite deposits. Rivers draining basins containing carbonates generally have total cation charges of 450 to 3000 microequivalents ([mu]eq) per Kter (L), and rivers draining basins containing evaporites may have total cation charges of greater than 70,000 [mu]eq per L near the salt sources (Stallard and Edmond 1983). The rich mineral content of Andean tributaries underpins the ecological productivity of downstream reaches. Black-water and clearwater tributaries draining lowland portions of the basin, by contrast, have total cation charges below 300 [mu]eq per L and are characteristically considered to have low ecosystem-scale productivity. The second distinguishing feature of the Andes is the intensity of its weathering regime, which increases the concentration of ions in solution. Among the Amazon tributaries that drain basins dominated by less-weatherable silicate rocks, Andean rivers have consistently higher total cation concentrations (Stallard and Edmond 1983). Few data exist that would allow us to estimate the proportional contribution of major ion fluxes to the main stem from the Andes. Robert Stallard's work demonstrates that solute concentrations are elevated in Andean rivers, but without measurements of discharge it is not possible to calculate fluxes. Furthermore, one-time flux measurements are not representative of annual or interannual contributions to the main stem. Unfortunately, no suitable data exist for Colombian, Ecuadorian, or Peruvian Andean tributaries, and thus no estimation can be made regarding the Andean contribution of major ions to flow in the Solimoes River from these countries. We may speculate, however, on the basis of the high ion concentrations in Andean rivers, that the Andean contribution to the main-stem solute load is dominant, especially for certain elements found preferentially in Andean lithologies. For the headwaters of the Madeira River in Bolivia, Andean fluxes can be estimated with some confidence, thanks to a 10-year data set (Guyot et al. 1992). Over the period of these data, the specific flux of total dissolved solids from Andean basins was 80 Mg per km^sup 2^ per year, while the specific flux from lowland Bolivian basins was 7 Mg per km^sup 2^ per year. The headwaters of the Madeira River contain few carbonate and evaporite deposits in comparison with the headwaters of the Solimoes River in Peru. Thus it is likely that the Peruvian Ancles contribute an even larger percentage of the major ions delivered to the main stem.

Organic matter

Andean-derived suspended sediments carry a significant amount of organic matter, 90% of which is made up of particles less than 63 micrometers ([mu]m) in diameter (Richey et al. 1990). Variations in the fluxes of fine particulate organic carbon (FPOC; particles 90%) cannot be physically separated from mineral material and is therefore probably physically bound to it (Keil et al. 1997). This physical association has been shown to reduce the rate of organic matter decomposition and enhance its preservation. Total organic carbon is approximately 1%, by mass, of suspended sediment in the main stem, constituting a flux of 5 to 14 teragrams (Tg) of carbon per year to the Atlantic Ocean (Richey et al. 1990).

Measurements show that more than 90% of particulate organic carbon (POC; > 0.5 [mu]m) in the main-stem Amazon River comes from Andean tributaries, but how much actually originates in the Andes Mountains? POC behaves more or less conservatively in the main stem, suggesting that it resists decay and is derived from distant sources (Richey et al. 1990). Just how refractory and how distant the sources are can be estimated from a suite of molecular, elemental, and isotopic techniques used to characterize the organic matter and to trace it back to its sources (Hedges et al. 1986,2000, Aufdenkampe et al. 2007). Concentrations of total lignin-derived phenols, carbon-to-nitrogen ratios, and stable carbon isotope ratios point to terrestrial plants, and more specifically the leaves of terrestrial plants, as the main source of main-stem organic matter. Algae and aquatic plants, so abundant on the extensive Amazonian floodplain, are important sources of labile organic matter, fueling microbial metabolism in the main stem, but do not persist in the system (Richey et al. 1990). The depletion of carbohydrates and the increasing abundances of nonprotein amino acids and diagnostic lignin-derived phenols confirm that the organic matter is highly degraded, especially the FPOC fraction. Moreover, these characteristic signatures extend up the Madeira and Solimoes rivers and into the Andean foothills (Hedges et al. 2000, Aufdenkampe et al. 2007). Richey and colleagues (2002) estimated that the main- stem Amazon River transports only 7% of the organic matter supplied to the river basinwide, supporting the finding that it also transports the most degraded and recalcitrant materials.

The isotopic data, however, provide the most definitive information on the age and general source area of particulate organic matter in the main stem and its Andean tributaries. For main- stem FPOC to have a true Andean source, much of it would have to be hundreds to thousands of years old. This is because little main- stem FPOC (and little of the fine sediment with which it is associated) is transported directly from the Andes; most is stored for varying periods of time in pouit-bar and floodplain sediments (Dunne et al. 1998). FPOC does, in fact, have the lowest levels of bomb carbon14 (^sup 14^C) of any organic matter fraction in the main- stem Amazon (+19 Delta^sup 14^C per thousand [%o])*, uggesting an average turnover time of hundreds of years (Hedges et al. 1986). Allowing for the dilution of the bomb ^sup 14^C signal by younger organic matter, this implies that a significant portion of main- stem FPOM may be Andean.

