environmental, social, and management drivers of soil nutrient mass balances in an extensive andean...

19
Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System S. J. Vanek 1 * and L. E. Drinkwater 2 1 Department of Crop and Soil Sciences, Cornell University, Ithaca, NY 14853, USA; 2 Department of Horticulture, Cornell University, Ithaca, New York 14853, USA ABSTRACT Sustainable nutrient cycling in agroecosystems com- bining grazing and crops has global ramifications for protecting these ecosystems and for the livelihoods they support. We sought to understand environ- mental, management, and social drivers of nutrient management and sustainability in Andean grazing/ crop systems. We assessed the impact of farmer wealth, fields’ proximity to villages, topography, and rangeland net primary productivity (NPP) on mass balances for nitrogen (N), phosphorus (P), and potassium (K) of 43 fields. Wealthier farmers applied greater total amounts (kg) of manure nutrients. However, higher manure application rates (kg ha -1 ) were associated with field proximity and NPP rather than wealth. Manure P inputs in far fields (> 500-m distant) were half those in near fields. Harvest exports increased with manure inputs (P < 0.001) so that balances varied less than either of these flows. Erosion nutrient losses in steeper far fields matched crop exports, and yields declined with increasing field slope (P < 0.001), suggesting that erosion reduces productivity. Balances for P were slightly positive in near and far fields (+2.2 kg P ha -1 y -1 , combined mean) when calculated without erosion, but zero in near fields and negative in far fields with erosion in- cluded (-6.1 kg P ha -1 y -1 in far fields). Near/far differences in both inputs and erosion thus drove P limitation. Crop K exports dominated K balances, which were negative even without accounting for erosion. Modeled intensification scenarios showed that remediating far field deficits would require P addition and erosion reduction. Management nested within environmental constraints (NPP, erosion) rather than socioeconomic status drives soil nutrient sustainability in these agroecosystems. Time-lags between management and long-term degradation are a principal sustainability challenge to farming in these montane grazing/crop agroecosystems. Key words: nutrient mass balances; Andes; soil erosion; rangeland; mixed cropping systems; time- lags; Bolivia; manure; phosphorus; potassium. INTRODUCTION In smallholder systems with a mix of grazing and cropland, traditional methods used by low-intensity smallholders have been characterized as sustainable, relying on primary productivity in grazing areas and transformation of soil nutrients into useable form by fallow vegetation and animals (Powell and others Received 19 February 2013; accepted 23 July 2013; published online 21 August 2013 Electronic supplementary material: The online version of this article (doi:10.1007/s10021-013-9699-3) contains supplementary material, which is available to authorized users. Author Contributions: SJV and LED conceived the study and wrote the paper. SJV conducted research and analyzed data. *Corresponding author; e-mail: [email protected] Ecosystems (2013) 16: 1517–1535 DOI: 10.1007/s10021-013-9699-3 Ó 2013 Springer Science+Business Media New York 1517

Upload: l-e-drinkwater

Post on 07-Feb-2017

213 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Environmental, Social,and Management Drivers of Soil

Nutrient Mass Balances in anExtensive Andean Cropping System

S. J. Vanek1* and L. E. Drinkwater2

1Department of Crop and Soil Sciences, Cornell University, Ithaca, NY 14853, USA; 2Department of Horticulture, Cornell University,

Ithaca, New York 14853, USA

ABSTRACT

Sustainable nutrient cycling in agroecosystems com-

bining grazing and crops has global ramifications for

protecting these ecosystems and for the livelihoods

they support. We sought to understand environ-

mental, management, and social drivers of nutrient

management and sustainability in Andean grazing/

crop systems. We assessed the impact of farmer

wealth, fields’ proximity to villages, topography, and

rangeland net primary productivity (NPP) on mass

balances for nitrogen (N), phosphorus (P), and

potassium (K) of 43 fields. Wealthier farmers applied

greater total amounts (kg) of manure nutrients.

However, higher manure application rates (kg ha-1)

were associated with field proximity and NPP rather

than wealth. Manure P inputs in far fields (> 500-m

distant) were half those in near fields. Harvest exports

increased with manure inputs (P < 0.001) so that

balances varied less than either of these flows. Erosion

nutrient losses in steeper far fields matched crop

exports, and yields declined with increasing field

slope (P < 0.001), suggesting that erosion reduces

productivity. Balances for P were slightly positive in

near and far fields (+2.2 kg P ha-1 y-1, combined

mean) when calculated without erosion, but zero in

near fields and negative in far fields with erosion in-

cluded (-6.1 kg P ha-1 y-1 in far fields). Near/far

differences in both inputs and erosion thus drove P

limitation. Crop K exports dominated K balances,

which were negative even without accounting for

erosion. Modeled intensification scenarios showed

that remediating far field deficits would require P

addition and erosion reduction. Management nested

within environmental constraints (NPP, erosion)

rather than socioeconomic status drives soil nutrient

sustainability in these agroecosystems. Time-lags

between management and long-term degradation are

a principal sustainability challenge to farming in these

montane grazing/crop agroecosystems.

Key words: nutrient mass balances; Andes; soil

erosion; rangeland; mixed cropping systems; time-

lags; Bolivia; manure; phosphorus; potassium.

INTRODUCTION

In smallholder systems with a mix of grazing and

cropland, traditional methods used by low-intensity

smallholders have been characterized as sustainable,

relying on primary productivity in grazing areas and

transformation of soil nutrients into useable form by

fallow vegetation and animals (Powell and others

Received 19 February 2013; accepted 23 July 2013;

published online 21 August 2013

Electronic supplementary material: The online version of this article

(doi:10.1007/s10021-013-9699-3) contains supplementary material,

which is available to authorized users.

Author Contributions: SJV and LED conceived the study and wrote the

paper. SJV conducted research and analyzed data.

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

Ecosystems (2013) 16: 1517–1535DOI: 10.1007/s10021-013-9699-3

� 2013 Springer Science+Business Media New York

1517

Page 2: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

1996; Pestalozzi 2000). However, traditional systems

face pressure from increasing human and animal

populations, breakdown of communal management

systems, and external market incentives (Mayer

1979; Lightfoot and Noble 2001; Pendleton and

Howe 2002; Baijukya and others 2005; Pacheco

2009). Frequently, intensification of these systems

increases erosion and truncates the regeneration of

soil fertility,1 so that fallows and grazing-derived

manure fail to replenish nutrient deficits. Degrada-

tion and declining productivity create a downward

spiral that undermines food security for smallhold-

ers. However, other research suggests that innova-

tion often averts degradation within smallholder

systems via practices such as agroforestry, fertilizer

and manure additions, efficient residue recycling,

and sustainable-grazing management, often sup-

ported by markets, regional policies, and local

knowledge and governance (Mortimore and Harris

2005; Schechambo and others 1999; Scherr 2000).

Mixed cropping systems utilizing rangelands are

an important setting in which to analyze potential

degradation by smallholders. The strategy of sup-

porting intensively managed crops by integrating

rangeland and animals is found worldwide from

uplands in Mexico and India to wide swaths of

African savanna and mountain regions (Powell and

others 1996; Saberwal 1996; Elias and others 1998;

Aganga and Mosimanyana 2001; Schlecht and

others 2004; Arriaga-Jordan and others 2005; Ruf-

ino and others 2011). Rangelands with significant

human habitation cover over 20% of the earth’s

non-wild land area (Ellis and others 2010). These

form an important grazing–cropping ‘‘anthrome’’

on marginal landscapes that are not suited for

intensive agriculture due to poor soils, steep

topography, or limited rainfall. Such systems con-

trast markedly with densely populated, intensive

smallholder systems that were the focus of the

Green Revolution, where confined livestock do not

interact with rangelands (Thorne and Tanner 2002).

Nutrient mass balances can be used to understand

the potential drivers of nutrient depletion in these

systems. Mass balances are an important tool of

ecosystem ecology whose use has expanded to study

managed ecosystems (Smaling and others 1996;

Baker and others 2001). As applied to agricultural

management, balances quantify net gain or loss of

nutrients over time for fields and other land units

and can identify which nutrients (for example, N, P)

or processes (for example, manure inputs, crop

export) most limit agroecosystem productivity,

suggesting leverage for innovations in practices

(Berry and others 2003; Baijukya and others 2005).

Mass balances have also identified crop rotations,

land uses, or spatial components of systems that are

most vulnerable to nutrient depletion (Elias and

others 1998; Wortmann and Kaizzi 1998; Lesschen

and others 2007), or suggested that relative wealth

or poverty drive degradation (Elias and others 1998;

Nkonya and others 2005; Yirga and Hassan 2006;

Cobo and others 2009).

In this study, we used nutrient balances to

evaluate the relative importance of environmental,

social, and management drivers on the sustainability

of soil nutrient stocks in Andean agroecosystems.

These are extensive, mixed crop–livestock systems

managed by smallholders with a shared land base

for grazing, combined with some degree of social

stratification. Households rely mainly on forage

production in extensive rangelands and food pro-

duction in crop fields dispersed within mountain

terrain at varying distances from a central village.

We hypothesized that the NPP of community

territories and soil erodibility from varying slopes of

fields would be environmental drivers of these bal-

ances. We predicted that NPP (kg ha-1) within a

community’s boundary would correlate positively

to manure nutrient inputs to fields because animals

graze freely on surrounding non-cropped lands.

Because of the steep topography we also expected

that soil erosion related to differing field slopes

would influence mass balances, with flatter fields

in areas near communities suffering smaller erosion

losses. As a social driver, we expected that farmer

wealth level indicated by land and animal owner-

ship would affect field nutrient balances, because

livestock is a means of capturing manure nutrients

from rangeland for application to cropland. We

expected that wealthier farmers with more animals

would apply both a greater total amount and at a

higher per-hectare rate on farmed fields. The cor-

responding lower manuring rates among poorer

farmers would create more negative balances and

thus greater degradation potential. Manuring rate

was expected to be the important management dri-

ver governing soil nutrient stocks, expressing social

factors of differential access to rangeland manure

nutrients, and also knowledge and management

strategies particular to each farm. We expected that

higher manuring rates on near versus far fields

would create gradients in soil nutrients and pro-

ductive capacity by creating more positive nutrient

1 We use ‘‘soil fertility’’ to refer broadly to soil produc-tive capacity which reflects not only nutrient availabilitybut also soil physical and chemical characteristics such aspore structure, water holding capacity, cation exchangecapacity, and pH which all impact plant productivity.

