Nitrogen Transformations In Flowpaths Leading From Soils to Streams In Amazon Forest and Pasture

Nitrogen Transformationsin Flowpaths Leading from Soils

to Streams in Amazon Forestand Pasture

Joaquın Chaves,1,2* Christopher Neill,1 Sonja Germer,3 Sergio GouveiaNeto,4 Alex V. Krusche,4 Adriana Castellanos Bonilla,4

and Helmut Elsenbeer3

1The Ecosystems Center, Marine Biological Laboratory, Woods Hole, Massachusetts 02543, USA; 2Department of Ecology

and Evolutionary Biology, Brown University, Providence, Rhode Island 02912, USA; 3Institut fur Geookologie, Universitat Potsdam,Potsdam, Germany; 4Laboratorio de Ecologia Isotopica, CENA – USP, Piracicaba, SP, Brazil

ABSTRACT

The modification of large areas of tropical forest to

agricultural uses has consequences for the move-

ment of inorganic nitrogen (N) from land to water.

Various biogeochemical pathways in soils and

riparian zones can influence the movement and

retention of N within watersheds and affect the

quantity exported in streams. We used the concen-

trations of NO3- and NH4

+ in different hydrological

flowpaths leading from upland soils to streams to

investigate inorganic N transformations in adjacent

watersheds containing tropical forest and estab-

lished cattle pasture in the southwestern Brazilian

Amazon Basin. High NO3- concentrations in forest

soil solution relative to groundwater indicated a

large removal of N mostly as NO3- in flowpaths

leading from soil to groundwater. Forest ground-

water NO3- concentrations were lower than in other

Amazon sites where riparian zones have been

implicated as important N sinks. Based on water

budgets for these watersheds, we estimated that 7.3–

10.3 kg N ha-1 y-1 was removed from flowpaths

between 20 and 100 cm, and 7.1–10.2 kg N ha-1 y-1

was removed below 100 cm and the top of the

groundwater. N removal from vertical flowpaths in

forest exceeded previously measured N2O emissions

of 3.0 kg N ha-1 y-1 and estimated emissions of NO

of 1.4 kg N ha-1 y-1. Potential fates for this large

amount of nitrate removal in forest soils include

plant uptake, denitrification, and abiotic N reten-

tion. Conversion to pasture shifted the system from

dominance by processes producing and consuming

NO3- to one dominated by NH4

+, presumably the

product of lower rates of net N mineralization and

net nitrification in pasture compared with forest. In

pasture, no hydrological flowpaths contained sub-

stantial amounts of NO3- and estimated N removal

from soil vertical flowpaths was 0.2 kg N ha-1 y-1

below the depth of 100 cm. This contrasts with the

extent to which agricultural sources dominate N

inputs to groundwater and stream water in many

temperate regions. This could change, however, if

Received 28 April 2008; accepted 7 July 2009; published online

10 September 2009

Author Contributions: Chaves: Collected and analyzed isotope sam-

ples, constructed ecosystem budgets, and wrote paper. Neill: Designed

study, co-wrote paper. Germer: Instrumented watersheds, collected field

samples, analyzed flowpath chemistry, and contributed to data analyses.

Neto: Instrumented watersheds, collected field samples, ran laboratory

chemistry, and organized flowpath chemistry. Krusche: Assisted with

watershed instrumentation and oversaw laboratory chemistry. Bonilla:

Instrumented watersheds, collected field samples, ran laboratory chem-

istry, and organized flowpath chemistry. Elsenbeer: Co-designed study,

supervised watershed instrumentation, and contributed to paper writing.

*Corresponding author; e-mail: [email protected]

Ecosystems (2009) 12: 961–972DOI: 10.1007/s10021-009-9270-4

� 2009 Springer Science+Business Media, LLC

961

pasture agriculture in the tropics shifts toward

intensive crop cultivation.

Key words: nitrogen; soils; Amazon; forest;

pasture; flowpaths; land use.

INTRODUCTION

Controls on the retention and movement of nitro-

gen (N) within watersheds are important ecosys-

tem variables because N often acts as a regulator of

plant production and thus can affect carbon storage

(Finzi and others 2007) and cycling of other plant

nutrients such as phosphorus (P) (Perring and

others 2008). At the same time, the processes that

control N retention are only partially effective at

compensating for increased anthropogenic N loads

to the terrestrial portions of watersheds, leading to

increasing concentrations of inorganic N in major

rivers of the world (Peierls and others 1991; Goo-

lsby and Battaglin 2001; Boyer and others 2002;

Mulholland and others 2008).

Fertilizer use and anthropogenic releases of N

from fossil fuel burning are growing rapidly in

tropical regions (Matthews 1994; Matson and oth-

ers 1999). Meanwhile, tropical forests are under-

going more rapid rates of conversion to agriculture

than any other biome (Matthews and others 2000;

Achard and others 2002; Fearnside 2005). Under-

standing processes that control N removal in trop-

ical landscapes is important because N in primary

moist tropical forests behaves largely as if vegeta-

tion and soils are N-saturated (Hall and Matson

1999). Foliar N, litterfall fluxes (Vitousek 1984),

and rates of N mineralization and nitrification in

soils are high (Robertson 1989; Neill and others

1997b; Verchot and others 1999). This leads to high

concentrations of NO3- in solution of mineral soil

in tropical forest (Neill and others 2001, 2006;

Markewitz and others 2004). This is consistent with

limitation of plant growth by an element other

than N (likely P in most cases), which allows a

‘‘leakier’’, open soil N cycle (Hall and Matson 1999;

Markewitz and others 2004). In contrast, replace-

ment of forest with pasture lowers rates of soil N

stocks and cycling, NO3- production, and solution

N concentrations on a variety of Amazon soils

(Neill and others 1997b, 2006; Verchot and others

1999; Markewitz and others 2004).

In tropical forest, concentrations of NO3- and

dissolved inorganic N (DIN: NO3- + NH4

+) in small

streams draining seemingly N-saturated forests,

such as those in the lowland Amazon, are low and

typically two orders of magnitude less than con-

centrations in soil solution (Lesack 1993; Williams

and Melack 1997; Neill and others 2001; Markewitz

and others 2004). It has been suggested that high N

removal occurs in flowpaths leading from soils to

streams, primarily by denitrification in saturated

riparian zones (McClain and others 1994; Brandes

and others 1996; Williams and others 1997).