The actual proportion of FPOC of Andean origin has been approximated using delta carbon-13 (delta^sup 13^C) stable isotopic ratios as a "fingerprint" of its origin. The delta^sup 13^C of plant leaves is positively correlated with elevation, and ratios in the Peruvian Andes have been found to range from about -30[per thousand] at 1000 to 2000 masl to -26[per thousand] at 4000 masl (TownsendSmall et al. 2005,2007). The values of leaves from prominent floodplain and upland forest trees along the main-stem river also average -30[per thousand], indicating that there is no clear isotopic separation of leaf delta^sup 13^C between lowland forests and Andean forests below 2000 masl of elevation (approximately 50% of the Andean Amazon area; table 1 ). Unlike plant leaves, however, there is a dear separation of FPOC delta^sup 13^C between Andean and lowland rivers, and this separation can be used to estimate the relative proportion of each in the main stem. FPOC in purely lowland rivers has delta^sup 13^C values consistently near -28.5[per thousand] (Quay et al. 1992). The delta^sup 13^C of FPOC discharged in the main-stem Amazon River at Obidos is - 27.4[per thousand] and thus indicates a mixture of the Andean and lowland sources. If the Peruvian value for delta^sup 13^C of FPOC exiting the Andes (approximately -26.5[per thousand]) is taken as the Andean end member and -28.5%o is taken as the lowland end member, FPOC at Obidos is a mixture of 50% Andean FPOC and 50% lowland FPOC. Alternatively, if the Bolivian end member of-25.5[per thousand] is used, FPOC at Obidos is a mixture of 33% Andean and 67% lowland FPOC (Quay et al. 1992, Hedges et al. 2000).

Interestingly, the delta^sup 13^ of FPOC in each of the major. Andean tributaries (the Solimoes and Madeira rivers) where they meet the main stem is -26.8[per thousand]. This suggests that these rivers carry FPOC that is largely of Andean origin and account for 82% of the FPOC input to the main stem. If only 30% to 50% of FPOC entering the Atlantic Ocean is of Andean origin, then there is a 50% to 70% reduction in Andean-derived FPOC in the main-stem section of the river. This reduction probably occurs through sediment exchange with the floodplain and gradual decomposition of Andean organic matter while in storage. Recent research using a dual-isotope approach (^sup 14^C and ^sup 13^C) estimated the degree of mineralization of Andean-derived FPOC with transport downstream and concluded that nearly all Andean FPOC was mineralized in the river and floodplain system (Mayorga et al. 2005). Taken together, the Andes largely regulate the particulate load to the main-stem Amazon River, not simply with respect to its particulate mineral load but also with respect to associated nutrients and organic matter.

The four major Andean tributaries contribute approximately 50% of the dissolved organic matter (DOM) input to the main stem (Richey et al. 1990), but unlike particulate organic matter, this DOM appears to derive largely from lowland sources. Neither mass-balance nor chemical-tracer approaches support important Andean contributions of DOM to the lowland or main-stem Amazon. DOM accumulates in swampy environments that are common throughout the lowland Amazon, and in rivers and streams that drain areas of spodosol soils (McClain and Richey 1996). In the central Brazilian Amazon, fluxes of DOM to groundwater in the spodosols characteristic of the Rio Negro subbasin are approximately 20 times greater than those in the oxisols characteristic of much of the rest of the lowland Amazon (McClain et al. 1997). In the Rio Negro basin, high groundwater DOM concentrations (approximately 3000 micromoles of carbon) also appear in surface water draining spodosols, whereas in oxisol terrains, fringing wetlands appear to be important sources of DOM. DOM concentrations are uniformly low in the few studies on Andean rivers (Guyot and Wasson 1994, Hedges et al. 2000, Saunders et al. 2006). In the Madeira subbasin, there is a distinct increase in DOC concentrations in rivers below 500 masl, and this additional DOC appears to derive from floodplains and wetlands such as those of the Bolivian Llanos de Mojos (Guyot and Wasson 1994).

Andean influences on the productivity of the main-stem Amazon

The productivity of the main-stem Amazon is tied to the productivity of its fioodplain, a system built of Andeanderived materials and fueled by mineral nutrients from the Andes (Melack and Forsberg 2001). Over a 2010-km reach of the Amazon main stem, the mean lateral flux of sediments ( 1570 to 2070 Tg per year) between the channel and adjoining fioodplain exceeds the downstream flux (1200 Tg per year), and approximately 500 Tg per year of upstream- derived sediment and associated nutrients accumulate on the floodplain and in channel bars (Dunne et al. 1998). This process builds the fertile floodpiain soils along Andean tributaries and the main stem. By contrast, floodplains along non-Andean, lowland tributaries are far more depleted in mineral nutrients. The Amazon River maintains year-round lateral exchanges with its floodpiain, and especially with its abundant lakes. The floodpiain is a highly productive system, with an estimated regional net production of 113 Tg of carbon per year occurring over an area of 67,900 km^sup 2^, from the Brazilian-Colombian border to near the river's mouth (figure 5; Melack and Forsberg 2001). This translates to 17 Mg carbon per hectare per year, which exceeds the productivity of upland Amazon forests by a factor of five; in fact, the Amazonian floodpiain is among the most productive ecosystems on Earth. The majority of primary productivity is attributed to macrophyte (65%) and floodplain forest (28%) communities. Subtracting estimates of carbon loss to respiration and burial, about 90 Tg carbon per year are available for export to the mainstem river, where the additional carbon fuels respiration (Melack and Forsberg 2001, Mayorga et al. 2005). Figures. Nutrients and mineral substrates carried by Andean tributaries and deposited on fioodplains fuel the highest primary productivity rates per hectare in the Amazon basin. This schematic illustrates the balance of organic carbon on the main-stem Amazon floodplain between 70.5[degrees]W (west) and 52.5[degrees]W (refer to figure 1 for extent). This balance indicates that large quantities (approximately 90 teragrams) of organic matter are returned to the river channel annually to fuel in-channel respiration. All quantities are for total organic carbon unless otherwise noted. Source: Melack and forsberg (2001) and Richey and colleagues (1990). Abbreviations: DOC, dissolved organic carbon; POC, paniculate organic carbon.