1518 S. J. Vanek and L. E. Drinkwater

Page 3: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

balances on near fields. We also expected mass

balances to show whether N, P, or K most con-

strains crop productivity in these systems. Lastly,

we used our data to create scenarios that model

management alternatives into the future. We

compared a status quo rotation scenario to balances

resulting from management changes that would be

expected to slow or accelerate degradation.

MATERIALS AND METHODS

Research Area

Field-scale nutrient balances were constructed for

43 farms in northern Potosi, Bolivia, an isolated

region with rainfed potato-cereal-pulse rotations

set within areas of rangeland and fallows (Fig-

ure 1). Livestock are a versatile economic asset and

a manure source for these farms (McCorkle 1990).

Sheep predominate at higher elevations and goats

at lower elevations, with some llamas at higher

elevations and bullocks throughout for traction.

Animals are grazed on rangeland year-round

(predominantly C3 grasses at high elevations and

shrubs at lower elevations; CIF-UMSS 2013), with

seasonal feeding of crop stubble and cultivated

forage oats. Average annual precipitation of

650 mm occurs between October and March (FAO

2010). Mean temperatures during this cropping

season range from 9.5 to 18.0�C varying by field

elevation (between 2,700 and 4,200 m asl). Agri-

cultural management regimes differ based on

proximity to villages: far fields in contiguous blocks

under community sectoral fallows undergo syn-

chronized crop rotations (Pestalozzi 2000). Erosion

prevention measures include stone retention walls

and shrub live barriers that reduce but do not

eliminate substantial field slopes, in contrast to

other areas in the central Andes where crops are

grown on terraces. Nearer to households, field

rotations are determined by each household with-

out synchronization. The dominant soil nutrient

input is manure derived from rangeland grazing

and feeding of crop residues. Crop rotations gen-

erally begin with potato, which receives the bulk of

manure in the rotation. A 2nd year with maize,

wheat, barley, fava beans, peas, or oca (Oxalis tu-

berosum) receives sporadic manuring. A final rota-

tion year without manure follows, usually a cereal,

forage oats, or Andean lupine (tarwi locally, Lupinus

mutabilis Sweet). On average, legumes occur once

per 25 years in far fields and once per 7 years in

near fields (unpublished survey data). Maize is

grown below 3,600 m elevation, sometimes in

continuous cropping alternating with wheat.

Wealth Ranking and Land and AnimalAssets of Farmers

Data on household asset levels were drawn from a

local project baseline study (Neighbors 2006).

Community members anonymously ranked peers

into three groups as those with ‘‘most’’, ‘‘less’’, and

‘‘very little’’ and estimated typical asset levels of

these groups in animal numbers and cropland area.

Farmers with fields sampled for nutrient balances

provided the number of animals they owned and

kg of seed planted for each crop. Total cropped area

for the farm was calculated by dividing these seed

amounts by local crop seeding rates (kg ha-1) given

by experienced farmers and project staff.

Remotely-Sensed NPP and SlopeGradient Data Within CommunityBoundaries

Annual NPP (g C m-2) on a 1,000-m grid, averaged

over the years 2000–2006 was downloaded from an

online database (MODIS, Zhao and others 2008).

Community boundaries were geo-referenced and

transferred to the NPP layer. Mean NPP was cal-

culated for the five to ten 1-km2 pixels covering

each community. To examine how village and field

locations influenced soil erosion potential, a digital

elevation model (DEM) at 30-m resolution (ASTER,

Figure 1. Study location, with latitude/longitude coor-

dinates.

Drivers of Andean Soil Nutrient Mass Balances 1519

Page 4: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

ERSDAC 2007) was used to create a histogram of

slope gradient (degrees slope, binned at 2-degree

intervals) for the entire community and also for

pixels in a 500-m radius of villages. The proportion

of highly erodible land in each community area was

calculated as the percentage of pixels with slope

greater than 21% (12� slope gradient). This slope

threshold represents highly erodible land in the

Revised Universal Soil Loss Equation (RUSLE; Re-

nard and others 1997) and was also chosen to create

the widest separation between communities with

flat and steep topography in a 45-community survey

dataset from another study (Jones 2011).

Field Selection and Overall BalanceApproach

One field per farm was identified in collaboration

with a random sample of farmers, stratified by

anonymous wealth ranking so that fields from all

three wealth levels in all communities were sam-

pled. The final sample contained ten each of

highest and lowest grouped-farmers, and 23 of the

middle group. Balances were calculated as inputs

minus outputs for an entire rotation divided by the

rotation length (giving kg ha-1 y-1). Rotations of

2 years were measured for near fields in the lowest

community, where continuous cropping of maize

and wheat takes place. Six-year rotations with

three cropped and three fallow years occurred in

most other fields, whereas some near fields at

higher elevations had 3-year crop sequences

without fallows. We used 3 years as the fallow

length for far fields based on farmers’ projections of

likely future fallows, and the median fallow length

(3.2 years) from a parallel survey of 297 house-

holds (Jones 2011).

Sampling during two sequential years was stag-

gered to capture 3 years of cropping reflecting

common local crop sequences. Manuring rates and

two sequential-cropped years were measured on

the 43 designated fields, during which all manuring

and most of the crop exports occur. Then, we

estimated cereal and lupine nutrient exports of a

3rd year by sampling a similar set of fields in their

3rd year of production following potatoes and a

2nd-year crop. These cereal fields were in the same

communities, near/far locations, and elevations as

those sampled for 2-year sequences. We fit P

exports from these fields to a random normal dis-

tribution and used draws from this distribution to

estimate P exports for the 3rd year, for fields in

which farmers later reported growing cereals in the

3rd year. To find N and K exports for these fields,

we multiplied these P exports by a similar draw

from N:P and K:P distributions of harvest nutrient

content ratios based on the sampled cereal fields.

When lupine occurred in the 3rd year, qualitative

yield estimates from farmers were converted to

exports of N, P, and K, and fixed N inputs using

local field experiment data (Vanek 2011).

Manure Inputs to Fields

Manure inputs of N, P, and K were measured in

two manuring systems used by farmers. At lower

elevations (2,500–3,400 masl), fields receive two to

four nights of manuring when animals are corralled

at night in fields prior to cropping (‘‘field-corral’’).

At elevations generally above 3,500 masl a second

‘‘fixed-corral’’ system is used: manure accumulates

over an entire year in fixed sleeping pens, and then

is transported to fields for direct application to crops

at seeding.

For the field-corral system, manure was gathered

from three stratified, randomly selected quadrats of

0.25 m2 within the corral after at least two nights of

corralling. Small amounts of soil (< 2%) adhering

to manure was included to maximize capture of

manure and urine nutrients. Manure was weighed

and subsampled for dry matter (DM) calculation

and nutrient analysis (below). Mean DM per

0.25 m2 for three quadrats gave a field-level

nutrient application rate. The nightly rate was

multiplied by the intended number of nights of

manuring to give a total manure input rate.

Our manure rate measurements are most accu-

rate for P inputs, which remain in a particulate

form in manure without gaseous losses. Our field-

corral N measurements may slightly underestimate

manure N inputs: samples were air-dried before

analysis from a moist field state, and ammonia-N in

urine may have entered the soil without being in-

cluded in the sample. Typical ammonia-N fractions

of about 25% of total N for ruminant manure are

an upper bound on this uncertainty (Antil and

others 2009). Urine K may likewise not be captured

by our measurements. We adjusted field-corral

manure K concentrations upward by a factor of

1.96, the ratio of fixed-corral to field-corral K

content of manure in a community that used both

manuring practices (that is, based on the same

animals and rangeland forages and likely to create

similar levels of K excretion in livestock urine). We

conducted a simple sensitivity analysis to test the

impact of imprecision in manure K content (see

results).

Fixed-corral manuring using manure from

household pens was measured late in the dry sea-

son and did not suffer N and K leaching, because

1520 S. J. Vanek and L. E. Drinkwater

Page 5: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

corral manure DM content was always greater than

50%. Small, regularly spaced piles of manure are

placed on fields at planting, so that manuring could

be measured by estimating the weight of 15–20

piles within rectangular control areas. Manure

weight in each rectangle was estimated by weigh-

ing two piles with a calibrated spring scale. Manure

bulk density was then calculated as mean weight

divided by mean volume of the piles (modeled as

cones with rounded tops, using their heights and

average diameters). The weight of remaining piles

in the field area was then found by multiplying the

volume of each pile (measured as for the weighed

piles) by the manure bulk density. Manuring rate

was computed as the total DM of the piles divided

by the rectangle area. A bulked sample was taken

from all piles in the field for nutrient analysis.

Seed Nutrient Inputs

Seeding rates (kg ha-1) were given by local farmers

and field staff: potatoes, 1,700; maize, 80; wheat

and barley, 100; oats, 120; fava beans, 120; and

lupine, 60. Seed nutrient inputs were the product

of these rates by mean crop nutrient contents (be-

low). For oat seed a literature value was used

(NRCS 2010).

Fixed N and Nutrient Deposition

N additions by legume crops (fava beans, lupine)

were estimated as net zero; local experiments with

lupine indicated that the mean shoot N fraction

(69%) equaled the fraction of N fixed (%Ndfa;

Vanek 2011) so that grain and straw N exports

cancelled the N-fixation input (Ndfa). This is con-

servative given higher estimates of %Ndfa for fava

(75%; Ross and others 2008), and higher lupine

%Ndfa for the most P-fertile sites in the area (80%;

Vanek 2011). Field experiments have shown a

16 kg N ha-1 input from lupine residues (Villarroel

and others 1986). Mean Ndfa inputs would however

be small on an annualized basis because legumes

are infrequently grown.

Three-year fallows in sampled fields were as-

signed an N-fixation credit of 4 kg N ha-1 y-1. This

was determined by sampling three fallow fields of

age 2–5 years for which shoot + root N equaled

24 kg ha-1, with %Ndfa of 50% calculated for

endemic range legumes (for example, Trifolium

amabile, unpublished data). We did not account for

additional fixed N left as manure during grazing on

wild legumes in fallows. This flow was likely much

smaller than manure N inputs during cropping.