Compared with riparian zones, which occur in

limited areas in watersheds, upland soils dominate

the landscape in terra firme lowland tropical forests

not subjected to seasonal inundation. Because soil

saturation is not required for denitrification, which

can occur in soils where anaerobic microsites exist

(Tiedje and others 1984), upland soil denitrification

losses could also be a significant path for N loss at

the ecosystem level. Multiple sources of evidence

suggest that a large fraction of the N2O emissions

from terra firme tropical forest soils with percentage

water-filled pore space greater than 50% can be

attributed to denitrification (Davidson and others

2000; Neill and others 2005). In tropical montane

forests in Hawaii, comparisons of N isotope mass

balances across a precipitation gradient suggested

that denitrification in soil accounted for 24–53% of

total ecosystem N loss (Houlton and others 2006).

We examined the fate of N in forest and pasture

watersheds of the Amazon basin. The Amazon

contains the world’s largest area of contiguous

tropical forest and now experiences steadily high

rates of deforestation and conversion to cattle pas-

ture and other agricultural uses (INPE 2008). In

this study, we examined the concentrations of

NO3- and NH4

+ in soil solution, groundwater, and

different hydrological flowpaths that lead to forest

and pasture stream flow over an entire rainy season

in two adjacent watersheds in the southwestern

Amazon Basin. We also quantified d15N values of

NO3- in each flowpath. We use the results to

estimate ecosystem N fluxes and losses in pathways

from upland soils to small order streams in con-

trasting forest and pasture land covers.

STUDY SITE

Field measurements were carried out at Rancho

Grande (10�18¢ S, 62�52¢ W, 143 m a.s.l.), a cattle

ranch in the Brazilian state of Rondonia, in the

southwestern Amazon Basin (Figure 1). Mean

annual precipitation is 2300 mm with a pro-

nounced dry season of less than 100 mm month-1

from June through August (Zimmermann and

962 J. Chaves and others

others 2006). Measurements were conducted on

forest and pasture watersheds of approximately

1 ha located 400 m apart. The forest watershed

contained open moist tropical forest with a high

density of palms typical of the region (Pires and

Prance 1986; Germer and others 2006). The pasture

was cleared in 1985, planted to pasture in 1986 and

had continuous cover of the African forage grass

Brachiaria humidicola since that time. It has been

actively grazed at about 1 animal ha-1 but never

fertilized, both typical of pastures in the region

(Neill and others 1997a). Both watersheds were

drained by zero-order, ephemeral streams that lack

streamside, saturated riparian zones. The area be-

longs to a morphostructural unit known as

‘‘Southern Amazon Dissected Highlands’’ (Planalto

Dissecado Sul da Amazonia, Peixoto de Melo and

others 1978), which is characterized by a very

pronounced topography with an altitudinal differ-

ential of up to 150 m. Remnant ridges of Pre-

cambrian basement rock, made up of granites and

gneisses of the Complexo Xingu (Leal and others

1978), are separated by flat valley floors of varying

width. Soil orders associated with this morpho-

structural unit are Ultisols and Oxisols in valleys

and Inceptisols and Entisols (Soil Survey Staff 1999)

along streams. Soils underlying the study water-

sheds are Kandiudults (Zimmermann and others

2006) based on clay content changes with depth

and cation exchange capacity in the subsoil, which

exceeds 2 m in depth. Soil clay contents in both

watersheds range from 20% in the A horizon to

40% in the B horizon. Soil pH in water (2.5:1) on

air-dried soil at 0–10 cm was 5.4 in the forest and

6.2 in the pasture. The corresponding cation

exchange capacities were 5.8 cmolc kg-1 in the

Figure 1. Location of the

study site: (A) Marker

shows the location of

Rancho Grande cattle

ranch in central

Rondonia. (B) Pasture

watershed. (C) Forest

watershed. Contour

intervals are every 0.5 m;

elevations are shown

relative to the watershed

outlets.

N Transformations in Amazon Soils 963

forest and 3.8 cmolc kg-1 in the pasture (Neill and

others 1997a). The stocks of N in pasture soils are

reduced by approximately 50% compared with the

adjacent forest (Neill and others 1997a).

We chose these watersheds after an extensive

survey of many potential sites. They are closely

matched in geomorphology, soil characteristics, and

elevation and are representative of the soils and

vegetation of Rondonia. Our earlier surveys of soil

hydraulic properties indicated that infiltration rates

were significantly lower in pasture, where average

rates were 13 times lower than those in forest, and

that infiltration rates decrease considerably at soil

depths of 12.5–20 cm where an impeding layer

occurs regardless of land cover (Zimmermann and

others 2006). Permeability at a depth of 20 cm is as

much as two orders of magnitude lower than infil-

trability (Zimmermann and others 2006). This leads

to a 20-fold increase in pasture streamflow relative

to forest streamflow (Chaves and others 2008;

Germer and others 2009).

MATERIALS AND METHODS

Water Balance

Hydrological measurements were made in both

watersheds from the end of the dry season in August

2004 through July 2005. A tipping bucket rain gauge

(Hydrological Services P/L, Liverpool, Australia) and

a Campbell logger installed in the pasture recorded

rainfall intensity. Forest throughfall quantities were

measured after each rain event at 20 trough collec-

tors installed 1 m above the ground distributed

throughout the watershed (Figure 1) and forest

stream flow was measured from the same events

from January to March 2005 (Germer and others

2006). Rainfall quantities were measured in two

trough collectors in the pasture watershed. Stream

flow was calculated from stage height measured

continually for the duration of the study with 50 cm

water level loggers (TruTrack, Christchurch, New

Zealand) at H-flumes (30.5 cm) at the outlet of each

watershed.

Estimated ranges for evapotranspiration (ET)

from dry to wet years for the forest and pasture

central Rondonia sites, Reserva Jaru (1387–

1629 mm y-1) and Nossa Senhora Ranch (1024–

1419 mm y-1), respectively, were derived from

Kabat and others (1997). Reserva Jaru is 100 km

northwest of Rancho Grande, and Nossa Senhora is

160 km southeast. Both lie in a zone of similar

annual rainfall. During the year of our study,

rainfall at our site was close to the 18-year average,

which should make the range of ET values appro-

priate for our computations. Groundwater recharge

ranges were calculated by difference from the other

components of the water balance measured in each

watershed and the range of ET values for each land

use type.