A portion of the supply of Andean nutrients to the floodplain can eventually be traced back into the main stem not only as labile organic matter but as part of myriad organisms that move between the floodplain and channel. Large numbers of fish move onto the floodplain annually to exploit its productivity and utilize its habitats (Goulding 1993). In fact, annual movements onto the floodplains of Andean-influenced white-water rivers are the most common form of migration among Amazon fishes and are critical to maintaining the region's fisheries (Goulding et al. 1997). Of the 24 species in the Brazilian Amazon that are most important to humans (in nutritional and economic terms), most migrate as part of their life cycle, and most rely to some extent on the resources delivered from the Andes (Araujo-Lima and Ruffino 2004). One of the most sought-after fish is the tambaqui (Colossoma macropomum), This omnivorous/frugivorous fish occurs over the length of white-water rivers but only in the lower reaches of black-water rivers. It feeds in flooded forests during high water and migrates back into the channel during low water. Tambaqui, like many other species, spawns along the margin of white-water rivers, and the larvae are washed onto floodplains by the rising waters. There they feed and seek shelter beneath the ubiquitous macrophyte beds (Araujo-Lima and Goulding 1997). A number of other characids important to Amazon fisheries (Brycon spp., Mylossoma spp., Myleus spp.) also follow this migration pattern (Araujo-Lima and Ruffino 2004), using the floodplain for feeding and nursery habitats and for transporting resources back to the river as they migrate. Isotopic tracers have shown that C3 macrophytes, floodplain trees, and phytoplankton account for 82% to 97% of the carbon in 35 species of adult fishes examined (Forsberg et al. 1993). Phytoplankton, while accounting for a small proportion of the total primary productivity on floodplains, represents the primary source of carbon to characiform fishes (AraujoLimaetal. 1986).

Migrations are also important in distributing the enhanced productivity of Andean-influenced white-water rivers and their floodplains to less productive black-water and clear-water environments. Many Amazon fish migrate from black-water and clear- water rivers to the main stem and other whitewater rivers to spawn. In fact, all commercially important species appear to spawn only in white waters (Goulding et al. 1997). During times of the year other than the spawning season, some move back into black-water and clear- water environments, and in the event of predation or death, the organic matter and nutrients of their bodies serve as subsidies to these less productive ecosystems. Jaraqui (Semaprochilodus spp.) is an example of a fish that migrates from black-water rivers into white-water rivers to spawn (figure 6a). These predictable migration routes are stalked by larger predators that congregate at the confluences of black-water and whitewater rivers, such as the Amazon River dolphin, or boto (Inia geoffrensis).

Many other fish use the main stem and its Andean tributaries as migration corridors, most notably large predatory catfish (Pimelodidae) moving upriver to Andean spawning areas. Catfish making long-distance migrations are quantitatively the most important predators in the river system, and they are also the most important species to fisheries along the river's length (Barthem and Goulding 1997). The most remarkable of these migrations is that of the dorado, or dourada, catfish (Brachyplatystoma spp.; figure 6b), which travels as far as 5000 km in one direction (Goulding et al. 2003). Statistical data on size classes along the entire length of the Amazon River reveal that dorado spawn in headwater regions (including Andean foothills) and that the young are washed downstream to nursery areas in the Amazon estuary (Barthem and Goulding 1997). Preadult dorado move upriver again, completing the approximately 8000-km migration over several years. Dorado and a number of other migrating catfish are heavily fished along the river, so their numbers are significantly reduced by the time they reach the rivers of the piedmont and Andean foothills.

In Andean piedmont regions, characins emerge as the most important fishery species in biomass; the most important among these is Prochilodus nigricans, known as boquichico in Peru. Boquichico is a fine-particle feeder that ingests detritus and algae, and has a maximum length of less than 40 centimeters. During the low-water season, it lives in floodplain lakes and channels of the Amazon piedmont, but at the initiation of rising water it leaves the floodplain and migrates en masse up Andean tributaries to spawn (Diaz-Sarmiento and Alvarez-Leon 2004). Collectively, the fish migrations illustrate the critical connections between the Andes and downstream biotic communities and ecological processes, as well as the importance of maintaining both lateral and longitudinal connectivity throughout the Amazon.

Enormous sediment loads, fluxes of nutrients and refractory organic matter, and ultimately the fertility of the expansive floodplains reflect the many influences of distant Andean mountain ranges on the main-stem Amazon and other white-water tributaries (figure 7). The river's character has been shaped by these materials for more than 10 million years, and its present form and host of diverse organisms are adapted to the annual and interannual cycles of Andean inputs. It is safe to say that the ecology of the modern Amazon main stem has been built on substrates and nutrients derived from the Andes, and that the decoupling of the mainstem Amazon from its mountain headwaters would lead to dramatic changes in the river- a pattern reflected in many of the world's other great rivers.