Deposition was minor and was not included in our

balance calculations (< 0.2 kg ha-1 of N and K per

year; Duncan Fairlie and others 2007; Lesschen and

others 2007).

Crop and Residue Exports

Crops and residues were weighed from three ran-

dom, stratified quadrats in each field (1 m2 for

cereals, 3 m2 for other crops). Harvested weight

was divided by the quadrat area, giving yield

(kg ha-1), and the mean of the three quadrat DM

yields was calculated. For maize, stalks and ears

with grain were weighed separately, and subsam-

ples of two ears and two stalks were taken to

measure DM and nutrient content. Dry subsample

grain and stalk weight allowed calculation of yields

from total ear and stalk weight. Maize stalk biomass

was added to cob and husk DM to calculate residue

export for each quadrat. For broadcast forage oats,

wheat, and barley, mature whole shoots from three

1-m2 quadrats per field were weighed. A random

subsample of 25 stems was dried and threshed to

calculate grain and straw DM yields per quadrat.

Maize and cereal residues were treated as exports

reflecting local practices of threshing and feeding

residues off-field. Crops and weeds were cut at

3-cm height to reflect post-harvest grazing of fields.

Moisture Content and Nutrient Analyses

Crop and manure samples were dried in paper bags

on pavement in strong sunlight (45�C), followed by

oven-drying at 58�C. Tubers were frozen before

drying. Dry weight over field fresh weight gave DM

content. N content was measured in ground sample

DM by combustion (LECO St. Joseph, MI). Total P

and K in crops and manures was determined using

nitric acid digestion followed by P and K analysis by

ICP-AES (Kalra 1998). Nitrogen, P and K flows

were then calculated as crop or manure DM stocks

(kg ha-1) multiplied by their nutrient concentra-

tions.

Soil Erosion, Gaseous, and LeachingLosses

To estimate erosion losses from fields, we calibrated

RUSLE (Renard and others 1997) for 1 year on six

fields growing cereal crops and five fields recently

placed in fallow. Fallow fields contained a mix of

crop residues and vegetation grazed by animals.

Measured erosion rates were regressed to RUSLE’s

topographic factor LS, representing slope length (L)

and gradient (S) from the soil loss equation:

Drivers of Andean Soil Nutrient Mass Balances 1521

Page 6: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Erosion Mg soil ha�1y�1� �

¼ R K LSð ÞC Pr:

Here R represents rainfall erosivity, K is soil

erodibility, C and Pr incorporate the effects of soil

cover and management practices, respectively. We

measured L (m) and S (% slope) on calibration

fields and those used for balances so that soil ero-

sion could be estimated as

Erosionmodeled ¼MlocalLS

with Mlocal the fitted slope of measured erosion rates

against LS (that is, equal to R K C Pr). We measured

erosion on the 11 calibration fields by modifying the

erosion pins method (Haigh 1977; Hudson 1993),

which measures the elevation change of the soil

surface compared to a datum to quantify erosion or

deposition. At each of four field replicates we

measured 15–20 microelevation changes over

1 year. We then multiplied this elevation change by

soil bulk density, corrected for stone content dif-

ferences in the initial and final bulk density sam-

pling cores. On the six calibration fields with cereal

crops, we also corrected for the elevation change of

soil resulting from settling and compaction of soil in

a tilled crop field in addition to that from soil ero-

sion. The detailed methods are given in the online

supplementary information (SI).

To determine N, P, and K erosion losses, the

erosion rate was multiplied by total soil N (com-

bustion) and P (perchloric acid digest), and 10.1

times the exchangeable soil K content (Kexch,

ammonium acetate method), a factor that repre-

sents long-term plant-available K and its relation to

Kexch in published data on 20 soils with similar

climate and mineralogy to those sampled here

(Andrist-Rangel and others 2007; Ogaard and

others 2004; Simonsson and others 2007; see SI for

details).

Nitrogen and K gaseous and leaching losses

were estimated with transfer functions in Less-

chen and others (2007) using N and K inputs, and

area-wide climate and soil data (see SI for detailed

functions).

Scenarios of Alternative Management

To model future impacts of management alterna-

tives, four scenarios of intensification in a far field

were compared (Table 6): (1) A status quo rotation;

(2) shortened fallows with no other changes; (3) a

‘‘legume/P’’ strategy in which shortened fallows

were compensated for by P addition and greater le-

gume use (Kihara and others 2010); (4) ‘‘integrated

intensification’’ combining P addition, legumes, and

erosion reduction (for example, live barriers, erosion

catch ditches). We modeled erosion in these sce-

narios using our RUSLE calibration with L = 15 m

and S = 10% slope. Mean far field manuring inputs

of N, P, and K from the field balances were used. Crop

yields were estimated using Monte Carlo simulation

in which random draws were taken from a 95%

confidence interval of our yield data modeled as a

normal distribution. We assumed that P addition to

forages and green manures would reduce erosion by

10% by enhancing soil cover (Vanek 2011). Under

‘‘integrated intensification’’ we assumed a 50%

reduction in soil erosion. Balances of N, P, and K for

scenarios were composed in an Excel spreadsheet

and replicated 200 times with random draws for crop

yields, with mean and standard deviation calculated

for each scenario.

Statistical Analyses

For data linking wealth ranking to land and animal

tenure, analysis of variance (JMP, SAS Institute,

Cary, NC) tested farmer wealth differences. Anal-

ysis of covariance was used to test categorical and

continuous predictors of manure and crop flows

and nutrient concentrations as well as whole bal-

ances for N, P, and K. Linear regression was used to

test the association between field slope and grain or

tuber yield standardized to the mean yield of each

crop. Pairwise comparisons were made using two-

sided t tests.

RESULTS

Nutrient Input Rates

The relation between farmer wealth and manure

inputs showed several trends. First, farmers ranked

by peers as those with ‘‘very little’’ had fewer

animals and less cropped land than those ranked in

wealthier groups (Table 1). However, manure

application rates (kg ha-1) for N and P did not

differ significantly among any wealth groups or in

proportion to animals owned. Relative poverty may

therefore not drive nutrient depletion of cropland

in this system. Both the number of animals and

cropped acreage increased with wealth level, so

that farmers with fewer animals also had less land

to manure, which would explain similar manuring

rates. However, the ratio of animals:land area also

increased with land area, and it was puzzling that

the wealthiest farmers with the highest ratios of

animals:land did not manure at higher rates. In

fact, none of the balance terms or whole balances

differed with wealth, so we combined wealth

groups in all other analyses. However, similar

1522 S. J. Vanek and L. E. Drinkwater

Page 7: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

manuring rates across wealth levels implied that

wealthier farmers with more land applied greater

total amounts of manure nutrients than the poorest

farmers.

Meanwhile, communities situated in environ-

ments with greater NPP had larger application rates

of N and P via manure (Figure 2). The highest-NPP

community had greater amounts of tree and shrub

browse for livestock, whereas the lowest NPP

community presented obvious rangeland degrada-

tion, corroborating the relation between lower re-

motely-sensed NPP and reduced access to animal

forage and manure. Manure N and P concentra-

tions were also greater in manures from higher-

NPP communities (Figure 2; Table 2).

Nitrogen inputs derived directly from biologically

fixed N (Ndfa) through legume crops was dwarfed

by manure N. Legume Ndfa did not differ between

near and far fields, and was approximately five

times smaller than manure N in far fields, and 20

times smaller in near fields due to the large differ-

ences in near/far manuring rates (Table 3).

Uncertainties in our estimates of Ndfa would be

unlikely to alter this comparison given these large

differences. However, Ndfa from rangeland forages

grazed by livestock is likely an important source of

N that is aggregated into the manure N flows.

Crop Yields and Nutrient Exports: FieldProximity and Rotation SequenceImpacts

Crop yields in sampled fields were highly variable

due to a variety of factors including soil nutrient

availability. Fresh potato yields varied tenfold, from

3 to over 30 Mg ha-1, and maize grain varied

100-fold, from less than 100 kg ha-1 to 5 Mg ha-1.

Biotic and abiotic stresses certainly played a role in

this yield variation, as indicated by collaborating

farmers. However, both nutrient concentrations

and total nutrient flows in exported crops suggested

that differences in soil nutrient levels between near

and far fields contributed to the trend for lower

yields in far fields as well as the steep drop in cereal

yields grown 3 years after manure was applied.

Averaged across potato, cereal, and maize crop

fields and rotation year, yields were 40% higher in

near than far fields (P = 0.0012, n = 111)

Crop nutrient contents also indicate that soil

nutrient availability was greater in near than far

fields. Potato tuber P and K and cereal grain N

concentrations were all higher in near than far

fields (Table 4), and when averaged across all

crops, P and K concentrations in crops were higher

in near than in far fields (P < 0.05, n = 151). Tab

le1.

Com

mu

nit

ySoci

alR

an

kin

gas

aPre

dic

tor

of

Liv

est

ock

Ten

ure

,C

ropped

an

dM

an

ure

dA

rea,M

an

uri

ng

Rate

,an

dTota

lM

an

ure

Nu

trie

nts

Applied

Wealt

hra

nk

ing

from

foca

lgro

up

act

ivit

y

Nu

mb

er

of

an

imals

Tota

lcr

op

ped

lan

dare

a,

ha

Lan

dare

ain

pota

toes/

maiz

e

(for

man

uri

ng)

Mean

man

uri

ng

rate

,

kg

ha

-1

y-

1T

ota

ln

utr

ien

tap

pli

ed

per

year,

kg,

est

imate

d

NP

NP

Most

72a

1.8

3a

0.6

0a

130

32

52a

12a

Less

56a

1.3

0b

0.4

6a

119

28

47a

11a

Very

litt

le23b

0.6

9c

0.2

3b

123

28

23b

5b

Sig

nifi

can

ceP

<0.0

01**

*P

<0.0

01**

*P

<0.0

01**

*n

sn

sP

<0.0

1**

P<

0.0

1**

Tot

al

Nan

dP

appli

edare

calc

ula

ted

ona

farm

basi

sm

ult

iply

ing

man

ure

appli

cati

onra

teby

the

repor

ted

are

ain

maiz

ean

dpot

ato

esth

at

are

pot

enti

al

crop

sfo

rm

an

ure

appli

cati

on.