Hydrochemistry

We sampled a subset of the precipitation events

that generated runoff in the forest (22 events) and

pasture (27 events) watersheds for chemical anal-

yses. Precipitation during those sampled events

accounted for approximately 30% of the total

precipitation recorded during the study. Rainfall

and throughfall samples were collected after each

event into 20 l acid-washed polyethylene contain-

ers (Germer and others 2006). Overland flow

samplers (Kirkby and others 1976) were placed at

three locations in each watershed. Stream water

was collected at the outlet of each watershed from

the H-flumes using automatic water samplers

(ISCO, Lincoln, Nebraska, USA). Rain, throughfall,

overland flow, and stream water samples were

recovered no later than 2 h after each event, or

alternatively the next morning for events that

ended after 9 pm (Germer and others 2007).

Samples of up to 1 l were transferred into poly-

ethylene acid-washed bottles and returned to the

field laboratory and stored on ice in coolers.

Samples of soil solution from the vadose zone,

above and below the texture break at the top of the

B horizon (20, 100 cm), were taken every 10 days

from tension lysimeters installed at the 10 instru-

ment nests in each watershed (Figure 1). Ground-

water was sampled once weekly from three wells in

each watershed. Wells were 4–6 m deep, 5 m apart,

and approximately 10 m from the outlet of each

watershed. Each well was bailed 10 times

(2.5 times well volume) and let recharge before

each sample was taken. Soil and groundwater

samples were handled and stored in the field as

those for event water samples.

Once in the field laboratory all samples were

filtered through Whatman GF/F filters (0.7 lm

pore size), preserved with approximately 5 mg of

thymol and stored frozen in 60 ml acid-washed

polyethylene bottles. Samples were then packed in

Styrofoam coolers and shipped to Centro de Ener-

gia Nuclear na Agricultura (CENA), at the Univer-

sity of Sao Paulo, in Piracicaba, SP, where they

arrived still partially frozen. Once received, the

samples were stored frozen until chemical analyses.

Samples for 15NO3- isotopic analyses were col-

lected during the second half of the rainy season,

from January to April. The timing, choice, and


amount of these samples depended on adequate

sample volumes and simultaneous sampling of

flowpath chemistry. Precipitation was sampled

from a rain event on February 23. Samples of

throughfall were obtained from eight collectors

during a rain event on February 17 and pooled into

four composite samples. Groundwater samples

were taken from each of the three wells at each

watershed on January 20 and processed individu-

ally. Samples of soil solution were collected from all

the lysimeters on February 26 and pooled into

composite samples; one sample from each soil

depth from each watershed, for a total of four

samples. Forest stream flow was collected on Feb-

ruary 9; the 15 samples collected by the ISCO

sampler on that flow event were pooled into seven,

consecutive two-sample composites plus an addi-

tional individual sample. Pasture stream flow was

obtained on January 29 from 21 ISCO sampler

bottles; samples were pooled into seven consecu-

tive three-sample composites, one sample was lost

to breakage. Four individual samples of forest

overland flow were collected during rain events on

February 19 and 21. Pasture overland flow samples

(2) were collected on February 23.

Laboratory Analyses

Concentrations of NO3- and NH4

+ were deter-

mined by ion chromatography (Dionex DX-500).

Stable isotope analyses of 15NO3- were carried out

in 1- to 4-l samples by raising the pH to eliminate

NH4+, subsequent reduction of NO3

- to NH4+ by

addition of Devarda’s Alloy, and diffusion of NH4+

onto acidified filters using the method of Sigman

and others (1997). Sample processing and diffu-

sions were carried out at the field laboratory. Iso-

topic ratios of the samples diffused onto the

acidified filters were measured on a Finnigan Delta

S continuous flow mass spectrometer at CENA.

Data Analysis

We used analysis of variance (ANOVA) to test for

differences in NO3- and NH4

+ concentrations be-

tween watersheds and among flowpaths. Data were

averaged by rain event or sampling date for each

flowpath and watershed, so that the mean of each

event or sampling date for each flowpath consti-

tuted a separate sample. Data were log-transformed

to minimize departures from normality. To test

differences between watersheds, we performed a

two-way factorial ANOVA with watershed and

flowpath as factors. To compare equivalent flow-

paths between watersheds, we excluded rain and

throughfall data from this analysis because

throughfall is exclusive to forest, and rain is com-

mon to both watersheds. Rain and throughfall data

were included separately in analyses to test for

differences among flowpaths within watersheds

using independent one-way ANOVAs for each

watershed; rain and throughfall data were included

in the forest analysis, and rain in the pasture

analysis. Further differences among flowpaths were

evaluated with Tukey’s HSD post hoc tests (Craw-

ley 2002). Differences in d15NO3- between water-

sheds were evaluated with Student’s t-test on data

averaged by flowpath; rain and throughfall data

were excluded as well for this analysis. Differences

in d15NO3- among flowpaths with sufficient sam-

ple numbers were also evaluated using t-tests. All

statistical analyses were carried out using the lan-

guage for statistical computing R version 2.3.1 (R

Development Core Team 2006).

We used the water budget, throughfall from Ger-

mer and others (2007), ranges of groundwater re-

charge, and the changes in median concentrations of

NO3- and NH4

+ from soil solution at 20–100 cm and

groundwater to estimate ecosystem N fluxes in forest

and pasture ecosystems. N losses from flowpaths

were calculated as the differences in fluxes between

flowpaths or soils depths. We assumed no intercep-

tion and change in throughfall chemistry in pasture.

The uncertainty in the N flux estimates was provided

by the uncertainty introduced in the groundwater

recharge by the range of ET values for both land uses.

N fluxes in surface runoff were calculated from the

concentrations of NO3- and NH4

+ in stream flow and

stream water discharge.

RESULTS

Concentrations of NO3- in forest flowpaths were

significantly higher than in pasture flowpaths

(F = 41.3, df = 4, 225, P < 0.001). Concentrations

of NO3- became progressively higher as water

passed through the forest canopy as throughfall

(median = 8.1 lmol l-1) and reached the forest

floor and became overland flow (med-

ian = 31.6 lmol l-1) or infiltrated into the soil

(Figure 2A). Forest soil solution at 20 cm had the

highest median concentration (176 lmol l-1) of all

the flowpaths sampled in both watersheds. Con-

centrations of NO3- were lower at 100 cm in forest

soil solution (median = 85 lmol l-1), and much

lower (1 lmol l-1) in groundwater. Groundwater

consistently had the lowest NO3- concentrations

and was the only flowpath in forest significantly

different from all others, including rain (Tukey test,

P < 0.001). Forest stream flow had a median NO3-

concentration of 34 lmol l-1.