Figure 6. Migrations of many Amazon fish are strongly influenced by the pursuit of resources and habitats tied to Andean tributaries, (a) Thejaraqui fSemaprochilodus insignis) is an example of species that, as adults, live mostly in black-water rivers or lakes, but migrate to white-water rivers to spawn. Juvenile jaraqui also use white-water floodplains as their nurseries, (b) The dourada (Portuguese) or dorado (Spanish) catfish (Brachyplatystoma 5pp.; B. rousseauxii in photo) are the farthest-migrating species known in the Amazon. They hatch in the Andean foothills, use the Amazon estuary as their nursery, and then migrate thousands of kilometers up Andean tributaries to spawn. Photographs: Michael Goulding.

Andean processes regulating fluxes to lowlands: A research frontier

The Andes exert strong influences on the main-stem Amazon, and these influences strengthen as one travels upstream along the major Andean tributaries. But what processes regulate the fluxes of Andean derived materials, and how do these processes vary spatially and temporally in the Andean Amazon? Unfortunately, little research to date addresses these questions, and obtaining regional numbers is exceedingly difficult. Nevertheless, current rates of land-use change in the Andean Amazon are among the highest in the Amazon basin; 40% or more of the region already has been significantly fragmented and otherwise affected by human alterations (EvaetaL 1998). How will land-use change and possible flow regulation alter fluxes of participates and solutes to the low-land Amazon, and what other forms of contamination might be emitted by growing mountain populations? Research addressing these human-related questions is still relatively restricted spatially in the Andean Amazon, but such research is essential for the coming decade if effective regional agreements are to be forged about the future of the Amazon basin. Figure 7. Andean influences on the ecology and btogeochemistry of the Amazon may be grouped into three interacting sets ofprocesses. Andean exports of water, sediment nutrients, and organic and biological material exert fundamental control and produce the white- water characteristics of Andean tributaries and the mainsteam Amazon itself. Floodplain building by these Andean-derived materials provides the substrate and nutrition fueling productive flooplain forests, macrophyte beds, and lakes. Fish migrate throughout these systems and along tributaries, capitalizing on the productivity of white-water river systems and transferring a small quantity of Andean-derived energy and nutrients to nutrient-poor black-water and clear-water systems.

Concerning sediment fluxes, it is important to note that instantaneous loads in lowland rivers are largely decoupled from those in mountain rivers. Where lowland Andean tributaries remain "white" with high sediment loads year-round, mountain rivers are generally dear during the dry season and white only during storm- runoff events (Townsend-Small et al. 2008). Their sediment fluxes may fluctuate greatly on daily or weekly timescales in response to individual storm and landslide events (Guyot et aL 1999), whereas lowland river fluxes, like their hydrographe, fluctuate according to dampened seasonal cycles. Meandering lowland rivers maintain their sediment loads by continually resuspending and depositing materials within their channels (Meade et al. 1985, Dunne et al. 1998), effectively mining sediments accumulated in the piedmont over long timescales through discrete depositional events (Aalto et al. 2003). To understand mountain-lowland linkages, one therefore needs to consider erosional processes over a broad range of timescales.

At timescales stretching into millions of years, and at the spatial scale of the entire mountain range, climate seems to exert a fundamental control on erosion processes in the Andean Amazon. Montgomery and colleagues (2001) analyzed the topographic, climatic, and tectonic variability of the entire Andes cordillera and concluded that morphology is more closely related to climate than to tectonic processes. Erosion from the mountain range over the past 25 million years has come predominantly from the northern Amazon Andes (north of 15[degrees] south), where historical rates of erosion are up to twice as high as in the drier southern portion of the Amazon Andes (southern Peru and Bolivia). Linked to this long-term erosional history, a striking and relevant geomorphological characteristic of the high Andes is a shift from steep-sided, V- shaped valleys to gently sloped, U-shaped valleys between 3000 and 3500 masl. Although much reduced in size today, glaciers have been important in shaping high Andean valleys. Moreover, the gentle valley slopes exposed by glacial retreat result in reduced physical erosion in the highest portions of the Andes.

At subregional spatial scales and shorter timescales, vegetation may assume a first-order control of erosion rates. Erosion rates in the Beni and Mamore river basins of Bolivia range from 521 to 6000 metric tons per km^sup 2^ per year and from 310 to 2600 metric tons per km^sup 2^ per year, respectively (Guyot et al. 1988). Topography, lithology, rainfall, and vegetation all play roles in explaining differences in erosion between basins, but vegetation plays the dominant role. Rates of erosion are greatest in the southernmost basins, where vegetation is sparse. In the north, where rainfall is greater but subbasins are heavily forested, erosion rates are considerably lower.

The controlling influence of vegetation on erosion at both subregional and hillslope scales is significant because land-use change is the most prolific form of anthropogenic disturbance in the Amazon (figure 8). Erosion is less intense in densely vegetated parts of the Andes, despite high rainfall on erosionprone slopes. The stabilizing effects of natural vegetation are lost, however, following deforestation, and land management practices become important variables in explaining fluxes of sediments, organic matter, and nutrients from newly created agricultural fields and pastures. Studies conducted in midelevation (2000 to 2500 masl) valleys of the Peruvian Amazon find increased fluxes of sediments, organic matter, and nutrients in rivers draining valleys with greater proportions of agriculture and pastures (Waggoner 2006). Similar trends have been observed in the Napo River basin of Ecuador, where clear correlations were found between overall river health and the level of anthropogenic alterations (Celi 2005). Continued investigations of land-use impacts on stream and river sediment loads are one of the most pressing research needs in the Andean Amazon today. Studies of land-use impacts on rivers and streams should emphasize riparian zones, both because they are control points for land-to-river material transfers (Naiman and Decamps 1997, Naiman et al. 2005) and because they are favored for agriculture in the Andean Amazon as a result of the relative fertility of their soils (McClain and Cossio 2003).