Soc

ial

ran

kin

gdata

wer

epart

ofa

pro

ject

base

lin

eev

alu

ati

onpro

vided

by

alo

cal

non

-gov

ern

men

tal

orga

niz

ati

on.

Drivers of Andean Soil Nutrient Mass Balances 1523

Page 8: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Taken together, differences in yields and nutri-

ent concentrations resulted in larger nutrient ex-

port amounts from near fields compared to far

fields; reflecting the greater nutrient additions

through manure applications (Tables 3, 5). Maize

P and cereal N, P, and K exports declined as the

time since manure was added lengthened from

two to 3 years and nutrients derived from manure

were removed by crop harvests and lost through

erosion (Table 5). Contrasts in P nutrient exports

between near and far fields and over the rotation

sequence were especially notable suggesting that

manure additions were a crucial source of P in

these agroecosystems.

Erosion Losses: Strong Driver of Near/FarField Differences

Erosion rates measured on six cereal and six fallow

fields ranged from 35 to 169 and 3 to 121 Mg ha-1,

respectively (Figure 3). Mean erosion on crop fields

was roughly three times that on fallow fields,

controlling for the effect of slope. Erosion flows of

N, P, and 10.1 Kexch were respectively 12.4, 8.0,

and 11.4 times the RUSLE LS factor on cropped

cereal fields and 3.9, 2.0, and 4.3 times the LS

factor on fallow fields, which represent regression

slopes fit to the erosion data in Figure 3.

Slope data from DEMs showed that areas near

villages were flatter on average than areas far from

the village (> 500-m distant): only 25% of far

areas versus 42% of near areas had slopes less than

12% (paired t test, P < 0.05, n = 6; based on his-

tograms of slope data for 30-m elevation pixels).

This was likely the basis for near/far differences in

sampled fields’ slopes. The range of soil slopes was

similar in near and far fields (0–33 and 3–34%,

respectively) but the distribution of near field

slopes was skewed towards flatness: 75% of near

fields, but only 41% of far fields, had slopes less

than 13%. As a result, far fields were steeper on

average than near fields (mean slopes of far vs.

near, 17.1� vs. 9.3�, t test P < 0.01, n = 43), and

modeled soil nutrient losses were greater in far

fields (Table 3). Yields of crops were negatively

correlated to field slope (Figure 4) consistent with

erosion nutrient exports on steeper fields limiting

crop productivity.

Mass Balances: Near versus Far FieldDifferences and Pervasive K Deficits

Manuring rates combined with erosion differences

created dramatic near versus far field differences in

nutrient balances. Manuring rates annualized

across the rotation were about twice as large in

near fields (< 500 m to dwellings) as in more

distant fields (Table 3). These annualized rates in

far fields are lower partly because they divide

manuring by more years of fallow in far fields.

However, even when we calculated manuring rates

on a cropped-year basis, rates were lower in far

compared to near fields (39 vs. 24 kg P ha1 y-1,

respectively, t = 13.8, P < 0.001, n = 43).

Meanwhile, higher erosion rates in far fields

meant that soil erosion matched crop harvest

exports in far field balances, accentuating the N and

P deficits (Table 3). Erosion and manuring rates

rather than distance from the community per se

drove these negative balances. Thus the steepest far

Figure 2. Dependence on remotely sensed community

NPP of: A manure N and P concentrations and B phos-

phorus application rate in manure.

1524 S. J. Vanek and L. E. Drinkwater

Page 9: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

fields had the most negative P balances, whereas

flatter far fields had lower erosion rates and P

balances intermediate between steep far fields and

flat, well-manured near fields. Balances for N and P

calculated without erosion were positive in near

fields, suggesting that erosion reduction in these fields

would ease N and P constraints to crop productivity.

Crop yield exports were positively correlated to

manure inputs, reinforcing this finding (Table 3).

Meanwhile, negative K balances, even without

erosion, resulted from substantial K exports in

potatoes and maize and cereal residues (Table 3).

In these balances uncertainties were larger for K

than for P inputs due to uncertainty regarding K

content of manure from corralling on fields. How-

ever, K deficits persisted when field-corral fields

were removed from the analysis, and even when

both manuring strategies were assigned the higher

K content from fixed corral manure (data not

shown). Crop exports therefore tend to produce K

deficits in these rotations regardless of uncertainties

in K inputs.

Scenarios for Reversing the Declineof Soil Nutrient Stocks

Modeled scenarios for intensification in nutrient

management demonstrated that increased pressure

on grazed rangelands would result from shortened

fallow lengths, and showed benefits from erosion

reduction and a legume/P strategy. The status quo

scenario used crop sequences and manuring rates

from this study, so that the resulting N and P def-

icits matched those in sampled far fields (Table 3).

By comparison, the shortened fallows scenario re-

tained these deficits but required 33% more man-

ure (Table 6), threatening rangeland forage

production and the ability to sustain present man-

ure input rates, as suggested by the correlation

between rangeland NPP and manure inputs (Fig-

ure 2). The legume/P and ‘‘integrated intensifica-

tion’’ scenario increased both N and P stocks

through erosion reduction, P from rock phosphate

(RP) addition and N through increased legume

cultivation (Table 6). In the legume/P scenario,

legume N from green manures prior to potatoes

allowed for the reduction by half of manure

applications, so that manure use did not exceed the

status quo. Oat/vetch forages in this scenario also

spared N from soil export, and substantial RP

additions in this scenario were intended to reha-

bilitate soil P in depleted outfields. In the ‘‘inte-

grated intensification’’ scenario, erosion reduction

doubled the positive impacts of the legume/P

strategy on soil N and P. Interestingly, addition of

RP and legume N in these latter two scenarios

exacerbated K deficits because manure K was re-

duced relative to crop K export.

DISCUSSION

Our results demonstrate that social and manage-

ment factors are nested within environmental

constraints in this Andean mixed rangeland/crop-

ping system, so that management and environment

together are the strongest determinants of soil

nutrient sustainability. Negative balances in far

fields are indicative of an unsustainable trajectory,

Table 2. Regression Slopes of N and P Concentration and Application Rate on MODIS Remotely-sensed NPPand Mean N and P Concentrations of Manure from Two Different Manuring Strategies

Regression coefficients of nutrient content and application rates to mean NPP (2000–2006)

Nutrient in manure N P K Units of coefficient

Coefficient and significance

Nutrient content of manure 4.1*** 0.04** 12.6*** g kg-1 nutrient content per

Mg C ha-1 y-1 NPP

Nutrient application rate

(in application year)

63.8* 14.0* 137*** kg ha-1 nutrient per

Mg C ha-1 y-1 NPP

Nutrient concentrations in manure from different manuring strategies

N content (g N kg-1) P content (g P kg-1) K content (g K kg-1)

Fixed household pen (fixed-corral; n = 44) 16.9ns (0.4) 4.4ns (0.4) 16.9* (0.8)

Animals corralled on field (field-corral; n = 18) 16.5ns (0.6) 3.9ns (0.6) 9.2* (1.2)

Significance of differences by two-sided t test: *P < 0.05; **P < 0.01; ***P < 0.001.

Drivers of Andean Soil Nutrient Mass Balances 1525

Page 10: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Tab

le3.

Nu

trie

nt

Bala

nce

Term

san

dB

ala

nce

sfo

rN

,P,

an

dK

inN

ear

an

dFar

Fie

lds

Bala

nce

term

sN

(kg

Nh

a-

1y

-1)

P(k

gP

ha

-1

y-

1)

K(k

gK

ha

-1

y-

1)

Near

(n=

21)

Far

(n=

22)

Sig

.N

ear

(n=

21)

Far

(n=

22)

Sig

.N

ear

(n=

21)

Far

(n=

22)

Sig

.

Man

ure

in36.3

(4.3

)17.3

(2.0

)**

*10.6

(1.3

)4.2

(0.7

)**

*31.0

(5.1

)10.6

(1.5

)**

*

Nfi

xati

on

in1.3

(0.7

)3.2

(1.0

)n

s–

––

Cro

pexport

ou

t30.8

(2.7

)12.4

(1.3

)**

*7.6

(0.6

)2.6

(0.4

)**

*53.0

(5.6

)23.1

(2.9

)**

*

Ero

sion

ou

t3.2

(0.9

)9.0

(1.7

)**

2.0

(0.9

)5.5

(1.7

)**

3.0

(0.8

)8.8

(1.7

)**

Leach

ing

an

dgase

ou

slo

sses

ou

t112.0

(1.6

)6.3

(0.8

)**

*–

–5.4

(0.9

)1.6

(0.8

)**

*

Corr

ela

tion

of

crop

export

sto

man

ure

inpu

ts(n

ear,

far

com

bin

ed)

R=

0.3

2,

n=

43

*R

=0.4

2,

n=

43

**R

=0.4

1,

n=

43

**

Wh

ole

bala

nce

san

dst

ati

stic

al

com

pari

son

toze

ro

Bala

nce

wit

hle

ach

ing/g

ase

ou

slo

sses,

wit

hou

tero

sion

-4.5

(2.8

)0

-1.4

(2.7

)0

ns

3.0

(1.1

)>

01.5

(1.1

)>

0n

s-

23.1

(4.9

)<

0-

9.4

(3.6

)<

0n

s

Bala

nce

wit

hero

sion

,n

ole

ach

ing/

gase

ou

slo

sses

8.3

(3.8

)>

0-

3.8

(3.3

)0

*-

0.3

(1.8

)0

-6.1

(1.4

)<

0*

-27.7

(6.3

)<

0-

22.7

(6.2

)<

0n

s

Bala

nce

wit

hlo

sses

an

dero

sion

-5.8

(3.8

)0

-10.3

(3.6

)<

0n

s-

0.3

(1.8

)0

-6.1

(1.4

)<

0*

-32.7

(6.0

)<

0-

24.3

(5.9

)<

0n

s

1E

stim

ate

du

sin

gtr

an

sfer

fun

ctio

ns

(Les

sch

enan

dot

her

s2007).