All flowpaths and stream flow in pasture had

median NO3- concentrations of 1 lmol l-1 or

lower (Figure 2A). With the exception of ground-

water, which also had the lowest NO3- values

(median = 0.2 lmol l-1) among flowpaths, con-

centrations of NO3- in pasture flowpaths were

indistinguishable from each other or precipitation

(Tukey test, P > 0.05).

Although mean NH4+ concentrations were

higher in forest flowpaths, they were not statisti-

cally different from those in pasture (F = 1.25, df =

4, 207, P > 0.05). Median NH4+ concentrations

ranged between 1 and 12.5 lmol l-1 (Figure 2B).

The highest NH4+ concentrations were generally in

precipitation and throughfall and the lowest con-

centrations were in groundwater regardless of

landcover (Figure 2B). Groundwater NH4+ was in

all cases significantly lower than other flowpaths

(Tukey test, P < 0.05).

In forest, NO3- in soil solution at both 20 and

100 cm increased at the onset of the rainy season

but otherwise showed little seasonal pattern

(Figure 3A). Forest groundwater NO3- concentra-

tions remained low throughout the year

(Figure 3A). Forest NH4+ concentrations in soil

solution were high in the dry season and decreased

steadily through the rainy season to levels compa-

rable to those in groundwater, where NH4+ con-

centrations remained fairly constant at slightly

more than 1 lmol l-1 (Figure 3B). In pasture, both

NO3- and NH4

+ concentrations in soil solution and

groundwater were less than 1 lmol l-1 in the dry

season and increased slightly during the rainy

season (Figure 3C, D). Concentrations of NO3- in

pasture groundwater increased from dry to wet

seasons (Figure 3C) but the concentration of NH4+

was similar throughout the year (Figure 3D).

The d15NO3- for all samples varied between -5.7

and 10.3& (Figure 4). All forest flowpaths were 15N-

enriched relative to precipitation (mean = 0.7&)

and ratios were higher in forest than in pasture

(t = 2.87, df = 8, P < 0.05). Forest soil solution at

20 cm had the highest d15NO3- (8.7&), although

our n = 1 precluded any statistical comparison with

values from other flowpaths. In forest, soil

(mean = 7.4&) and ground waters (mean = 8.4&)

had the highest isotopic ratios and overland flow

(mean = 2.8&), and streamflow (mean = 2.9&)

was the lowest, whereas throughfall spanned the

data range (mean = 5.2&, min = 2.1&, max =

10.3&). In pasture, all flowpaths except ground-

water (mean = 5.9&) were 15N-depleted relative to

precipitation. Pasture stream d15NO3- (mean =

-1.7&) was significantly lower than in the forest

stream (t = 5.12, df = 12, P < 0.001).

Runoff response to rainfall, both in quantity and

frequency, was much higher in pasture than in

forest (Figure 5). Total incident rainfall recorded for

the period of study was 2286 mm, close to the 1984–

2003 annual average (2300 mm; Germer and others

Figure 2. Box plots (min,

25%, median, 75%, and

max) of concentration

(lmol l-1) of (A) NO3-

and (B) NH4+ in direct

precipitation (Precip.),

forest throughfall (TF),

overland flow (OF), soil

solution (20, 100 cm),

groundwater (GW), and

streamflow (Stream) in

the forest (F) and pasture

(P) watersheds.


2006 and Table 1). Forest throughfall and stream

flow were estimated at 89% and 8% of precipitation

during this study (Germer and others 2006). Stream

flow was 1% of precipitation (24 mm) in the forest

and 18% of precipitation (416 mm) in the pasture.

The ranges for groundwater recharge were in 582–

824 mm y-1 in forest and 451–846 mm y-1 in

pasture.

The fluxes of inorganic N within forest and pas-

ture ecosystems differed widely despite identical

inputs. In forest, NH4+ fluxes increased slightly in

throughfall then decreased through soils and to

groundwater (Table 1). Forest NO3- fluxes also

increased slightly in throughfall but increased dra-

matically in soils before decreasing in groundwater

(Table 1). In forest, 94% of the DIN flux through

Figure 3. Average

concentrations

(lmol L-1) of (A) NO3-

and (B) NH4+ in forest,

and (C) NO3- and

(D) NH4+ in pasture, in

soil solution at 20 and

100 cm, and groundwater

for each sampling date.

Figure 4. d15N values of NO3- (&) in direct precipitation, throughfall, overland flow stormflow, soil solution (20,

100 cm), groundwater, and streamflow in the forest and pasture watersheds. Open circles represent individual measure-

ments, solid squares the mean, and error bars represent ±1 standard deviation. Abbreviations are as in Figure 2.


soil occurred as NO3-. Differences between fluxes

in groundwater and soil solution indicated that

7.3–10.3 kg N ha-1 y-1 was lost as water traveled

between soil solution at 20–100 cm, and an addi-

tional 7.1–10.2 kg N ha-1 y-1 was lost as water

moved from soil solution at 100 cm to ground-

water. Differences in assumptions about annual

water balance were important only for calculating

NO3- fluxes in soils and did not change the con-

clusions about where NO3- was lost from soil

flowpaths. In pasture, the largest change in N

fluxes was a decrease from throughfall and soil

Figure 5. (A)

Precipitation (mm day-1),

(B) forest, and (C) pasture

daily stream runoff

(mm day-1) during the

period of study.

