Green mulch decomposition and nitrogen release from leaves of two Inga

spp. in an organic alley-cropping practice in the humid tropics

Humberto A. Leblanca,*, Pekka Nygrenb,1, Robert L. McGrawa

aDepartment of Agronomy, University of Missouri, 210 Waters Hall, Columbia, MO 65211, USAbCenter for Agroforestry, University of Missouri, Columbia, MO, USA

Received 27 August 2004; received in revised form 13 April 2005; accepted 10 May 2005

Abstract

Inga edulis Mart and Inga samanensis Uribe are promising yet little studied legume trees for use in agroforestry on acidic soils. The

objective of this study was to analyze the decomposition and N release processes of green mulch from these species. Litterbags filled

with leaves from each species were placed on the ground in an organic maize (Zea mays L.) alley-cropping experiment at the time of

maize sowing and collected every 2 weeks over a 20 week period, and measured for dry matter, N, hemicellulose, cellulose, and lignin

content. Three types of models were applied to the data, according to the characteristics of each component, to analyze the

decomposition dynamics of whole leaves and leaf components: a negative exponential decay function, an inverted Michaelis–Menten

function, and a linear regression. Initial decay of I. samanensis mulch was faster than I. edulis mulch. However, the recalcitrant fraction

was about half of the initial litter mass in both Inga spp. Hemicellulose disappeared almost completely from the litter during the 20-

week incubation period, while no significant lignin decay occurred. After a slow start, cellulose partially decayed following linear

kinetics. The half-life of labile N, estimated as a Michaelis–Menten parameter, was 10 weeks in I. samanensis and ca. 24 weeks in

I. edulis litter. Polyphenol content was significantly higher in I. edulis. Litter of I. edulis and I. samanensis may be classified as ‘low-

quality’ and ‘medium-quality’ mulch, respectively. Due to the relatively large recalcitrant mulch fraction, both Inga spp. may promote C

sequestration and long-term N accumulation in soil.

q 2005 Elsevier Ltd. All rights reserved.

Keywords: Inga edulis; Inga samanensis; Decay models; Litterbags; Mulch quality; Polyphenol content

1. Introduction

Depletion of soil organic matter is a serious threat to

agricultural production and food security in many tropical

regions. Decreases in soil organic matter leads to a decline

in agricultural and biomass productivity, poor environmen-

tal quality, soil degradation and nutrient depletion, and

finally to food insecurity (Lal, 2004). This regressive

process concerns both large-scale commercial agriculture

that targets short-term benefits in the global market place

0038-0717/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.soilbio.2005.05.012

* Corresponding author. Current address: EARTH University, Agr-

onomy, Mercedes de Guacimo, Limon, 4442-1000 San Jose, Costa Rica.

Tel.: C506 713 0000; fax: C506 713 0001.

E-mail address: hleblanc@earth.ac.cr (H.A. Leblanc).1 Present address: Department of Forest Ecology, University of Helsinki,

P.O. Box 27, 00014 Helsinki, Finland.

and small-holder agriculture where the tradition of

sustainable agriculture is broken under population pressure,

rapidly changing natural and social environment, and

degrading soils. Reverting the degradation of tropical soils

calls for sustainable agricultural practices including no-till

farming, application of compost and mulch, legume cover

crops, and agroforestry (Lal, 2004).

Organic agriculture that is defined as ‘a holistic

production management system, which promotes and

enhances ecosystem health, including biological cycles

and soil biological activity’ (FAO, 2002) is a promising

alternative for many tropical cropping systems in Latin

America. The annual value of organic market in the USA

was $10 billions in 2002, and it is doubling every 2–3 years.

World organic trade grows by 20–30% per year (Geier,

2003). Since organic farmers cannot compensate for a loss

in soil fertility by inputs of industrial fertilizers, the building

and maintenance of soil fertility is a central objective of

organic agriculture (FAO, 2002). Green mulch produced by

Soil Biology & Biochemistry 38 (2006) 349–358

www.elsevier.com/locate/soilbio

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358350

pruning legume trees in agroforestry is a good source of N

(Mafongoya and Nair, 1997; Aihou et al., 1999; Tossah

et al., 1999) that is compatible with organic agriculture.

Green mulch also increases soil organic matter reserves,

promotes C sequestration, and recycles other nutrients (Lal,

2004).