All

figu

res

are

an

nu

ali

zed

acr

oss

the

len

gth

ofa

rota

tion

oftw

o,th

ree,

or6

years

.N

ear

fiel

ds

are

loca

ted

less

than

500-m

dis

tan

tfr

omfa

rmer

dw

elli

ngs

,w

her

eas

far

fiel

ds

are

grea

ter

than

500-m

dis

tan

t.T

oil

lust

rate

the

impact

sof

dif

fere

nt

loss

path

ways

,bala

nce

sare

give

nw

ith

out

eros

ion

an

dw

ith

Nan

dK

leach

ing

an

dga

slo

sses

,w

ith

eros

ion

an

dw

ith

out

thes

elo

sses

,an

dw

ith

all

loss

esan

der

osio

n.

Bala

nce

dif

fere

nce

sfr

omze

row

ere

ass

esse

dw

ith

atw

o-si

ded

tte

stat

P=

0.0

5co

nfiden

ce.

Sig

nifi

can

ceis

als

osh

own

for

the

dif

fere

nce

bet

wee

nn

ear

an

dfa

rfiel

ds:

*P

<0.0

5;

**P

<0.0

1;

***P

<0.0

01.

1526 S. J. Vanek and L. E. Drinkwater

Page 11: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Tab

le4.

Nu

trie

nt

Con

cen

trati

on

sof

Cro

ps

an

dR

esi

du

es

inR

ota

tion

al

Sequ

en

ces

an

dN

ear

vers

us

Far

Fie

lds

Cro

pan

dfr

act

ion

Year

of

rota

tion

Pvalu

efo

r

dif

fere

nce

Pro

xim

ity

tovil

lage

Pvalu

efo

r

dif

fere

nce

N(fi

eld

s)

On

e

(man

ure

ap

pli

ed

)

Tw

oT

hre

eN

ear

Far

Nco

nce

ntr

ati

on

gN

kg

-1

Pota

totu

bers

9.8

(0.2

)–

––

10.1

(0.4

)9.8

(0.3

)n

s63

Maiz

egra

in13.1

(1.0

)10.9

(0.9

)–

0.0

9n

s11.0

(0.7

)13.1

(1.4

)n

s21

Maiz

est

over

3.9

(0.3

)3.6

(0.2

)–

ns

3.8

(0.2

)3.7

(0.4

)n

s21

Cere

als

gra

in–

16.2

(0.4

)16.6

(0.9

)n

s17.7

(0.7

)15.1

(0.5

)0.0

03**

23

Cere

als

stra

w–

2.9

(0.2

)2.7

(0.3

)n

s3.1

(0.3

)2.5

(0.2

)0.0

8n

s23

Fora

ge

oats

–4.5

(0.6

)6.3

(0.8

)0.0

8n

s6.1

(1.0

)4.8

(0.5

)n

s7

Pco

nce

ntr

ati

on

gP

kg

-1

Pota

totu

bers

1.7

(0.0

4)

––

–1.8

(0.1

)1.6

(0.1

)0.0

08**

63

Maiz

egra

in3.7

(0.2

)3.0

(0.2

)–

0.0

02**

3.4

(0.1

)3.2

(0.2

)n

s21

Maiz

est

over

1.6

(0.3

)1.1

(0.2

)–

ns

1.6

(0.2

)1.1

(0.4

)n

s21

Cere

als

gra

in–

3.5

(0.2

)3.4

(0.4

)n

s3.5

(0.3

)3.3

(0.2

)n

s23

Cere

als

stra

w–

0.7

(0.1

)0.4

(0.1

)0.1

0n

s0.6

(0.2

)0.4

(0.3

)n

s23

Fora

ge

oats

–1.8

(0.3

)2.0

(0.4

)n

s2.5

(0.5

)1.3

(0.3

)0.0

4*

7

Kco

nce

ntr

ati

on

gK

kg

-1

Pota

totu

bers

18.0

(0.3

)–

––

19.2

(0.4

)17.4

(0.3

)0.0

08**

63

Maiz

egra

in6.5

(0.4

)5.0

(0.3

)–

0.0

1**

5.7

(0.3

)5.6

(0.6

)n

s21

Maiz

est

over

14.8

(0.8

)15.9

(0.6

)–

ns

15.9

(0.5

)14.8

(0.9

)n

s21

Cere

als

gra

in–

4.9

(0.2

)4.6

(0.4

)n

s4.9

(0.2

)4.7

(0.3

)n

s23

Cere

als

stra

w–

9.5

(0.3

)7.7

(0.5

)0.1

1n

s9.9

(0.5

)7.4

(0.3

)0.0

2*

23

Fora

ge

oats

–14.9

(0.5

)14.4

(0.8

)n

s16.6

(0.9

)12.8

(0.5

)n

s7

Nea

rfiel

ds

are

loca

ted

less

than

500-m

dis

tan

tfr

omfa

rmer

dw

elli

ngs

an

dfa

rare

grea

ter

than

500-m

dis

tan

t.Sig

nifi

can

cep

valu

efo

rdif

fere

nce

sby

two-

sided

tte

stis

als

osh

own

:*P

<0.0

5;

**P

<0.0

1;

***P

<0.0

01.

Drivers of Andean Soil Nutrient Mass Balances 1527

Page 12: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Tab

le5.

Yie

ldan

dN

utr

ien

tE

xport

sof

Cro

ps

inR

ota

tion

al

Sequ

en

ces

an

din

Near

vers

us

Far

Fie

lds

Cro

pan

dfr

act

ion

Year

of

rota

tion

Pvalu

efo

r

dif

fere

nce

Pro

xim

ity

tovil

lage

Pvalu

efo

r

dif

fere

nce

N(fi

eld

s)

On

e(m

an

ure

ap

pli

ed

)T

wo

Th

ree

Near

Far

Yie

ld(M

gh

a-

1)

Pota

to13

(1.0

)–

––

17

(1.5

)11

(1.2

)<

0.0

01**

*63

Maiz

e2.0

(0.3

)2.1

(0.3

)–

Ns

2.6

(0.3

)2.1

(0.4

)0.0

6n

s21

Cere

als

(wh

eat,

barl

ey)

–2.5

(0.2

)1.2

(0.2

)<

0.0

01**

*1.7

(0.2

)1.8

(0.2

)n

s30

NE

xport

(kg

Nh

a-

1)

Pota

to34

(2.6

)–

––

41

(4.3

)26

(2.2

)<

0.0

1**

63

Maiz

e46

(7.3

)34

(7.3

)–

Ns

49

(5.6

)30

(9.4

)0.1

0n

s21

Cere

als

(wh

eat,

barl

ey,

fora

ge

oats

)–

56

(5.0

)24

(3.5

)<

0.0

01**

*41

(6.0

)32

(2.9

)0.1

4n

s30

PE

xport

(kg

Ph

a-

1)

Pota

to6

(0.5

)–

––

7(0

.7)

4(0

.3)

<0.0

01**

*63

Maiz

e16

(2.2

)10

(2.2

)–

0.0

5*

18

(1.7

)8

(2.8

)0.0

1**

21

Cere

als

(wh

eat,

barl

ey,

fora

ge

oats

)–

11

(1.9

)5

(1.4

)<

0.0

01**

*9

(2.4

)6

(1.2

)0.0

5*

30

KE

xport

(kg

Kh

a-

1)

Pota

to60

(4.2

)–

––

78

(6.0

)48

(4.6

)<

0.0

01**

*63

Maiz

e91

(13.0

)71

(13.0

)–

0.0

6n

s101

(10.0

)61

(16.8

)0.0

6n

s21

Cere

als

(wh

eat,

barl

ey,

fora

ge

oats

)–

56

(8.7

)27

(6.4

)0.0

2*

43

(9.4

)35

(6.1

)n

s30

Nea

rfiel

ds

are

thos

ele

ssth

an

500-m

dis

tan

tfr

omfa

rmer

dw

elli

ngs

an

dfa

rare

grea

ter

than

500-m

dis

tan

t.Sig

nifi

can

ceP

valu

efo

rdif

fere

nce

sby

two-

sided

tte

stis

als

osh

own

:*P

<0.0

5;

**P

<0.0

1;

***P

<0.0

01.

1528 S. J. Vanek and L. E. Drinkwater

Page 13: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

a feature of subsistence farming in developing

regions where soil nutrient deficits are common

(Vitousek and others 2009). Within this panorama

of overall deficits, we provide a counterexample to

the linkage often purported between relative pov-

erty and soil degradation. We also suggest that time

lags between the year-to-year focus of farmers on

production goals and the decadal consequences of

erosion and rangeland degradation define the

central challenge to agroecosystem sustainability

for extensive, montane smallholder systems.

Erosion and Rangeland NPP:Environmental Constraintson Management

Erosion is a complex process driven by climate, soil,

topographic, and management factors that has long

been recognized as a major vulnerability for agri-

culture, because soil losses under annual tillage

tend to exceed soil formation rates (Lal 1990;

Montgomery 2007). This is especially true in

mountainous regions where soil erosion from

agriculture on slopes is a central threat to sustain-

ing food production (Lal 1998; Kaihura and others

1999). The erosion rates measured in our study

(29–134 Mg ha-1 in cereal crops, 3–122 Mg ha-1

in fallows) are very high and are comparable to

other Andean erosion data (Alegre and others

1990; Zimmerer 1993; Romero-Leon 2005). The

range of erosion rates within each management

class, crops versus fallow, reflects the impact of

slope, and contributed to drastically different N and

P balances among fields depending on their prox-

imity to flatter village areas. Compared to most

subsistence systems, erosion poses a greater threat

to crop production and food security in these

Andean systems. The rates we measured are several

times those from Sahelian-mixed crop/livestock

systems with flatter topography (5–21 Mg ha-1; Pieri

1989) and also exceed the median rate reported for a

large agricultural data set (� 18 Mg ha-1; Mont-

gomery 2007). A similar pattern of very high erosion

was seen in an Ethiopian montane system, where

erosion comprised 80% of P losses and resulted

Figure 4. Regression of standardized yield data from

114 harvest samplings over 2 years against field slope.

Yields were standardized to the mean and standard

deviation of each crop. Summary statistics for dry grain

or tuber yield (kg ha-1) of each crop is shown in legend:

number of fields sampled (n); mean yield (l) and stan-

dard deviation (s).