It was recognized early on that concentrations of major ions and trace elements in Andean Amazon rivers were linked to the lithologies of the major subbasins, and subsequent work has supported this link (Sobieraj et aL 2002). The most focused impacts that humans have on major ions and trace-element fluxes from the Andes is through mining, which is widespread at higher elevations. Contamination of soils and vegetation by heavy metals has been documented near mines and downstream of mining operations (Hudson- Edwards et al. 2001). Accumulations of metals in river invertebrates have even been measured downstream of the point at which contamination of bottom sediments is no longer detectable (Bervoets et al. 1998). Mercury contamination from placer gold-mining operations is a significant concern in many Amazonian areas, and mercury accumulations in fish and in the hair of riverine people have been linked to gold-mining operations as far as 150 km upstream in the upper Beni subbasin of Bolivia (Maurice-Bourgain et al. 1999). Although of considerable local concern, the current impacts from mining appear to be limited to river reaches immediately downstream of mining sites. Expansion of mining activities, however, may eventually lead to significant changes in the fluxes of heavy and trace metals to adjoining Amazon lowlands. Quantifying the composition, magnitude, and ecological consequences of increased heavy metal fluxes is an important need in the Andean Amazon.

The dependence of lowland river corridors on sediments and nutrients derived from the Andes requires unobstructed connectivity between the two regions. No major Andean tributary to the Amazon is currently dammed, although Brazil is pursuing plans to build two major dams on the Madeira River. Hydroelectric installations are common, however, on streams and small rivers close to major mining operations, to urban areas, or to other significant human settlements. Peru has five significant hydroelectric projects under way in its Amazon region, and the Peruvian Ministry of Energy and Mines has identified dozens more potential dam sites, some on prominent rivers such as the Maranon, Huallaga, Tambo, and Urubamba. Dams trap large volumes of sediment, and could cause major readjustments over the long term in the geomorphology of downstream river sections and the eventual sediment starvation of some downstream reaches. The iD, effects of dams on river organisms and riparian environments are well known (e.g., Dudgeon et al. 2006) and could be especially destructive in the Andean Amazon, where biodiversity is high and many fish species migrate annually between mountains and the lowland rivers and floodplains. Far too little is known at this point about the extent to which riverine organisms and riparian environments rely on open linkages between mountains and adjacent lowlands in the western Amazon. It is therefore impossible to predict what the short- and long-term consequences of widespread dam building would be. We suspect, on the basis of evidence presented here and evidence from other regions with numerous dams, that eventually the consequences would be severe, as they have been for other rivers (e.g., the Columbia River in the United States).

Figure 8. The Oxapampa Valley in central Peru illustrates a number of the forces threatening the ecological health of Andean and downstream river reaches, including the deforestation and cultivation of steep slopes and the urban development of narrow valley bottoms. Future damming of valleys such as this could significantly affect downstream fluxes of sediments and nutrients. Photograph courtesy of Thomas Saunders.

A wild card in all discussions of future scenarios in the Andean Amazon is the effect of climate change, including the feedbacks between land use and climate. There is already strong spatial variability in today's Andean climate, due to the area's topographic complexity. Even though the response of Andean environments to El Nino/La Nina events is complicated, the trend is toward heavier than normal rainfall (Kane 2000), resulting in increased landslide intensity. This may not be the case, however, in the future. Rainfall in the Andean Amazon is sensitive to the water balance of the lowland Amazon, and this balance is expected to change in predictable ways. Because rain in the Andean Amazon is ultimately derived from the Atlantic Ocean, it must be transported across the lowland Amazon basin in westward-moving air masses. During this westward movement, moisture cycles between the atmosphere and land surface, and estimations are that roughly 55% of the rain falling in the Amazon basin is derived from evapotranspiration within the basin (Marengo and Nobre 2001). For the eastern slopes of the Andes, the percentage of rainfall derived from evapotranspiration is probably higher. Consequently, continued deforestation should lead to reduced levels of precipitation in the Andean Amazon (Chagnon and Bras 2005). Both elevated carbon dioxide (CO2) and the conversion of forest to managed uses are predicted to reduce evapotranspiration and thus the amount of water moving westward toward the Andes. Elevated CO2 alone is predicted to reduce evapotranspiration in the Amazon by about 4% through reductions in stomatal conductance, and mis should also reduce rainfall. Conversion of forest to pasture across the entire Amazon basin is predicted to reduce evapotranspiration by as much as 20% (Lean et al. 1996). These changes in the regional water balance will certainly affect terrestrial and aquatic ecosystems of the Andean Amazon and thereby fundamentally alter the mountain-to-lowland fluxes discussed here. As investigations of these questions proceed at a basin scale, and as confidence in predicted changes increases, Andean policymakers should carefully examine local impacts.