A prerequisite for developing management strategies for

legume green mulch is a clear understanding of the factors

that govern the decomposition process (Vanlauwe et al.,

1997). Nitrogen content (Vanlauwe et al., 1997), polyphenol

content (Palm and Sanchez, 1990), polyphenol to N ratio

(Oglesby and Fownes, 1992), lignin plus polyphenol to N

ratio (Handayanto et al., 1994; Vanlauwe et al., 1997), and

C to N ratio and lignin to N ratio (Cadisch and Giller, 1997;

Vanlauwe et al., 1997) of tree tissue used for green mulch

have been associated with its decomposition and N release

rates.

Pruning residues used for green mulch can be classified

based on N release rate (Mafongoya et al., 1997). ‘High-

quality’ prunings have an N content higher than 2.5%,

contain less than 15% lignin, and less than 4%

polyphenols (Palm et al., 2001). However, these materials

may release N too quickly to be taken up completely by

the crop (Mafongoya et al., 1997). If lignin content is

higher than 15%, polyphenol higher than 3%, and N less

than 2.5%, N can be immobilized (Palm et al., 2001).

‘Low-quality’ plant materials release N too slowly to meet

crop demands, but they may have a longer residual

influence in the soil. ‘Medium-quality’ mulch is expected

to release N in synchrony with crop demand (Mafongoya

et al., 1997).

Inga spp. (Mimosaceae) are potential components of

low-input sustainable agriculture because they have high

biomass productivity and a tolerance to acid soils (Hands,

1998). In addition to providing nutrients, green prunings

of Inga spp. form a permanent mulch cover that helps to

control weeds and breaks the erosive force of heavy

rains of the humid tropics. Latin American farmers

have traditionally used Inga edulis Mart as a shade tree

for coffee (Coffea arabica L.) and cacao (Theobroma

cacao L.) plantations. I. edulis has recently attracted

attention in agroforestry because of its rapid growth in

poor acidic soils (Hands, 1998). Many other Inga spp.

have the characteristics needed for use in agroforestry,

like Inga samanensis Uribe. It grows naturally near rivers

and in the borders of wet forests. This species has

potential as a shade tree and for firewood production

(Zamora and Pennington, 2001).

The objectives of this experiment were to (1) determine

the decomposition rate of I. edulis and I. samanensis leaves

in an organic alley-cropping system under humid tropical

conditions, (2) study the fate of various compounds in

leaves during the decomposition process, and (3) fit

mathematical models for analyzing and predicting leaf

decomposition and N release rate of these two species.

2. Materials and methods

2.1. Study site

The study was conducted at the EARTH University

Organic Farm in conjunction with a maize (Zea mays L.)

alley-cropping experiment using I. edulis and I. samanensis

as hedgerow trees. EARTH University is located in the

Caribbean coastal plain of Costa Rica (10810 0 N, 83837 0 W,

95 m a.s.l.). The climatic zone is classified as a premontane,

wet forest, basal belt transition (Bolanos and Watson, 1993).

Annual rainfall averages 3464 mm and is evenly distributed

throughout the year. Annual mean temperature is 25.1 8C.

The soil is classified as Thaptic Hapludand with the

following characteristics: pH 5.12 in water; organic C

3.47%, N 0.59% (Kjeldahl); exchangeable acidity 0.3 cmol

(C) kgK1; Ca 4.2, Mg 1.4, and K 0.15 cmol (C) kgK1; P

14.2 mg kgK1 (modified Olsen); Cu 28.1, Fe 109, Zn 18,

and Mn 5.8 mg kgK1.

The maize alley-cropping experiment was established in

1999. Seedlings of I. samanensis and I. edulis were planted

in separate 25!25 m plots, in east–west oriented rows at

4-m between-row and 0.5-m within-row density totaling

5000 trees haK1. Experiment was arranged in randomized

complete blocks with 4 replicates per species. The trees

were pruned approximately every 6 months leaving 5–10%

of the foliage. Maize was sown immediately after pruning at

the rate of 40,000 plants haK1. Maize rows were 1 m apart

and two seeds were sown every 50 cm within the row. Two

maize cycles were cropped on the site before this study.