Table 1. Annual N Fluxes as N-NO3-, N-NH4

+, and Dissolved Inorganic N (DIN: NO3- + NH4

+) Calculatedfrom Changes in Median Concentration Between Soil Depths and Groundwater and the EstimatedGroundwater Recharge for Each Watershed

Flowpath Forest Pasture

Water

(mm)

NH4+

(kg ha-1)

NO3-

(kg ha-1)

DIN

(kg ha-1)

Water

(mm)

NH4+

(kg ha-1)

NO3-

(kg ha-1)

DIN

(kg ha-1)

Rain 2286 4.6 0.8 5.4 2286 4.6 0.8 5.4

Evapotranspiration 1387–1629 1024–1419

Throughfall 2236 5.7 2.1 7.8 2286 4.6 0.8 5.4

Soil solution 20 cm 0.4–0.5 14.3–20.8 14.7–20.8 0.1 0.1 0.2

Soil solution 100 cm 0.4–0.6 7.0–9.9 7.4–10.5 0.1 0.1 0.2

Groundwater 582–824 0.2 0.1 0.3 451–846 0 0 0

Stream water 24 0 0.2 0.2 416 0.3 0.1 0.4

Interception was estimated as precipitation minus the sum of throughfall and stream flow (Germer and others 2006). Evapotranspiration (ET) estimates were for the centralRondonia forest at Reserva Jaru and pasture at Nossa Senhora (Kabat and others 1997). Througfall fluxes are from Germer and others (2007). Groundwater recharge wascalculated as the difference of precipitation and the sum of the other balance terms in each watershed.


solution (Table 1). Little N was lost as water moved

from soils to groundwater because N fluxes

through soils were extremely low. The export of

DIN in stream water was 0.2 kg N ha-1 y-1 from

forest and 0.4 kg ha-1 y-1 from pasture (Table 1).

The range of groundwater recharge estimates for

the pasture resulted in little variation in estimates

of N fluxes from soil to groundwater due to the low

DIN levels at all soil depths.

DISCUSSION

There were important differences in the location,

form, and magnitude of DIN fluxes through forest

and pasture watersheds. In forest, fluxes of NH4+

decreased and fluxes of NO3- increased as water

moved from throughfall to shallow soil solution.

This led to high DIN fluxes through soil solution. In

pasture, fluxes of both NH4+ and NO3

- decreased

from throughfall to soils, and DIN fluxes through

soils were very low. This pattern is consistent with

high rates of net N mineralization and net nitrifi-

cation in soils of Amazon forest compared with

pasture (Neill and others 1997b; Verchot and others

1999). Davidson and others (2007) showed that old

growth Amazon forests exhibit abundant soil NO3-

and high rates of N cycling, but young Amazon

forests regrowing on abandoned pastures have low

soil NO3- and low soil N availability that reflect

high N losses and inducement of plant N limitation

during deforestation and pasture management. Our

forest showed the same high soil NO3- availability

and our pasture exhibited the same low soil N

availability consistent with plant N limitation at the

extreme young forest end of that gradient.

In forest, soil solution NO3- was high and

showed little seasonal variation during the year,

but soil solution NH4+ declined following wetting of

soils after the first major rains in November. Sta-

bilization of wet season concentrations just below

those in rain and throughfall suggested dilution of

soil solution by rain during the wettest period.

Higher seasonal rates of net nitrification in wet soils

(Neill and others 1995) may also contribute to de-

cline in NH4+ after the dry season. The increase in

NO3- in pasture from dry to wet seasons may also

be influenced by rain inputs; only in this case, soil

solution NO3- was likely depleted by plant demand

during the dry season but replenished by rain and

approached concentrations in rain after the begin-

ning of the wet season. Groundwater concentra-

tions varied little, probably because different travel

times for soil solution resulted in water reaching

the groundwater throughout the year.

Our results indicated a large removal of N (14.5–

20.5 kg ha-1 y-1), mostly as NO3-in hydrological

flowpaths leading from soil solution at 20 cm to

groundwater in the forest watershed. Evidence for

this comes from both the decline in concentration

of NO3- from soil solution to groundwater and the

vertical water movement from the unsaturated

zone to groundwater as derived from the water

balance. Lower concentrations of NO3- in ground-

water cannot be explained by lateral transfers of

soil solution to streams above the regional

groundwater because water export as stream flow

in the forest watershed amounted to only 1% of

annual precipitation (2286 mm), whereas vertical

flowpaths moved an estimated range of 582–

824 mm from soil solution to groundwater.

We cannot unequivocally establish the fate of the

N that disappeared from the forest soil-to-ground-

water flowpath, but N fluxes at this site and work

in other tropical forests suggest that denitrification

is an important component of this loss. Emissions of

N2O from forest soil at Rancho Grande have been

reported at 3.0 kg N ha-1 y-1 (Garcia-Montiel

and others 2004), whereas forest NO fluxes at

nearby Fazenda Nova Vida were estimated at

1.4 kg N ha-1 y-1 (Neill and others 2005). Com-

bined, these fluxes could account for gaseous N gas

loss of 4.4 kg ha-1 y -1, which is 43–62% of the

calculated N loss below 100 cm. Because the

potential contribution to gaseous N flux from deni-

trification to N2 in Amazon forest soil remains

unquantified and poorly constrained, denitrifica-

tion potentially could account for all of the

observed hydrologic N loss in soil flowpaths.

Mature tropical forests are largely presumed to be

P rather than N limited (Denslow and others 1987;

Herbert and Fownes 1995; Vitousek and Farrington

1997; Vitousek 2004; Davidson and others 2007).

Observations from multiple tropical forest sites

suggest that stocks of N and leaching and gaseous

losses are high (Martinelli and others 1999) and that

cycling of N is not conservative (Vitousek 1984). N

concentrations and fluxes in our forest watershed

were consistent with that view. We suggest that

plant N demand may contribute to the concentra-

tion gradient and N lost from hydrologic flowpaths

in shallow soil; however, plant N demand is un-

likely to explain losses between soils at 100 cm and

groundwater. This is because fine roots in Amazon

forest are most abundant in the top 30 cm of soil

and an order of magnitude lower below 1 m

(Nepstad and others 1994). In a forest in French

Guiana fine roots below 50 cm contributed little

or no significant Ca and Sr uptake to the whole

tree economy (Poszwa and others 2002). If root


distribution in our forest site is comparable to that at

these locations, plant uptake might account for

some of the 7.3–10.3 kg ha-1 y-1 N loss between

20 and 100 cm. However, it would be unlikely to

explain uptake from deeper depths where root

biomass is dramatically lower and where invest-

ment in N uptake would require greater plant

investment compared with uptake from shallower

depths where N in soil solution is abundant.