Figure 3. Measured soil erosion rates of N, P, and K regressed against RUSLE LS factor, for (A) six cereal fields and

(B) five fallow fields in the study area. Linear regressions shown were used to estimate erosion losses of total N, total P, and

10.1 Kexch for nutrient balances, based on the LS factors of balance fields. ANOVA table (inset) gives significance of RUSLE

LS factor as a linear predictor and of field type (fallow vs. cereal) as a categorical predictor.

Drivers of Andean Soil Nutrient Mass Balances 1529

Page 14: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Tab

le6.

Sce

nari

os

for

Net

An

nu

alize

dN

utr

ien

tB

ala

nce

sover

18

Years

of

Rota

tion

on

aFie

ldw

ith

10%

Slo

pe

Sce

nari

oA

nn

uali

zed

bala

nce

,k

gh

a-

1y

1

Mean

(SD

),200

run

s

Cro

pro

tati

on

an

dfe

rtil

ity

inp

uts

Fall

ow

an

dgre

en

man

ure

ssh

ow

nin

bol

d

NP

K

1.

Sta

tus

qu

oro

tati

on

0.2

(2.5

)-

4.5

(0.6

)-

6.6

(4.1

)R

ota

tion

:P–W

–Fo-T

-ff-

P-M

-Fo-f

ff-P

-Fb-W

-fff

Man

ure

:410-8

5-3

90

kg

ha

-1

N–P-K

over

18

years

2.

Sh

ort

en

ed

fall

ow

s0.5

(3.4

)-

4.6

(0.8

)-

5.5

(4.7

)R

ota

tion

:P–W

–Fo-f

-P-M

-Fo-f

f-P-F

b-F

o-f

f-P–W

-T-f

Man

ure

:520-1

15-5

10

kg

ha

-1

N–P-K

over

18

years

3.

Legu

me/P

stra

tegy

4.8

(2.2

)2.7

(0.8

)-

14.1

(4.8

)R

ota

tion

:P–W

-F/V

rp-G

m-P

-M-T

rp-f

f-P-F

b-F

/V-f

-Gm

-

P–W

-T-f

Man

ure

:410-8

5-3

90

kg

ha

-1

N–P-K

over

18

years

Rock

ph

osp

hate

:160

kg

ha

-1

Pas

RP

iny

3,4

,7,

an

d13

4.

‘‘In

tegra

ted

inte

nsi

fica

tion

’’

(Legu

me/P

plu

sero

sion

redu

ctio

n)

9.8

(2.2

)5.8

(0.8

)-

12.9

(4.8

)R

ota

tion

:P–W

-F/V

rp-G

m-P

-M-T

rp-f

f-P-F

b-F

/V-f

-Gm

-P–W

-T-f

Man

ure

:410-8

5-3

90

kg

ha

-1

N–P-K

over

18

years

Rock

ph

osp

hate

:160

kg

ha

-1

Pas

RP

iny

3,4

,7,

an

d13

Ero

sion

man

agem

en

t:re

du

cest

atu

squ

oero

sion

by

50%

Cro

ps

inro

tati

on:

P,

pot

ato

;W

,w

hea

t;F

o,fo

rage

oat;

T,

tarw

i;M

;m

aiz

e;F

b,

fava

bea

n;

F/V

,fo

rage

oat

wit

hve

tch

;G

m,

tarw

igr

een

man

ure

;f,

fall

ow.

Res

ult

ssh

own

are

mea

ns

an

dst

an

dard

dev

iati

onof

200

run

sof

rota

tion

wit

hra

ndom

dra

ws

for

crop

yiel

ds.

Th

est

atu

squ

oro

tati

onre

pea

tsin

6-y

ear

cycl

es,

wit

h3

years

ofcr

oppin

gan

d3

years

offa

llow

,an

dm

an

ure

once

per

cycl

e.In

ten

sifica

tion

via

‘‘sh

orte

ned

fall

ows’

’gr

ows

pot

ato

esfo

ur

tim

esin

18

years

inst

ead

ofth

ree,

wit

ha

33%

incr

ease

inm

an

ure

appli

cati

on.

Th

e‘‘

legu

me/

P’’

stra

tegy

use

sa

year

ofgr

een

man

ure

bef

ore

pot

ato

,an

dro

ckph

osph

ate

(RP

)addit

ion

tofo

ur

legu

me

crop

sin

the

rota

tion

.‘‘

Inte

grate

din

ten

sifica

tion

’’u

ses

legu

mes

,R

Paddit

ion

,an

der

osio

nre

du

ctio

nw

ith

grass

con

tou

rbarr

iers

an

dse

dim

ent

captu

retr

ench

esso

that

eros

ion

ish

alv

ed.

1530 S. J. Vanek and L. E. Drinkwater

Page 15: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

in large cropland P deficits (Haileslassie and others

2005).

After erosion, rangeland NPP was a second

important environmental driver and point of vul-

nerability for these cropping systems. The positive

correlation of rangeland NPP to both application

rates and nutrient content of manure illustrates its

importance as the primary source of manure-

derived nutrients. This NPP-manure linkage has

been observed in other subsistence farming sys-

tems: one Sahelian rangeland/cropping system

with low, rainfall-limited NPP had manuring rates

only one tenth of those we measured here (Powell

and others 1996), suggesting that NPP of range-

lands globally places constraints on farmer nutrient

management. To understand these constraints,

systemic nutrient balances at the community and

rangeland level are needed to quantify the forage

and manure nutrient flows that rangelands can

sustainably provide based on N fixation, mineral

weathering, and nutrient deposition.

Manure Impacts on Soil NutrientBalances: Field Proximityand Management

In these Andean systems, community decisions

about village siting affect the location of fields for

crop production, creating a complex landscape of

intensively managed fields within a matrix of

communal rangelands that sets the stage for near/

far soil nutrient gradients that correspond with

differences in soil productive capacity due to

steepness and soil depth. Villages occupy flatter

land, suggesting that their locations evolve toward

the best soils, as demonstrated in other managed

landscapes (Imhoff and others 1997). Farmers’

agronomic management practices are nested with-

in the landscape and rely heavily on the steepest

lands which serve as nutrient reservoirs that are

tapped via manure from grazing.

Within this landscape, higher manuring rates in

near fields reinforced near/far erosion differences

and accentuated N and P deficits in far fields, mir-

roring near/far soil P contrasts in other smallholder

systems (Rowe and others 2006; Phiri and others

2010). Landscape factors explain these near/far

gradients. Most simply, far fields are less accessible

in difficult terrain. Farmers in our study hauled

manure up to 1.2 km, perhaps dissuading high

manuring rates. Also, fallows are longer in far than

near fields so that farmers reduce manuring rates to

account for the nutrients from fallow vegetation

(Pestalozzi 2000). Smallholders also achieve greater

returns to their manure from higher crop yields in

flatter near fields (Figure 4). Farmers tend to favor

these fields with manure, whereas far fields are

used to diversify climate and pest risks to crops

(Goland 1993).

Over time, management nested within landscape

gradients is likely to reinforce contrasts in soil

productive potential, a trend also seen in African

systems with near/far management gradients (Gil-

ler and others 2006). Strengthened soil nutrient

gradients in rangeland/cropping systems enable

farmers to produce higher yields in the best soils.

However, as management intensifies in these sys-

tems, reduced manuring rates combined with high

erosion will accelerate soil nutrient depletion in far

fields causing yields to decline.

Examining the stoichiometry of different flows in

the balance also helps to explain patterns of

depletion in nutrients: if inputs and outputs have

unmatched stoichiometries then depletion or sur-

plus can occur. The N:P stoichiometry of manure

inputs was approximately 3:1 (Table 3), which is

slightly less than that of crop exports (�4:1) and

more than that of erosion (�2:1). This shows how

far fields could become P-limited because erosion

accounts for a greater share of nutrient exports,

whereas flatter near fields where harvest exports

dominate the balances retain proportionately

greater P, a conclusion that is substantiated by

lower available P in more marginal fields (Vanek

2011). Nevertheless, these differences in N:P ratios

are not enormous, so that the large erosion N losses

in far fields and minimal direct inputs of N via N

fixation in cropped fields also result in negative N

balances in far fields.

In contrast to these small differences in N:P ra-

tios, K exports from tuber crops and cereal residues

dominated the K balances. This can be attributed to

the high K:P ratio of crop exports, which at

approximately 7:1 greatly exceeded the K:P ratio of

manure or erosion (� 3:1 and 1.5:1, respectively,

Table 3). Potassium balances were so dominated by

crop harvests that they did not vary with erosion

and field proximity. In spite of these K deficits,

the stronger contrasts in P than K crop exports

(Table 5) suggest that P is currently more limiting

than K in these systems, which is reinforced by the

unresponsiveness of Bolivian highland soils to K

addition (Valente and Oliver 1993). However, we

also observed insufficient levels of exchangeable K

(< 125 mg K kg-1) in 12 of 17 fields during soil

sampling in the project area, suggesting that K

limitation may be anticipated in the future

(unpublished data).

Drivers of Andean Soil Nutrient Mass Balances 1531

Page 16: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

The Weak Impact of Relative Wealthon Manuring Rates: Do Poorer FarmersCause Degradation?

Compared to other subsistence farming systems

that have been studied, manuring rates in these

Andean communities were surprisingly consistent

across farms that varied in wealth, even as the ratio

of animals:land increased with wealth. For example

in Ethiopia and Zimbabwe, wealthier farmers ap-

plied greater rates of fertilizer and manure (Elias

and others 1998; Cobo and others 2009), a trend

that was generally true across Africa (Cobo and

others 2010). Compared to Sahelian rangeland–

cropping systems, a far greater proportion of

farmers in our study had access to animal manure

(Harris 1999; Augustine 2003). In one Sahelian

community, Achard and Banoin (2003) found that

32% of farmers had no livestock, and among

farmers with animals only 37% owned enough to

apply appreciable rates of manure. These wealthier

farmers applied 5.5 kg P ha-1 y-1 as manure,

similar to the mean rate we report (Table 3).