The green mulch decomposition experiment was con-

ducted from May to September 2002. A 5-month period was

selected because it was 1-month longer than the 4-month

maize intercropping with Inga trees. In May 2002, leaves of

I. samanensis and I. edulis were collected from the alley

cropping experiment and sun-dried in a glasshouse for

3 days. Litterbags measuring 25!25 cm with 2-mm mesh

were filled with 42 g (dry mass) of I. edulis or I. samanensis

leaves. Litterbags were placed on the ground in the center of

the alleys of the plot where the material was collected at the

time of maize sowing. Each of the four replicates consisted

of 10 litterbags. One litterbag per species and replicate was

collected every 2 weeks from week 2–20. Trees were

allowed to regrow, but were lightly pruned 6 weeks after

maize sowing to reduce shading. Only small branches

overhanging the alleys were removed and the material was

left in the alleys. This is a normal practice in alley cropping,

and the microclimate and soil conditions during the

experiment were representative to the studied cropping

system.

2.2. Chemical analysis

Leaf material remaining in each litterbag was dried at

60 8C for 72 h to determine dry matter. The leaves were

ground to pass through a 686-mm screen. Nitrogen

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358 351

concentration was determined using the standard Kjeldahl

technique (Bremner, 1996).

Lignin, cellulose, and hemicellulose were determined at

the Animal Nutrition Laboratory at CATIE, Costa Rica,

using methods described by Lopez et al. (1992), which are

similar to those described by Van Soest (1963). Neutral

detergent fiber (NDF) and acid detergent fiber (ADF)

contents were determined. Hemicellulose was calculated as

NDF minus ADF. The ADF, which contains cellulose,

lignin, and the mineral fraction, was treated with KMnO4 to

oxidize the lignin. Lignin was calculated as ADF minus the

sample after oxidation. The residual sample was ashed in a

furnace muffle to remove the cellulose, leaving the mineral

fraction. Cellulose was determined as the sample after

oxidation minus the mineral fraction. Polyphenols were

extracted from leaves in a 50% aqueous methanol solution

for 1 h in a 80 8C water bath, analyzed by the Follin–Denis

method, and reported as percentage of catechol (Anderson

and Ingram, 1998).

2.3. Mathematical and statistical analysis

The experimental design was a split-plot in time

(Cochran and Cox, 1992), with species as main plots and

times as subplots. There were four replicates per species at

each sampling date.

Three formulations of the negative exponential decay

function were fitted to the data. The double exponential

form (Paul and Clark, 1996) estimates the sample mass of

component i remaining after incubation time t, Mi(t)

MiðtÞ Z Ml expðKkltÞCMr expðKkrtÞ (1)

where Ml is initial mass of labile fraction; kl is the

decomposition coefficient of Ml; Mr is initial mass of

recalcitrant fraction; kr is the decomposition coefficient of

Mr.

In the case of an incubation time that is too short for

observing significant changes in the recalcitrant fraction, or

when part of a decaying component is immobilized, Eq. (1)

simplifies to a single exponential decay model with

intercept (SEI model):

MiðtÞ Z Ml expðKkltÞCMr (2)

This formulation means that decomposition of Mr is not

considered or it refers to the immobilized fraction of the

decaying substrate. For a component that decays completely

within the observation time, a simple negative exponential

decay function applies (SE model):

MiðtÞ Z Ml expðKkltÞ (3)

The inverted Michaelis–Menten model (MM model) can

also be applied in the case where Eq. (2) or (3) applies. The

parameters of the MM model have different interpretations

than those of negative exponential decay function, and can

thus be applied to the same data set for analyzing different

aspects of the decay process. The remaining mass of a

sample or component i after incubation time t, Mi(t) is

estimated

MiðtÞ Z M0 1 Kait

bi C t

� �(4)

where M0 is the initial mass of component i and ai and bi are

Michaelis–Menten parameters for component i. The

equation is derived from enzyme or microbial growth

kinetics, in which case ai is interpreted as the maximum

reaction or growth rate and bi is the substrate concentration

in which half of ai is reached (Thornley and Johnson, 1990;

Holmberg, 1981). The inversion calls for a redefinition of

the parameters. When applied to a data set that also follows

kinetics of Eq. (2), i.e. there is a sizable recalcitrant or

immobilized fraction, ai may be redefined as the labile

fraction or equivalent to ratio of Ml in Eq. (2) to initial

sample mass, and bi is the time when half of the labile

fraction decays.