The higher d15NO3- values measured in forest

soil solution and groundwater relative to precipi-

tation and other forest flowpaths provided some

indication that denitrification could be one of the

pathways for the estimated N losses. In Hawaiian

forests, preferential 14N loss through denitrification

explained elevated d15N values inputs in bulk soil

N, and soil solution NO3- relative to those in pre-

cipitation (Houlton and others 2006). In that study,

stream N loss had high d15N values as well and

could not account for the 15N enrichment in soil N.

In our forest watershed, N losses to both stream

water and groundwater were 15N-enriched relative

to precipitation (Figure 4). Therefore, a gaseous

preferential 14N loss would be necessary to account

for the higher N isotope ratios in soil solution and

groundwater. A well-constrained N isotope budget,

which is precluded by our low sample number,

would be necessary to provide unequivocal evi-

dence for denitrification as the major N sink

mechanism.

Despite large differences in DIN fluxes within

forest and pasture watersheds, export of DIN in

stream water was low and similar (0.2–

0.4 kg N ha-1 y-1) in forest and pasture. In small

watersheds in the central Amazon, abrupt decreases

of NO3- concentration and 15N enrichment of NO3

-

along groundwater toposequences from upland

forest to stream riparian zones have been interpreted

to indicate denitrification and N removal in the

streamside zone, leading to low export of N in stream

water (McClain and others 1994; Brandes and others

1996). One prominent difference between these

sites and our study site is that ours lacked a saturated

riparian zone. Our results, in contrast, suggest that N

removal occurs in terra firme soils and is likely dis-

tributed throughout the watershed, rather than

concentrated in riparian zones.

Compared to N dynamics in other Amazon

watersheds, the groundwater NO3- concentrations

we measured in forest were much lower than the

approximately 30–50 lmol l-1 measured in the

upland portions of watersheds in the central

Amazon (McClain and others 1994; Brandes and

others 1996; Williams and others 1997) or in deep

soil solution in the eastern Amazon (Markewitz

and others 2004). These differences may be a

function of differences in the hydraulic properties

of soils among these locations. Our study was on

Ultisols, which exhibit sharp downward increases

in clay content and decreases in macroporosity

(Elsenbeer 2001) and therefore relatively lower

hydraulic conductivity (Zimmermann and others

2006). These other studies were on Oxisols, where

deep and potentially rapid flowpaths leading to

groundwater may predominate (Elsenbeer 2001).

Differences in the location and magnitude of N

cycling in tropical watersheds with soils of different

hydraulic properties have been documented at the

Luquillo Experimental Forest in Puerto Rico

(McDowell and others 1992). At Luquillo, NO3-

concentrations in upland groundwater were lower

under an Ultisol, with an impeding layer, than

under an Inceptisol, where subsoil was dominated

by sands of high hydraulic conductivity. This con-

trast, though currently based on few sites, suggests

that in the Amazon, unsaturated vadose zones

could be a major site of NO3- removal in forests on

soils with low hydraulic conductivities, whereas

soils with rapid water infiltration allow a larger

proportion of NO3- from soil solution to reach the

groundwater.

It is also possible that the observed decline in

NO3- from soil solution to groundwater could be

caused by retention of NO3- on anion exchange

sites deep in the soils. This has been reported in

tropical locations with volcanic soils, such as the

Caribbean lowlands of Costa Rica (Matson and

others 1987), where this process accounted for 12%

of N tracer additions. We did not directly measure

NO3- retention deep in the soil profile in our

watersheds. However, several basic features of these

Ultisols suggest that the presence of the significant

anion exchange capacity needed to promote

retention of NO3- on forest soil exchange sites is

unlikely. These features include shallow contact

with bedrock and relatively high pH (5.4), signifi-

cant cation exchange capacity (5.8 cmolc kg-1) and

73% base saturation in the surface (0–10 cm)

horizon (Neill and others 1997a).

Patterns of N fluxes in our watersheds show

some fundamental differences compared with pat-

terns in the temperate zone where most of our

understanding of watershed-level N fluxes has

been developed. First, high NO3- fluxes through

soils in forest do not lead to high groundwater N

concentrations and large stream water losses from

hydrologic flowpaths. Second, loss or retention of

NO3- in forest occurs largely between soil solution

and groundwater, not in riparian zones. Third,

conversion of forest to typical unfertilized Amazon


pasture agriculture does not cause increases in

watershed N inputs and stream water N losses, but

instead results in lower N fluxes through soils and

into groundwater and streams. This contrasts with

the predominant pattern in the temperate zone,

where watersheds with higher proportions of agri-

cultural land typically receive large N inputs and

have higher riverine N exports (Howarth and others

1996; Boyer and others 2002). This could change as

the sources of anthropogenic N deposition, such as

N oxides from fossil fuel combustion and N fertil-

ization in intensive agriculture expand in the tropics

(Matson and others 1999). Understanding the

locations and mechanisms of N retention across

differences in tropical soils and across upland and

riparian soilscapes will help to understand

watershed N response to these potential future

changes in land use and land management.

ACKNOWLEDGEMENTS

We thank Alexandra A. Montebelo, Gustavo G.

Baldi, and Maıra O. Bezerra at the Centro de Energia

Nuclear na Agricultura in Piracicaba for conducting

the chemical analyses. Shelby Hayhoe, Sonya

Remington, Mathew Shamey, Tobias Vetter, and

Lisa Werther helped during the field work phase of

this study. The subject editor and two anonymous

reviewers provided helpful criticism and insights

that greatly improved the final product. Support for

this study was provided by the US National Science

Foundation (DEB-0315656), the NASA LBA Pro-

gram (NCC5-690) and by grants from Brazilian

agencies FAPESP (# 03/13172-2) and CNPq

(# 420199/2005-5). Special thanks to the Schmitz

family for allowing us to work on their land.

REFERENCES

Achard F, Eva HD, Stibig HJ, Mayaux P, Gallego J, Richards T,

Malingreau JP. 2002. Determination of deforestation rates of

the world’s humid tropical forests. Science 297:999–1002.

Boyer EW, Goodale CL, Jaworski NA, Howarth RW. 2002.

Anthropogenic nitrogen sources and relationships to riverine

nitrogen export in the northeastern U.S.A. Biogeochemistry

57–58:137–69.

Brandes J, McClain M, Pimentel T. 1996. 15N evidence for the

origin and cycling of inorganic nitrogen in a small Amazonian

catchment. Biogeochemistry 34:45.