However, fully a third of farmers in these com-

munities lacked animals and thus were without

any means to replenish nutrients in their fields, so

that in this region, poorer farmers may degrade

cropland soil fertility by mining soil nutrient stocks

and accelerating soil loss. In the communities we

studied, social mechanisms allow farmers without

herds to pursue a shared standard of soil nutrient

management by exchanging their labor for man-

ure. This transfer of manure to poorer families may

account for the relatively consistent rates of man-

ure application across wealth classes, although we

are not sure based only on this sample of farmers

whether such social mechanisms and equal

manuring rates are prevalent across the region.

Nevertheless, due to similar application rates,

wealthier farmers gather and apply greater total

amounts of manure across their larger holdings,

and may in fact cause greater rangeland degrada-

tion than poor farmers. This would be especially

true if larger total manure harvests reinforce

wealth differences and inequality in herd sizes over

time (Giller and others 2006). Examining the

depletion rates of nutrients from rangeland by

wealthier farmers might contrast with common

assumptions about the impact of poorer farmers on

nutrients in cropland. Mass balances of whole

communities utilizing common-pool resources are

challenging to calculate. They could however

clarify the role of wealth in driving degradation and

also point to sustainable grazing and manure yields

of rangelands as we suggest above.

Lastly, similar manuring rates across wealth

levels and pulses of manure nutrients coordinated

with crop rotations suggest a community knowl-

edge system regarding soil nutrient management.

Andean farmers’ knowledge of soil fertility regen-

eration in fallows (Pestalozzi 2000) and manage-

ment systems of African smallholders (Boesen and

Friis-Hansen 2001) also exemplify this complex

and widely held management knowledge. Ironi-

cally, widely held strategies may insulate farmers

from perceiving range and cropland degradation,

exacerbating the time lag effects considered next.

Time Lags: Modeling as a Learning Toolto Forecast Agroecosystem Trajectories

It is likely that the tendency of farmers in these

systems to overlook soil-degradation stems from

time lags between management and the impact on

the environment. Degradation arising from the

delayed consequences of management on ecosys-

tem services occurs in other managed ecosystems

such as fisheries (Devine and Haedrich 2011) and

grazing management (Zhou and others 2011). In

these Andean systems soil management is oriented

towards food production on a yearly timescale,

whereas soil erosion and rangeland degradation

accrues over decadal timescales to eventually im-

pact production. Manure application is sufficient at

present to mitigate erosion impacts on crop yields.

As a result, soil degradation is not as swiftly

apparent and cannot be factored into the short-

term decisions. Lower yields on steeper slopes

related to erosion deficits represent a signal to

managers who might respond by reversing erosion.

However, crop yields vary around this trend

because of many other biotic and abiotic factors

(Terrazas and others 1998), and above-average

production occurs even on steep slopes where

erosion deficits will accrue in the long run, espe-

cially if crops receive manure (Figure 3). The ero-

sion signal may not be sufficiently perceived for

decades. Farmers may also ignore degradation if

yields decline slowly enough that they redefine

yield benchmarks, as examples from managed

fisheries indicate (Bunce and others 2008). Beyond

these issues of perception is the farmers’ ability to

respond. If farmers respond by increasing manur-

ing rates or expanding cropped area rather than

erosion reduction, rangeland degradation will

accelerate. The current response subsidizes crop

production at shorter time scales and smaller spatial

scales by sacrificing the longer-term resilience

of rangeland, a pattern identified by Carpenter

and others (2001). Zimmerer (1993) also described

1532 S. J. Vanek and L. E. Drinkwater

Page 17: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

labor shortages driven by national-scale policies

that constrain Andean farmers’ response to erosion.

The modeled scenarios we present clarify the

likely outcomes of long-term processes. Shortened

fallows without compensatory changes in man-

agement will result in severe degradation (Table 6).

The legume/P strategy would slow the decline in

soil nutrient stocks by replacing some manure with

legumes and RP. Improved legume forages and

management that access greater soil P might also

improve forage quality and animal-based liveli-

hoods, and also boost soil and manure nutrient

stocks, and soil cover in marginal crop fields and

rangeland, helping to reverse negative soil N and P

balances (for example, Meneses 1998). However,

only erosion reduction along with inputs to redress

past depletion will improve the long-term P and K

balances of these systems. Farmers’ short-term fo-

cus on crop production set within widely-held

knowledge systems may impede the promotion of

soil rehabilitation. In response, short-term positive

outcomes could incentivize changes in practices:

green manures would reduce the effort for manure

transport and the aggregate manure requirement;

live barriers with appropriate species could improve

forage resources; RP would improve legume yields

and enhance fixed N inputs. Nutrient balance

models can support farmer learning about long-

term processes so that the influence of important

drivers like erosion and rangeland NPP is amplified

in local knowledge systems. Removing barriers to

long-term investments in soil fertility by farmers

may also be needed, in the form of community

credit schemes, access to markets that incentivize

soil rehabilitation, or community or regional gov-

ernment assistance.

ACKNOWLEDGMENTS

We acknowledge MODIS 17A3 NPP data from

National Aeronautics and Space administration

(NASA) and the University of Montana Numerical

Terradynamic Simulation Group, and ASTER DEM

data (a product of NASA and the Japanese Ministry

of Economy, Trade, and Industry- METI). We

received invaluable support from Fulbright and

Fulbright-Hays fellowships and the McKnight

Foundation Collaborative Crop Research Program.

We thank World Neighbors Bolivia and collabo-

rating farmers for focal group data and sampling

assistance, and Ann Piombino and Keith Jenkins

for invaluable lab and GIS assistance respectively.

We also thank Marissa Weiss, Jennifer Blesh, Sean

Berthong and anonymous reviewers for sugges-

tions that greatly improved the manuscript.

REFERENCES

Achard F, Banoin M. 2003. Fallows, forage production and

nutrient transfers by livestock in Niger. Nutr Cycl Agroecosyst

65:183–9.

Aganga AA, Mosimanyana N. 2001. Gender impact on sheep

and goat production in Botswana. A case of Gaborone region.

J Agric Tropics Subtrop 102:15–18.

Alegre JC, Felipe-Morales C, LaTorre B. 1990. Soil erosion

studied in Peru. J Soil Water Conserv 45:417–20.

Antil RS, Janssen BH, Lantinga EA. 2009. Laboratory and

greenhouse assessment of plant availability of organic N in

animal manure. Nutr Cycl Agroecosyst 85:95–106.

Arriaga-Jordan CM, Pedraza-Fuentes AM, Nava-Bernal EG,

Chavez-Mejia MC, Castelan-Ortega OA. 2005. Livestock ag-

rodiversity of Mazahua smallholder Campesino systems in the

highlands of Central Mexico. Hum Ecol 33:821–45.

Augustine DJ. 2003. Long-term, livestock-mediated redistribu-

tion of nitrogen and phosphorus in an East African savanna.

J Appl Ecol 40:137–49.

Baijukya F-P, de-Ridder N, Masuki K-F, Giller K-E. 2005.

Dynamics of banana-based farming systems in Bukoba dis-

trict, Tanzania: changes in land use, cropping and cattle

keeping. Agric Ecosyst Environ 106:395–406.

Baker LA, Hope D, Xu Y, Edmonds J, Lauver L. 2001. Nitrogen

balance for the central Arizona–Phoenix (CAP) ecosystem.

Ecosystems 4:582–602.

Berry P-M, Stockdale E-A, Sylvester-Bradley R, Philipps L, Smith

K-A, Lord E-I, Watson C-A, Fortune S. 2003. N, P and K

budgets for crop rotations on nine organic farms in the UK.

Soil Use Manage 19:112–18.

Boesen J, Friis-Hansen E. 2001. Soil fertility management in

semi-arid agriculture in Tanzania: farmers’ perceptions and

management practices. CDR Working Papers, 31 p.

Bunce M, Rodwell LD, Gibb R, Mee L. 2008. Shifting baselines in

fishers’ perceptions of island reef fishery degradation. Ocean

Coast Manage 51:285–302.

Carpenter S, Walker B, Anderies JM, Abel N. 2001. From met-

aphor to measurement: resilience of what to what? Ecosys-

tems 4:765–81.

CIF-UMSS (Centro de Investigacion en Forrajes: Universidad

San Simon). 2013. Guıa ilustrada de especies forrajeras nativas

de la zona andina en Bolivia. Cochabamba: Universidad

Mayor San Simon. p 191.

Cobo JG, Dercon G, Monje C, Mahembe P, Gotosa T, Nya-

mangara J, Delve RJ, Cadisch G. 2009. Cropping strategies,

soil fertility investment and land management practices by

smallholder farmers in communal and resettlement areas in

Zimbabwe. Land Degrad Dev 20:492–508.

Cobo JG, Dercon G, Cadisch G. 2010. Nutrient balances in

African land use systems across different spatial scales: a re-

view of approaches, challenges and progress. Agric Ecosyst

Environ 136:1–15.

Devine JA, Haedrich RL. 2011. The role of environmental conditions

and exploitation in determining dynamics of redfish (Sebastes

species) in the Northwest Atlantic. Fish Oceanogr 20:66–81.

Duncan Fairlie T, Jacob DJ, Park RJ. 2007. The impact of

transpacific transport of mineral dust in the United States.

Atmospheric Environ 41:1251–66.

Elias E, Morse S, Belshaw D-G-R. 1998. Nitrogen and phos-

phorus balances of Kindo Koisha farms in southern Ethiopia.

Agric Ecosyst Environ 71:93–113.

Drivers of Andean Soil Nutrient Mass Balances 1533

Page 18: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Ellis EC, Klein Goldewijk K, Siebert S, Lightman D, Ramankutty

N. 2010. Anthropogenic transformation of the biomes, 1700 to

2000. Glob Ecol Biogeogr 19:589–606.

ERSDAC. 2007. ASTER Global Digital Elevation Model. http.://

www.ersdac.or.jp/GDEM/E/index.html.

FAO. 2010. SD Dimensions Special: Global Climate Maps. SD

Dimensions. FAO Sustainable Development Department.

http://www.fao.org/sd/EIdirect/climate/EIsp0002.htm.

Giller KE, Rowe EC, de Ridder N, van Keulen H. 2006. Resource

use dynamics and interactions in the tropics: scaling up in

space and time. Agric Syst 88:8–27.