The relationships of the parameters of Eqs. (2) and (4)

are:

Ml zaM0 (5)

Mr zð1 KaÞM0 (6)

In a decay process that follows Michaelis–Menten type

kinetics as described by Eq. (4), the mass of component i

that decays in time b, Mi(b), is:

MiðbÞ Z aM0=2 (7)

Eqs. (1)–(4) were fitted to the data of decomposition of

whole leaves or their components according to the behavior

of each specific component (Paul and Clark, 1996). Simple

linear regression was applied for the decomposition of the

most recalcitrant leaf components, cellulose, and lignin.

The non-linear models were fitted using the Newton least

squares iterative method of SAS v. 8.02 (SAS Institute,

1999). The model type that was applied in each case was

selected according to high r2 value, narrow confidence

interval of parameter values, and equal distribution of

residuals around zero when plotted against incubation time.

Only the best-fit models are shown. The models were fitted

by species. The significance of the differences between the

models fitted by species was tested by the analysis of

residual variance (Mead et al., 2002). The core of the

analysis was to determine if the subsets of data differed so

much that model parameters had to be fitted independently

for both species.

3. Results

3.1. Chemical characteristics of leaves

Leaves of I. samanensis and I. edulis had similar N, total

C, and cellulose contents (Table 1). I. edulis had

Table 1

Average chemical composition (% of dry matter) and average ratios of main chemical components of leaves of Inga samanensis and Inga edulis at the

beginning of field incubation in litterbags

Species Total N Total C C:N Hemicellulose Cellulose Polyph.

(Catechol)

Polyph.:N Lignin Lignin:N (LigninC

Polyph.):N

I. samanensis 2.94 54.8 19.6 19.4 32.0 3.14** 1.1** 29.8 6.71** 7.76**

I. edulis 2.70 54.3 21.7 26.3 33.2 5.02** 2.03** 18.9 11.86** 13.89**

**Statistically significant difference between species (Student’s t-test at p%0.01).

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358352

significantly higher polyphenol content (Student’s t-test

with p%0.01). Ratios related to N release were calculated

using the chemical compounds listed above. I. edulis had a

similar C:N ratio as I. samanesis, but I. edulis had

significantly higher polyphenol:N and (ligninCpolyphe-

nol):N ratios (p%0.01). Lignin:N was the only ratio that was

significantly higher in I. samanensis (p%0.01) (Table 1).

3.2. Leaf decomposition

The analysis of variance for leaf mass in litterbags

showed significant differences between species and weeks

(p%0.01). Mulch of I. samanensis decomposed faster than

mulch of I. edulis. Eq. (1) did not fit adequately to the

litterbag data on total mass loss, because kr was 0 for both

species. Both the MM model (Eq. (4)) and the SEI model

(Eq. (2)) produced a good fit to the data (Tables 2 and 3;

Fig. 1). The analysis of residual variance (Mead et al., 2002)

indicated that the fit of both models differed significantly

between species (p%0.01).

Table 2

Parameter values of the inverted Michaelis–Menten model (Eq. (4)) for predictin

function of incubation time in litterbags under humid tropical conditions

Component Species M0 (g) a

Total mass I. samanensis 41.83 (39.98–43.68) 0

I. edulis 42.09 (41.01–43.16) 0

Hemicellulose I. samanensis 8.288 (7.548–9.028) 0

I. edulis 11.25 (9.915–12.58)

Total N I. samanensis 1.246 (1.187–1.304) 0

I. edulis 1.226 (1.157–1.294) 0

n.s.: non-significant parameter value. The asymptotic confidence interval of each pa

for hemicellulose.

Table 3

Parameter values of the exponential decay equations (Eq. (2) for total mass and tota

samanensis and Inga edulis leaves as a function of incubation time in litterbags u

Component Species Ml (g)

Total mass I. samanensis 20.16 (18.61–21.91)

I. edulis 15.82 (14.59–17.04)

Hemicellulose I. samanensis 6.980 (5.909–8.051)

I. edulis 10.62 (9.043–12.20)

Total N I. samanensis 0.6091 (0.5315–0.6867)

I. edulis 0.5701 (0.3276–0.8125)

N.A.: not applicable to this compound. The asymptotic confidence interval of each

22 for hemicellulose.

The MM and SEI models gave about the same

fractionation to labile and recalcitrant components; i.e.

aM0zMl according to Eq. (5) for total mass (Tables 2 and 3).