The Amazon River system is unique in many ways because of its size and orientation along the equator, but the controls by its Andean headwaters are not unique. In fact, many of the mountain- lowland linkages we have discussed should be relevant to other major river systems. Similar controls are certainly observed in the adjoining Orinoco River system (Edmond et al. 1996, Jepson and Winemiller 2007) and are likely to be important in the major rivers draining the Himalayas, namely the Indus, Ganges, Brahmaputra, and Mekong. The fundamental ecological importance of these linkages stresses the need to manage even the world's largest rivers in a basin context.

Although our knowledge of the nature and magnitude of mountain- lowland linkages in the Amazon basin can serve to inform research and management in the Amazon and in basins around the world, much remains to be learned. Research in recent decades has illuminated the nature and magnitude of mountain-lowland linkages along the mainstem Amazon river, but investigations in the Andes lag for behind. Researchers still know little about the fluxes of sediments and associated nutrients from the Andes on a regional scale, and even less about the spatial and temporal variability in those fluxes. We know equally little about the degree to which river organisms depend on habitat and other resources of Andean rivers during annual and multiyear migrations. In the midst of our incomplete ecological knowledge, the Andes are being rapidly transformed into a managed landscape where rivers are modified and where montane forests and high-altitude grasslands are converted to pastures and agricultural fields. Filling these knowledge gaps is an immediate scientific challenge with important ramifications for the sustainability of the Amazon River basin as a whole. Brazil, the downstream beneficiary of Andean inputs from its upstream neighbors, should take special interest in these issues. Over the long term, the most productive components of the Brazilian Amazon River system are also the most vulnerable to poor management decisions in the Andes. Brazil's own plans for large-scale hydroelectric development, new road building, and agricultural intensification should pay similar consideration to the important hydrological and ecological linkages uniting the larger basin.


We wish to acknowledge our colleagues and collaborators in the Andean Amazon who have informed and influenced our understanding of Andean-Amazon linkages, especially Jay Brandes, Remigio Galarraga, Michael Goulding, Jean Loup Guyot, Carlos Llerena, Jose Efrain Ruiz, Richard Chase Smith, and Amy Townsend-Small. We thank the Inter- American Institute for Global Change Research, the US National Science Foundation, and the Andrew W. Mellon Foundation for supporting our research in the Amazon basin. Daniel Gann and Anna Boyette provided critical support with graphics. Michael Goulding, Margi Moss, and Thomas Saunders contributed photos. This manuscript was improved by the comments of John Melack and three anonymous reviewers.

References cited

Aalto R, Maurice-Bourgoin L, Dunne T, Montgomery DR, Nittrouer CA, Guyot JL. 2003. Episodic sediment accumulation on Amazonian flood plains influenced by El Nino/Southern Oscillation. Nature 425:493-497.

Allegre CJ, Dupre B, Negrel P, Gaillardet J. 1996. Sr-Nd-Pb isotope systematics in Amazon and Congo river systems: Constraints about erosion processes. Chemical Geology 131:93-112.

Araujo-Lima CARM, Goulding M. 1997. So Fruitful a Fish: Ecology, Conservation, and Aquaculture of the Amazon's Tabaqui. New York: Columbia University Press.

Araujo-Lima CARM, Ruffino ML. 2004. Migratory fish of the Brazilian Amazon. Pages 233-302 in Carolsfietd J, Harvey B, Ross C, Baer A, eds. Migratory Fishes of South America: Biology, Fisheries and Conservation Status. Victoria (Canada): World Fisheries Trust, World Bank, International Development Research Centre.

Araujo-Lima CARM, Forsberg BR, Victoria RL, Martinelli LA. 1986. Energy sources for detritivorous fishes in the Amazon. Science 234:1256-1258.

Aufdenkampe AK, Mayorga E, Hedges JL, Llerenac C, Quay PD, Gudeman J, Krusche AV, Richey JE. 2007. Organic matter in the Peruvian headwaters of the Amazon: Compositional evolution from the Andes to the lowland Amazon mainstem. Organic Geochemistry 38: 337- 364.

Barthem R, Goulding M. 1997. The Catfish Connection: Ecology, Migration, and Conservation of Amazon Predators. New York: Columbia University Press.

Bates HW. 1863. The Naturalist on the River Amazon. London: John Murray.

Berner RA, Rao JL. 1994. Phosphorus in sediments of the Amazon river and estuary: Implications for the global flux of phosphorus to the sea. Geochimica et Cosmochimica Acta 58:2333-2339.

Bervoets L, Solis D, Romero AM, Van Damme PA, Ollevier E 1998. Trace metal levels in chironomid larvae and sediments from a Bolivian river: Impact of mining activities. Ecotoxicologyand Environmental Safety 41:275-283.

Carvalho NO, da Cunha SB. 1998. Estimativa da carga solida do rio Amazonas e seus principals tributaries para a foz e oceano: Uma retrospective. A Agua em Revista 6:44-58.

Celi JE. 2005. The vulnerability of aquatic systems of the Upper Napo River Basin (Ecuadorian Amazon) to human activities. Master's thesis. Florida International University, Miami.

Chagnon FJF, Bras RL 2005. Contemporary climate change in the Amazon. Geophysical Research Litters 32: L13703. doi:10.1029/ 2005GL022722

Devol AH, Hedges JI- 2001. Organic matter and nutrients in the mainstem Amazon River. Pages 275-306 in McClain ME, Victoria RL, Richey JE, eds. The Biogeochemistry of the Amazon Basin. New York: Oxford University Press.