Goland C. 1993. Field Scattering as agricultural risk manage-

ment: a case study from Cuyo Cuyo, Department of Puno,

Peru. Mountain Res Dev 13:317–38.

Haigh MJ. 1977. The use of erosion pins in the study of slope

evolution. In: Finlayson B, Ed. British Geomorphological Re-

search Group, Technical Bulletin 18. Norwich, England: Geo

Books. p 31–49.

Haileslassie A, Priess J, Veldkamp E, Teketay D, Lesschen JP.

2005. Assessment of soil nutrient depletion and its spatial

variability on smallholders’ mixed farming systems in Ethiopia

using partial versus full nutrient balances. Agric Ecosys

Environ 108:1–16.

Harris F. 1999. Nutrient management strategies of small-holder

farmers in a short-fallow farming system in north-east Nigeria.

Geogr J 165:275–85.

Hudson NW. 1993. Field measurement of soil erosion and run-

off. Food and Agriculture Organization of the United Nations,

Rome. 139 p

Imhoff ML, Lawrence WT, Elvidge CD, Paul T, Levine E, Pri-

valsky MV. 1997. Using nighttime DMSP/OLS images of city

lights to estimate the impact of urban land use on soil re-

sources in the United States. Remote Sens Environ 59:105–17.

Jones A. 2011. Overcoming barriers to improving infant and

young child feeding practices in the Bolivian Andes: the role

of agriculture and rural livelihoods. Doctoral dissertation,

Cornell University, Ithaca, NY

Kaihura FBS, Kullaya IK, Kilasara M, Aune JB, Singh BR, Lal R.

1999. Soil quality effects of accelerated erosion and manage-

ment systems in three eco-regions of Tanzania. Soil Tillage Res

53:59–70.

Kalra YP. 1998. Handbook of reference methods for plant anal-

ysis. CRC Press, Boca Raton, FL. 300 p

Kihara J, Vanlauwe B, Waswa B, Kimetu JM, Chianu J, Bationo

A. 2010. Strategic phosphorus application in legume-cereal

rotations increases land productivity and profitability in wes-

tern Kenya. Exp Agric 46:35–52.

Lal R. 1990. Soil erosion in the tropics: principles and manage-

ment. New York: McGraw-Hill. 580 p

Lal R. 1998. Soil erosion impact on agronomic productivity and

environment quality. Crit Rev Plant Sci 17:319–464.

Lesschen JP, Stoorvogel JJ, Smaling EMA, Heuvelink GBM,

Veldkamp A. 2007. A spatially explicit methodology to

quantify soil nutrient balances and their uncertainties at the

national level. Nutr Cycl Agroecosyst 78:111–31.

Lightfoot C, Noble R. 2001. Tracking the ecological soundness of

farming systems: instruments and indicators. J Sustain Agric

19:9–29.

Mayer E. 1979. Land-use in the Andes: ecology and agriculture

in the Mantaro Valley of Peru with special reference to

potatoes. Lima, Peru: International Potato Center. 115 p

McCorkle CM, Ed. 1990. Improving Andean sheep and Alpaca

production: recommendations from a decade of research in

Peru. Columbia, Missouri: University of Missouri-Columbia.

Meneses R. 1998. Asociacion de cereales menores con legumi-

nosas y momentos de corte para produccion de forraje.

Compendio de trabajos presentandos por el Proyecto Rhizo-

biologıa (Cochabamba) en eventos y publicaciones de otras

instituciones.

Montgomery DR. 2007. Soil erosion and agricultural sustain-

ability. Proc Nat Acad Sci 104:13268–72.

Mortimore M, Harris F. 2005. Do small farmers’ achievements

contradict nutrient depletion scenarios for Africa? Land Use

Policy 22:43–56.

Neighbors World. 2006. Linea de base, proyecto Heifer de se-

guridad alimentaria Norte de Potosı. Cochabamba, Bolivia:

Vecinos mundiales. 28 pp

Nkonya E, Kaizzi C, Pender J. 2005. Determinants of nutrient

balances in a maize farming system in eastern Uganda. Agric

Syst 85:155–82.

NRCS. 2010. Crop Nutrient Tool. Natural Resources Conserva-

tion Service, U.S. Department of Agriculture, Beltsville. http://

plants.usda.gov/npk/main.

Pacheco P. 2009. Smallholder livelihoods, wealth and defores-

tation in the Eastern Amazon. Hum Ecol 37:27–41.

Pendleton LH, Howe EL. 2002. Market integration, develop-

ment, and smallholder forest clearance. Land Econ 78:1–19.

Pestalozzi H. 2000. Sectoral fallow systems and the management

of soil fertility: the rationality of indigenous knowledge in the

high Andes of Bolivia. Mt Res Dev 20:64–71.

Phiri AT, Njoloma JP, Kanyama-Phiri GY, Snapp S, Lowole MW.

2010. Maize yield response to the combined application of

Tundulu rock phosphate and Pigeon Pea residues in Kasungu,

Central Malawi. Afr J Agric Res 5:1235–42.

Pieri CJ. 1989. Fertility of soils: a future for farming in the West

African Savanna. Berlin: Springer. 348 p

Powell JM, FernandezRivera S, Hiernaux P, Turner MD. 1996.

Nutrient cycling in integrated rangeland/cropland systems of

the Sahel. Agric Syst 52:143–70.

Renard KG, Foster GR, Weesies GA, McCool DK, Yoder DC.

1997. Predicting soil erosion by water: a guide to conservation

planning with the Revised Universal Soil Loss Equation

(RUSLE), Agriculture Handbook 703. Beltsville, MD: USDA

Agricultural Research Service. 384 p

Romero-Leon C. 2005. A multi-scale approach for erosion

assessment in the Andes. The Hague: Wageningen University.

147 p

Ross SM, Izaurralde RC, Janzen HH, Robertson JA, McGill WB.

2008. The nitrogen balance of three long-term agroecosystems

on a boreal soil in western Canada. Agric Ecosyst Environ

127:241–50.

Rowe E, Vanwijk M, Deridder N, Giller K. 2006. Nutrient allo-

cation strategies across a simplified heterogeneous African

smallholder farm. Agric Ecosyst Environ 116:60–71.

Rufino MC, Dury J, Tittonell P, van Wijk MT, Herrero M,

Zingore S, Mapfumo P, Giller KE. 2011. Competing use of

organic resources, village-level interactions between farm

types and climate variability in a communal area of NE Zim-

babwe. Agric Syst 104:175–90.

Saberwal VK. 1996. Pastoral politics: Gaddi grazing, degradation,

and biodiversity conservation in Himachal Pradesh, India.

Conserv Biol 10:741–9.

1534 S. J. Vanek and L. E. Drinkwater

Page 19: Environmental, Social, and Management Drivers of Soil Nutrient Mass Balances in an Extensive Andean Cropping System

Schechambo F, Sosoveli H, Kisanga D. 1999. Rethinking natural

resource degradation in semi-arid sub-Saharan Africa: the

case of semi-arid Tanzania. Dar Es Salaam, Tanzania: Overseas

Development Institute. 58 p.

Scherr SJ. 2000. A downward spiral? Research evidence on the

relationship between poverty and natural resource degrada-

tion. Food Policy 25:479–98.

Schlecht E, Hiernaux P, Achard F, Turner MD. 2004. Livestock

related nutrient budgets within village territories in western

Niger. Nutr Cycl Agroecosyst 68:199–211.

Smaling E-M-A, Fresco L-O, De-Jager A. 1996. Classifying,

monitoring and improving soil nutrient stocks and flows in

African agriculture. AMBIO 25:492–6.

Terrazas F, Suarez V, Gardner G, Thiele G, Devaux A, Walker T.

1998. Diagnosing potato productivity in farmers’ fields in

Bolivia, Working paper 1998-5. Social Science Department,

International Potato Center (CIP), Lima, Peru

Thorne PJ, Tanner JC. 2002. Livestock and nutrient cycling in

crop–animal systems in Asia. Agric Syst 71:111–26.

Valente JF, Oliver R. 1993. Fertisuelos: evaluacion de la fertili-

dad de los suelos del antiplano, valle central y los llanos de

Bolivia. Rome: FAO. 123 p

Vanek S. 2011. Legume-phosphorus synergies in mountain

agroecosystems: field nutrient balances, soil fertility gradients,

and effects on legume attributes and nutrient cycling in the

Bolivian Andes. Doctoral dissertation, Cornell University.

Villarroel J, Augstburger F, Meneses R. 1986. Fixation and

contribution of nitrogen to the soil by Lupinus mutabilis, and

its effects on following barley. Proceedings of the Fourth

International Lupin Conference. p 308.

Vitousek PM, Naylor R, Crews T, David MB, Drinkwater LE,

Holland E, Johnes PJ, Katzenberger J, Martinelli LA, Matson

PA, Nziguheba G, Ojima D, Palm CA, Robertson GP, Sanchez

PA, Townsend AR, Zhang FS. 2009. Nutrient imbalances in

agricultural development. Science 324:1519–20.

Wortmann CS, Kaizzi CK. 1998. Nutrient balances and expected

effects of alternative practices in farming systems of Uganda.

Agric Ecosyst Environ 71:115–29.

Yirga C, Hassan RM. 2006. Poverty soil conservation efforts

among smallholder farmers in the central highlands of Ethi-

opia. South Afr J Econ Manage Sci 9:244–61.

Zhao M, Nemani R, Running S. 2008. ftp://ftp.ntsg.umt.edu/

pub/MODIS/Mirror/MOD17A3.LATEST/Improved_MOD17A3_

C5.1_GEOTIFF_1km/,fileNpp_QC_1km_C5.1_mean_00_to_

06.tif. University of Montana Numerical Terradynamic

Simulation Group, Bozeman, MT.

Zhou ZY, Li FR, Chen SK, Zhang HR, Li GD. 2011. Dynamics of

vegetation and soil carbon and nitrogen accumulation over

26 years under controlled grazing in a desert shrubland. Plant

and Soil 341:257–68.

Zimmerer KS. 1993. Soil-erosion and labor shortages in the

Andes with special reference to Bolivia, 1953–1991: implica-

tions for conservation-with-development. World Dev

21:1659–75.

Drivers of Andean Soil Nutrient Mass Balances 1535