The parameter b of MM model showed that half of the labile

fraction decomposed in ca. 6 and 7.5 weeks in litter of

I. samanensis and I. edulis, respectively (Table 2; Fig. 1). The

SEI model indicated an important recalcitrant fraction (Mr)

for I. samanensis, 20 g (49% of the initial sample mass), and

for I. edulis, 26.12 g (62%) (Table 3). Comparison of the

parameter M value with the initial mass of cellulose and

lignin in the litterbags provides a rough means for evaluating

SEI model behavior. The sum of initial cellulose and lignin

mass was 25.9 g in I. samanensis and 21.8 g in I. edulis.

3.3. Decomposition of polysaccharides and lignin

Hemicellulose decayed exponentially during the 20-

week incubation period and disappeared almost completely.

Half of the initial hemicellulose mass decayed in 2.8 and 2.5

weeks in I. samanensis and I. edulis, respectively (Tables 2

g remaining sample mass of Inga samanensis and Inga edulis leaves as a

b (week) r2

.5537 (0.4492–0.6301) 5.983 (3.075–8.892) 0.89

.5035 (0.4492–0.5578) 7.458 (4.930–9.986) 0.95

.8527 (0.7692–0.9362) 2.800 (1.506–4.094) 0.93

1.018 (0.9152–1.121) 2.457 (1.224–3.690) 0.92

.6855 (0.5512–0.8198) 9.752 (4.635–14.87) 0.89

.7881 (0.2537–1.322) 24.43n.s. (K4.360–53.23) 0.72

rameter is given in parentheses; nZ44 for total mass and total N, and nZ22

l N; Eq. (3) for hemicellulose) for predicting remaining sample mass of Inga

nder humid tropical conditions

Mr (g) k (weekK1) r2

20.00 (–) 0.0981 (0.0837–0.1125) 0.87

26.12 (25.00–27.24) 0.1478 (0.1161–0.1795) 0.95

N.A. 0.0856 (0.0601–0.1112) 0.78

N.A. 0.1974 (0.1414–0.2535) 0.87

0.6370 (0.5599–0.7142) 0.1286 (0.0842–0.1730) 0.89

0.6613 (0.3994–0.9232) 0.0714 (0.0103–0.1325) 0.72

parameter is given in parentheses; nZ44 for total mass and total N, and nZ

Fig. 1. Total mass loss (a and b), loss of hemicellulose (c and d), and loss of nitrogen (e and f) from leaves of Inga samanensis (left graphs) and I. edulis (right

graphs) during 20-week field incubation period in litterbags under humid tropical conditions. The trend lines based on inverted Michaelis–Menten model (MM

model; Eq. (4)) and negative exponential models (exp model) are also shown. The exp model is negative exponential model with intercept (SEI; Eq. (2)) for

mass (a and b) and N loss (e and f) and negative single exponential model (SE; Eq. (3)) for hemicellulose (c and d). Note that the y-axis scale is set to show the

range of variation during the 20-week-period, and it starts from 0 only for hemicellulose.

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358 353

and 3; Fig. 1). The MM model and the SE model (Eq. (3))

fitted well to the data on hemicellulose decomposition in

leaves (Tables 2 and 3; Fig. 1). Analysis of residual variance

indicated that the MM models fitted differed significantly

between species (p%0.05) but the SE models were not

significantly different.

The best-fit equation to cellulose decomposition was a

linear regression. The slope was significant for both species

(p%0.01) indicating significant cellulose decomposition

during the field incubation (Fig. 2). The regressions differed

significantly between species (analysis of residual variance

at p%0.05). During the 20-week period, 40 and 28% of

Fig. 2. Linear regression model for cellulose decay in leaves of Inga samanensis (a) and I. edulis (b) during 20-week field incubation period in litterbags under

humid tropical conditions.

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358354

the initial cellulose content disappeared from the leaves of

I. samanensis and I. edulis, respectively.

No apparent changes in lignin content occurred during

the field incubation period. The slope of the linear

regression was not significant for lignin and the r2 was

low, indicating that leaf lignin did not significantly

decompose in 20 weeks.

3.4. Nitrogen release

The analysis of variance for mass of N remaining in leaf

litter showed that there were significant differences between

species and between weeks (P%0.01). More N was released

from the leaves of I. samanensis than from the leaves of

I. edulis. Both MM and SEI models fitted satisfactorily to the

data. The analysis of residual variance indicated that models

fitted by species differed significantly (p%0.05) in both cases.