Chaves JE, Neill C, Germer S, Gouveia Neto S, Krusche AV,

Elsenbeer H. 2008. Land management impacts on runoff sources

in small Amazon watersheds. Hydrol Process 22:1766–75.

Crawley MJ. 2002. Statistical computing: an introduction to data

analysis using S-Plus. Chichester: Wiley.

Davidson EA, Keller M, Erickson HE, Verchot LV, Veldkamp E.

2000. Testing a conceptual model of soil emissions of nitrous

and nitric oxides. Bioscience 50:667–80.

Davidson EA, de Carvalho CJR, Figueira AM, Ishida FY, Ometto

JPHB, Nardoto GB, Saba RT, Hayashi SN, Leal EC, Vieira ICG,

Martinelli LA. 2007. Recuperation of nitrogen cycling in

Amazonian forests following agricultural abandonment. Nat-

ure 447:995–8.

Denslow JS, Vitousek PM, Schultz JC. 1987. Bioassays of nutrient

limitation in a tropical rain forest soil. Oecologia 74:370–6.

Elsenbeer H. 2001. Hydrologic flowpaths in tropical rainforest

soilscapes—a review. Hydrol Process 15:1751–9.

Fearnside PM. 2005. Deforestation in Brazilian Amazonia: his-

tory, rates, and consequences. Conserv Biol 19:680–8.

Finzi AC, Norby RJ, Calfapietra C, Gallet-Budynek A, Gielen B,

Holmes WE, Hoosbeek MR, Iversen CM, Jackson RB, Kubiske

ME, Ledford J, Liberloo M, Oren R, Polle A, Pritchard S, Zak

DR, Schlesinger WH, Ceulemans R. 2007. Increases in nitro-

gen uptake rather than nitrogen-use efficiency support higher

rates of temperate forest productivity under elevated CO2.

Proc Natl Acad Sci USA 104:14014–19.

Garcia-Montiel D, Melillo JM, Steudler PA, Tian H, Neill C,

Kicklighter D, Feigl BJ, Piccolo MC, Cerri CC. 2004. Emissions

of N2O and CO2 from terra firme forests in Rondonia, Brazil.

Ecol Appl 14:S214–20.

Germer S, Elsenbeer H, Moraes JM. 2006. Throughfall and

temporal trends of rainfall redistribution in an open tropical

rainforest, south-western Amazonia (Rondonia, Brazil). Hy-

drol Earth Syst Sci 10:383–93.

Germer S, Neill C, Krusche A, Neto S, Elsenbeer H. 2007. Sea-

sonal and within-event dynamics of rainfall and throughfall

chemistry in an open tropical rainforest in Rondonia, Brazil.

Biogeochemistry 86:155–74.

Germer S, Neill C, Vetter T, Chaves J, Krusche AV, Elsenbeer H.

2009. Implications of long-term land-use change for the

hydrology and solute budgets of small catchments in

Amazonia. J Hydrol 364:349–63.

Goolsby DA, Battaglin WA. 2001. Long-term changes in con-

centrations and flux of nitrogen in the Mississippi River Basin,

USA. Hydrol Process 15:1209–26.

Hall SJ, Matson PA. 1999. Nitrogen oxide emissions after

nitrogen additions in tropical forests. Nature 400:152–5.

Herbert D, Fownes J. 1995. Phosphorus limitation of forest leaf

area and net primary production on a highly weathered soil.


Houlton BZ, Sigman DM, Hedin LO. 2006. Isotopic evidence for

large gaseous nitrogen losses from tropical rainforests. Proc

Natl Acad Sci USA 103:8745–50.

Howarth RW, Billen G, Swaney D, Townsend A, Jaworski N,

Lajtha K, Downing JA, Elmgren R, Caraco N, Jordan T, Ber-

endse F, Freney J, Kudeyarov V, Murdoch P, Zhao-Liang Z.

1996. Regional nitrogen budgets and riverine N & P fluxes for

the drainages to the North Atlantic Ocean: natural and human

influences. Biogeochemistry 35:75–139.

INPE. 2008. Monitoramento da floresta Amazonica Brasileira

por satelite: Projeto PRODES. Sao Jose dos Campos: Instituto

Nacional de Pesquisas Espacias.

Kabat P, Dolman AJ, Ashby M, Gash JHC, Wright IR, Culf AD,

Calvet JC, Delire C, Noilhan J, Jochum A, Silva Dias MA,

Fisch GA, Santos Alval RC, Nobre A, Prince SD, Steininger M.

1997. The large scale biosphere—atmosphere experiment: use

of integrated modelling for experimental design. Report 145,

Wageningen.

Kirkby M, Callan J, Weyman D, Wood J. 1976. Measurement

and modeling of dynamic contributing areas in very small


catchments. Working Paper No. 167, University of Leeds,

School of Geography, Leeds, p 39.

Leal JWL, Silva GH, dos Santos DB, Texeira W, de Lima MIC,

Fernandes CAC, Pinto A. 1978. Geologia—Folha SC.20 Porto

Velho. Projeto Radambrasil 17:219–355.

Lesack LFW. 1993. Export of nutrients and major ionic solutes

from a rain forest catchment in the central Amazon basin.

Water Resour Res 29:743–58.

Markewitz D, Davidson EA, Moutinho P, Nepstad D. 2004.

Nutrient loss and redistribution after forest clearing on a

highly weathered soil in Amazonia. Ecol Appl 14:177–99.

Martinelli LA, Piccolo MC, Townsend AR, Vitousek PM, Cuevas

E, McDowell W, Robertson GP, Santos OC, Treseder K. 1999.

Nitrogen stable isotopic composition of leaves and soil: tropical

versus temperate forests. Biogeochemistry 46:45.

Matson PA, Vitousek PM, Ewel JJ, Mazzarino MJ, Robertson GP.

1987. Nitrogen transformations following tropical forest fell-

ing and burning on a volcanic soil. Ecology 68:491–502.

Matson PA, McDowell W, Townsend AR, Vitousek PM. 1999.

The globalization of N deposition: ecosystem consequences in

tropical environments. Biogeochemistry 46:67–83.

Matthews E. 1994. Nitrogenous fertilizers: global distribution of

consumption and associated emissions of nitrous oxide and

ammonia. Global Biogeochem Cycles 8:411–40.