Diaz-Sarmiento JA, Alvarez-Leon R. 2004. Migratory fish of the Colombian Amazon. Pages 303-334 in Carolsfield J, Harvey B, Ross C, Baer A, eds. Migratory Fishes of South America: Biology, Fisheries and Conservation Status. Victoria (Canada): World Fisheries Trust, World Bank, International Development Research Centre.

Dudgeon D, et al. 2006. Freshwater biodiversity: Importance, status, and conservation challenges. Biological Reviews 81:163-182.

Dunne T, Merles LA, Meade RH, Richey JE, Forsberg BR. 1998. Exchanges of sediment between the flood plain and channel of the Amazon River in Brazil. Geological Society of America Bulletin 110:450-467.

Edmond JM, Palmer MR, Measures CI, Brown ET, Huh Y. 1996. Fluvial geochemistry of the eastern slope of the northeastern Andes and its foredeep in the drainage of the Orinoco in Colombia and Venezuela. Geochimica et Cosmochimica Acta 60: 2949-2976.

Eva HD, Glinni A, Janvier P, Blair-Myers C. 1998. Vegetation Map of South America at 1:5,000,000. Luxembourg (Luxembourg): European Commission. TREES Publications Series D2, EUR 18658 EN.

Forsberg BR, Araujo-Lima CARM, Martinelli LA, Victoria RL, Bonassi JA. 1993. Autotrophic carbon sources for fish of the Central Amazon. Ecology 74:643-652.

Gibbs RJ. 1967. The geochemistry of the Amazon river system, part 1 : The factors that control the salinity and the composition and concentration of suspended solids. Geological Society of America Bufletin 78:1203-1232.

Goulding M. 1993. Flooded forests of the Amazon. Scientific American 266: 114-120.

Goulding M, Smith NJH, Mahar D. 1997. Floods of Fortune: Ecology and Economy along the Amazon. New York: Columbia University Press.

Goulding M, Caflas C, Barthem R, Forsberg B, Ortega H. 2003. Amazon Headwaters-Rivers, Wildlife, and Conservation in Southeastern Peru. Lima (Peru): Eco News and Grafica Biblos.

Guyot JL, Wasson JG. 1994. Regional pattern of riverine dissolved organic carbon in the Bolivian Amazonian drainage basin. Limnology and Oceanography 39:452-458.

Guyot JL, Bourges J, Hoordbecke R, Roche MA, Calle H, Cortes J, Guzman MCB. 1988. Exportation de matieres en suspension des Andes vers l'Amazonis par k Rio Beni, Bolivie. Pages 443-452 in Bordas MP, Walling DE, eds. Sediment Budgets-Proceedings of the Porto Afegre Symposium. Wellington (CT): IAHS Press. IAHS publication no. 174.

Guyot JL, Quintanilla J, Callidonde M, Calle H. 1992. Distribution regional de la hidroquimica en la cuenca Amazonica de Bolivia. Pages 133-144 in Roche MA, Bourges J, Salas E, Diaz C, eds. Seminario sobre el PHICAB. La Paz (Bolivia): Programme Hydrologique et Climatologique de Bolivie.

Guyot JL, Jouanneau JM, Wasson JG. 1999. Characterisation of river bed and suspended sediments in the Rio Madeira drainage basin (Bolivian Amazonia). Journal of South American Earth Sciences 12:401- 410.

Hedges JI. Ertel JR, Quay PD, Grootes PM, Richey JE, Devol AH, Farwell GW, Schmidt FW, Salati E. 1986. Organic carbon-14 in the Amazon River system. Science 231:1129-1131.

Hedges JI, et al. 2000. Organic matter in Bolivian tributaries of the Amazon River: A comparison to the lower mainstem. Limnology and Oceanography 45:1449-1466. Hudson-Edwards KA, Macklin MG, Miller JR, Lechler PJ. 2001. Sources, distribution and storage of heavy metals in the Rio Pilcomayo, Bolivia. Journal of Geochemical Exploration 72:229-250.

Jepson DB, Winemiller KO. 2007. Basin geochemistry and isotopic ratios of fishes and basal production sources in four neotropical rivers. Ecology of Freshwater Fish 16:267-281.

[JRC] Joint Research Centre, European Commission. 2000. Global Land Cover 2000. (26 February 2008; www-gmi.jrc.it/glc2000f)

Junk WJ, ed. 1997. The Central Amazon Floodplain: Ecology of a Pulsing System. Berlin: Springer.

Kane RP. 2000. El Nino/La Nina relationship with rainfall at Huancayo, in the Peruvian Andes. International Journal of Climatology 20:63-72.

Keil RG, Mayer LM, Quay PD, Richey JE, Hedges JI. 1997. Loss of organic matter from riverine particles in deltas. Geochemica et Cosmochimica Ada 61:1507-1511.

Lean J, Bunton CB, Nobre CA, Rovmtree PR. 19%. The simulated impact of Amazonian deforestation on climate using measured ABRACOS vegetation characteristics. Pages 549-576 in Gash JHC, Nobre CA, Roberts JM, Victoria RL, eds. Amazonian Deforestation and Climate. New York: waey.