The MM model gave higher estimates for labile N mass than

the SEI model; i.e. aM0OMl (Tables 2 and 3, respectively).

The parameter b value of MM model for I. edulis was not

significant (Table 2). The confidence interval of b for

I. samanensis was wide, but the value was significant. The

confidence intervals of parameters M0 and a of the MM

model (Table 2) were narrow indicating a reliable fit. The

labile N mass estimated to decay in b week according to

Eq. (7) was 0.427 g for I. samanensis. The observed loss of

N was 0.471G0.065 g 10 weeks. Thus, the estimate for

I. samanensis was within one standard deviation (SD) of the

observed value. The respective calculation was not done for

I. edulis because the b value was not significant.

The confidence intervals of all parameters of the SEI

model were narrow (Table 3). The reliability of SEI model

parameter values can be further evaluated by comparing the

value of the parameter Mr with N content by the end of

the incubation period. The N content retained in the

litterbags over the 20-week period was 0.702G0.077 g for

I. samanensis and 0.866G0.019 g for I. edulis. Thus, the Mr

estimate was within 1SD from the observed mean for

I. samanensis but a clear underestimation for I. edulis

(Table 3).

The decomposition of N compounds in the leaves of

I. edulis started slowly, and the N mass increased

significantly (Duncan’s Multiple Range Test at p!0.05)

from the beginning to weeks 2 and 4 (Fig. 1). Colonies of

fungi were observed in the litterbags, but fungal biomass

was not quantified.

Considering a typical application of green mulch of

5 Mg haK1 containing 145 kg [N] haK1, I. samanensis and

I. edulis would release 68 and 52 kg [N] haK1 in 20 weeks,

respectively (Fig. 3).

3.5. General composition

The changes in major litter components, cellulose, lignin,

and hemicellulose, and N compounds, during the 20-week

field incubation were estimated using the best-fit models for

each component (Fig. 4). The general mass loss from

I. samanensis and I. edulis leaves was mainly due to decay

of hemicellulose, some cellulose, and N degradation. Lignin

was an important litter component, but it was not

significantly affected by microbial decomposition within

20 weeks.

Carbon release from decomposing mulch was estimated

using the C content of hemicellulose, cellulose and organic

N compounds, and their decay rates estimated according to

the MM model. Considering an average mulch application

of 5 Mg haK1, I. samanensis and I. edulis would release 909

and 876 kg [C] haK1 in 20 weeks, respectively (Fig. 3).

4. Discussion

4.1. Decay of total C, polysaccharides, and lignin

The litterbag incubation experiment was designed for

simulating the fate of green mulch of I. samanensis

Fig. 3. Cumulative release of carbon (a) and nitrogen (b) from 5 Mg of

green mulch of Inga samanensis and Inga edulis estimated according to the

inverted Michaelis–Menten model (Eq. (4)). This amount of mulch has

initial N content of 145 kg.

Fig. 4. Changes in mass of main leaf compounds of Inga samanensis (a) and

I. edulis (b) in 20 weeks estimated according to the linear regression for

cellulose (see Fig. 2), and inverted Michaelis–Menten model (Eq. (4)) for

other compounds.

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358 355

and I. edulis after application as green manure for maize in

an organic alley-cropping practice under humid tropical

conditions. Litter decomposition resulted in complete

disappearance of hemicellulose and partial N release and

cellulose decay from the leaves during a maize cropping

period. No significant decay of lignin was observed in 20

weeks. Significant differences between species were

observed in both general decay and N release rate.

A three-step decay model (i.e. the sum of three negative

exponential models), which describes the decomposition of

labile, slowly decomposing, and recalcitrant fractions,

respectively, is often used for analyzing the microbial

decay of organic matter (Paul and Clark, 1996). However,

the SEI model, which includes the initial mass of slowly

decomposing and recalcitrant fraction as a constant

(parameter Mr in Eq. (2)) gave the best-fit to the data in

our case (Fig. 1; Table 3). The fit of the SEI model to

the general decay dynamics of green mulch may be

explained in the light of the behavior of hemicellulose,

cellulose, and lignin that formed a total of 81.2 and 78.4% of

dry matter of I. samanensis and I. edulis litter, respectively.