Matthews E, Rohweder M, Payne R, Murray S. 2000. Pilot

analysis of global ecosystems: forest ecosystems. Washington,

DC: World Resources Institute.

McClain ME, Richey JE, Pimentel TP. 1994. Groundwater

nitrogen dynamics at the terrestrial-lotic interface of a small

catchment in the Central Amazon basin. Biogeochemistry

27:113–27.

McDowell W, Bowden WB, Asbury CE. 1992. Riparian nitrogen

dynamics in two geomorphologically distinct rain forest

watersheds: subsurface solute patterns. Biogeochemistry

18:53–75.

Mulholland PJ, Helton AM, Poole GC, Hall RO, Hamilton SK,

Peterson BJ, Tank JL, Ashkenas LR, Cooper LW, Dahm CN,

Dodds WK, Findlay SEG, Gregory SV, Grimm NB, Johnson SL,

McDowell WH, Meyer JL, Valett HM, Webster JR, Arango CP,

Beaulieu JJ, Bernot MJ, Burgin AJ, Crenshaw CL, Johnson

LT, Niederlehner BR, O’Brien JM, Potter JD, Sheibley RW,

Sobota DJ, Thomas SM. 2008. Stream denitrification across

biomes and its response to anthropogenic nitrate loading.

Nature 452:202–5.

Neill C, Piccolo MC, Steudler PA, Melillo JM, Feigl BJ, Cerri CC.

1995. Nitrogen dynamics in soils of forests and active pastures

in the western Brazilian Amazon Basin. Soil Biol Biochem

27:1167–75.

Neill C, Melillo JM, Steudler PA, Cerri CC, De Moraes JM, Pic-

colo MC, Brito M. 1997a. Soil carbon and nitrogen stocks

following forest clearing for pasture in the southwestern

Brazilian Amazon. Ecol Appl 7:1216–25.

Neill C, Piccolo MC, Cerri CC, Steudler PA, Melillo JM, Brito M.

1997b. Net nitrogen mineralization and net nitrification rates

in soils following deforestation for pasture across the south-

western Brazilian Amazon Basin landscape. Oecologia 110:

243–52.

Neill C, Deegan L, Thomas S, Cerri CC. 2001. Deforestation for

pasture alters nitrogen and phosphorus in small Amazonian

streams. Ecol Appl 11:1817–28.

Neill C, Steudler P, Garcia-Montiel ADC, Melillo JM, Feigl BJ,

Piccolo MC, Cerri CC. 2005. Rates and controls of nitrous

oxide and nitric oxide emissions following conversion of forest

to pasture in Rondonia. Nutr Cycl Agroecosyst 71:1–15.

Neill C, Piccolo M, Cerri CC, Steudler PA, Melillo JM. 2006. Soil

solution nitrogen losses during clearing of lowland Amazon

forest for pasture. Plant Soil 281:233–45.

Nepstad DC, de Carvalho CR, Davidson EA, Jipp PH, Lefebvre

PA, Negreiros GH, da Silva ED, Stone TA, Trumbore SE, Vieira

S. 1994. The role of deep roots in the hydrological and carbon

cycles of Amazonian forests and pastures. Nature 372:666–9.

Peierls BL, Caraco NF, Pace ML, Cole JJ. 1991. Human influence

on river nitrogen. Nature 350:386.

Peixoto de Melo D, da Costa RC, Natali Filho T. 1978. Geo-

morfologia—Folha SC.20 Porto Velho. Projeto Radambrasil

16:187–250.

Perring MP, Hedin LO, Levin SA, McGroddy M, de Mazancourt

C. 2008. Increased plant growth from nitrogen addition

should conserve phosphorus in terrestrial ecosystems. Proc

Natl Acad Sci USA 105:1971–6.

Pires JM, Prance GT. 1986. The vegetation types of the Brazilian

Amazon. In: Prance GT, Lovejoy TM, Eds. Key environments:

Amazonia. Oxford: Pergamon. p 109–29.

Poszwa A, Dambrine E, Ferry B, Pollier B, Loubet M. 2002. Do

deep tree roots provide nutrients to the tropical rainforest?


R Development Core Team. 2006. R: a language and environ-

ment for statistical computing. Vienna, Austria: R Foundation

for Statistical Computing.

Robertson GP. 1989. Nitrification and denitrification in humid

tropical ecosystems. In: Proctor J, Ed. Mineral nutrients in

tropical forest and savanna ecosystems. Cambridge (MA):

Blackwell Scientific. p 55–70.

Sigman DM, Altabet MA, Michener R, McCorkle DC, Fry B,

Holmes RM. 1997. Natural abundance-level measurements of

the nitrogen isotopic composition of oceanic nitrate: an

adaptation of the ammonia diffusion method. Marine Chem

57:227–42.

Soil Survey Staff. 1999. Soil taxonomy—a basic system of soil

classification for making and interpreting soil surveys.

Washington, DC: US Government Printing Office.

Tiedje JM, Sexstone AJ, Parkin TB, Revsbech NP. 1984. Anaer-

obic processes in soil. Plant Soil 76:197–212.

Verchot LV, Davidson EA, Cattanio JH, Ackerman IL, Erickson

HE, Keller M. 1999. Land use change and biogeochemical

controls of nitrogen oxide emissions from soils in eastern

Amazonia. Global Biogeochem Cycles 13:31–46.

Vitousek PM. 1984. Litterfall, nutrient cycling, and nutrient

limitation in tropical forests. Ecology 65:285–98.

Vitousek PM. 2004. Nutrient cycling and limitation. Princeton

(NJ): Princeton University Press.

Vitousek PM, Farrington H. 1997. Nutrient limitation and soil

development: experimental test of a biogeochemical theory.


Williams MR, Melack JM. 1997. Solute export from forested and

partially deforested catchments in the central Amazon. Bio-

geochemistry 38:67.

Williams MR, Fisher TR, Melack JM. 1997. Solute dynamics in

soil water and groundwater in a central Amazon catchment

undergoing deforestation. Biogeochemistry 38:303.

Zimmermann B, Elsenbeer H, De Moraes JM. 2006. The influ-

ence of land-use changes on soil hydraulic properties: impli-

cations for runoff generation. Forest Ecol Manag 222:29–38.


Documents

Nitrogen Transformations In Flowpaths Leading From Soils to Streams In Amazon Forest and Pasture