Marengo JA, Nobre CA. 2001. General characteristics and variability of climate in the Amazon basin and its links to the global climate system. Pages 17-41 in McClain ME, Victoria RL, Richey JE, eds. The Biogeochemistry of the Amazon Basin. New York: Oxford University Press.

Maurice-Bourgoin L, Quiroga I, Guyot JL, Malm 0.1999. Mercury pollution in the upper Beni river, Amazonian basin: Bolivia. Amhio 28: 302-306.

Mayorga E, Aufdenkampe AK, Masiello CA, Krusche AV, Hedges JI, Quay PD, Richey JE, Brown TA, 2005. Young organic matter as a source of carbon dioxide outgassing from Amazonian rivers. Nature 436:538- 541.

McClain ME, Cossio RE. 2003. The use and conservation of riparian zones in the rural Peruvian Amazon. Environmental Conservation 30:242-248.

McClain ME, Richey JE. 1996. Regional-scale linkages of terrestrial and lotie ecosystems in the Amazon basin: A conceptual model for organic matter. Archiv fur Hydrobiologie (suppl.) 113:111- 125.

McClain ME, Richey JE, Brandes JA, Pimentel TP. 1997. Dissolved organic matter and terrestrial-lotic linkages in the central Amazon basin of Brazil Global Biogeochemical Cycles 11:295-311.

McClain ME, Victoria RL, Richey JE, eds. 2001. The Biogeochemistry of the Amazon Basin. New York: Oxford University Press.

Meade RH. 1994. Suspended sediments of the modem Amazon and Orinoco rivers. Quaternary International 21:29-39.

Meade RH, Dunne T, Richey JE, Santos UdM, Salati E. 1985. Storage and remobilization of sediment in the lower Amazon River of Brazil, Science 228:488-490.

Meggars BJ. 1984. The indigenous peoples of Amazonia, their cultures, land use patterns and eflects on the landscape and biota. Pages 627-648 in Sioli H, ed. The Amazon: Limnology and Landscape Ecology of a Mighty Tropical River and Its Basin. Hingham (MA): fGuwer Academic.

Melack JM, Forsberg BR. 2001. Biogeochemistry of Amazon floodptain lakes and associated wetlands. Pages 235-274 in McClain ME, Victoria RL, Richey JE, eds. The Biogeochemistry of the Amazon Basin. New York: Oxford University Press.

Mena CA, Bilsborrow R, McClain ME. 2006. Socioeconomic drivers of deforestation in the Napo River Basin of Ecuador. Environmental Management 37:802-815.

Montgomery DR, Balco G, WHlett SD. 2001. Climate, tectonics, and the morphology of the Andes. Geological Society of America Bulletin 29: 579-582.

Naiman RJ, Decamps H. 1997. The ecology of interfaces: Riparian zones. Annual Review of Ecology and Systematics 28:621-658.

Naiman RJ, Decamps H, McClain MR 2005. Riparia: Ecology, Conservation, and Management of Streamside Communities. New York: Elsevier.

Quay PD, Wilbur DO, Richey JE, Hedges JI, Devol AH, Martinelli LA, 1992. Carbon cycling in the Amazon River: Implications from the ^sup 13^C composition of paniculate and dissolved carbon. Limnology and Oceanography 37:857-871.

Richey JE, Victoria RL. 1993. C, N, and P export dynamics in the Amazon River. Pages 123-140 in Wollast R, Mackenzie FT, Chou L, eds. Interactions of C1 N, P, and S Biogeochemical Cycles and Global Change. Berlin: Springer.

Richey JE, Hedges JI, Devol AH, Quay PD. 1990. Biogeochemistry of carbon in the Amazon River. Limnology and Oceanography 35:352-371.

Richey JE, Melack JM, Aufdenkampe AK, Ballester VM, Hess L. 2002. Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO2. Nature 416:617-620.

Saunders TJ, McClain ME, Llerena CA. 2006. The biogeochemistry of dissolved nitrogen, phosphorus, and organic carbon along terrestrialaquatic flowpaths of a montane headwater catchment in the Peruvian Amazon. Hydrological Processes 2ft 2549-2562.

Sioli H, ed. 1984. The Amazon: Limnology and Landscape Ecology of a Mighty Tropical River and Its Basin. Dordrecht (Netherlands): W. Junk.

Sobieraj JA, Elsenbeer H, McClain M. 2002. The cation and silica chemistry of a Subandean river basin in western Amazonia. Hydrological Processes 16:1353-1372.

Stallard RF, Edmond JM. 1983. Geochemistry of the Amazon, 2: The influence of geology and weathering environment on the dissolved load. Journal of Geophysical Research 88:9671-9688.

Townsend-Small A, McClain ME, Brandes JA. 2005. Contributions of carbon and nitrogen from the Andes Mountains to the Amazon Riven Evidence from an elevational gradient of soils, plants, and river material Limnology and Oceanography 50:672-685.

Townsend-Small A, Noguera IL, McClain ME, Brandes JA. 2007. Radiocarbon and stable isotope geochemistry of organic matter in die Amazon headwaters, Peruvian Andes. Global Biogeochemical Cycles 21: GB2029. doi:10.1029/2006GB002835

Townsend-Small A, McClain ME, Hall B, Llerena CA, Noguera JL, Brandes JA. 2008. Contributions of suspended organic matter from mountain headwaters to the Amazon River: A one-year time ser