Although not always considered as part of the labile fraction

due to its complexity (Paul and Clark, 1996), decay of

hemicellulose followed the SE model, and it decomposed in

20 weeks. Cellulose decay started slowly (Fig. 2), but over

the 20-week field incubation period, a linear regression

best described the general trend. However, the slope was

so low that cellulose mostly contributed to the parameter

Mr. Lignin did not significantly decay within 20 weeks in the

litter of either species.

The value of parameter Mr of the SEI model was an

underestimate of 23% for I. samanensis and an overestimate

of 16% for I. edulis of the initial mass of cellulose and

lignin. The differences between the Mr and the observed

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358356

values were partly caused by the fraction of cellulose

decayed in 20 weeks (Fig. 2), and partly by random

variation.

The MM model is a continuous function for describing

the decay of the labile and slowly decomposing fractions. It

has been used successfully to predict the decomposition of

woody legume nodules (Nygren et al., 2000). The MM

model fitted satisfactorily to the data on the disappearance

rate of organic matter from the litter bags and the

hemicellulose decay rate (Table 2). The relationships

between the parameters of SEI and MM model predicted

by Eq. (5) held (Section 3.2). The half-lives of the labile

fraction predicted by the parameter b (Table 2), 6 and 7.5

weeks for I. samanensis and I. edulis, respectively, were

strongly affected by the short half-life of hemicellulose,

2.5–2.8 weeks under the studied humid tropical conditions

(Table 2).

After the 20-week period, an average of 59% of the

initial dry mass remained in litterbags of I. samanensis and

67% of I. edulis. Palm and Sanchez (1990) working under

similar tropical conditions, found that around 70% of the

original leaf mass of I. edulis remained after 20 weeks. They

fitted a SE model for analyzing the decomposition rate of

leaves of three legume tree species, including I. edulis; the

r2 values were higher than 0.83, or slightly lower in

comparison to the fit of the SEI model used in this study.

4.2. Nitrogen dynamics

Both SEI and MM models described satisfactorily the

general trends in N disappearance from the litterbags

(Fig. 1; Tables 2 and 3). Significant differences between

species were observed in N dynamics of decomposing litter.

The parameter b of MM model was not significant for

I. edulis (Table 2) indicating that half of the final labile N

mass did not decay in 20 weeks. Half of the labile N in

I. samanensis litter decayed in approximately 10 weeks. This

indicates considerably faster N release from I. samanensis

litter than from I. edulis litter.

The b of MM model, or the estimate of N half-life in

green mulch, is the most sensitive parameter to short-term

deviations from general process kinetics (Thornley and

Johnson, 1990). Thus, the failure to accurately estimate the

b value, and the slow start of N release from I. edulis may

indicate important N immobilization in microbial biomass.

Colonies of fungi were observed in the litterbags. Fungal

biomass was not quantified in this study, but it is possible

that the fungi immobilized part of N in the litterbags. Fungal

colonies were also observed in I. samanensis, and some N

immobilization may have occurred in both species.

However, further studies are needed for testing the

hypothesis on N immobilization.

Differences in N release rate between I. samanensis and

I. edulis were expected because of the different leaf

polyphenol contents (Palm and Sanchez, 1990). Leaves of

I. edulis that had the lower N release rate contained more

polyphenol (Table 1). Also the ratios of polyphenol to N and

(ligninCpolyphenol) to N were higher for I edulis, the

species with a lower N release rate, as has been suggested in

other studies (Oglesby and Fownes, 1992; Handayanto et

al., 1994; Vanlauwe et al., 1997).

4.3. Agronomic considerations

The leaf decomposition rate of both Inga spp. was slow

in comparison to other tree species commonly used in

tropical agroforestry. In an Erythrina sp., 21% of leaf dry

matter remained after a 20-week-field incubation period

(Palm and Sanchez, 1990). In Gliricidia sepium (Jacq.)

Kunth ex Walp. and Leucaena leucocephala (Lam) De Wit.,

about 60% of leaf dry matter remained after 2-mo-field

incubation period (Handayanto et al., 1994).

The sizable recalcitrant fraction indicates that the green

mulch of the two Inga spp. may promote C sequestration in

soil. Much of the initial loss of ca. 900–1000 kg [C] haK1

from the decaying mulch is probably released to atmosphere

as CO2, but the recalcitrant fraction may bind C to microbial

biomass and soil organic matter for long periods. The

apparent immobilization of N is probably connected to the

slow decay of the recalcitrant fraction (Resh et al., 2002).

The average cumulative N release was 47% for

I. samanensis and 36% for I. edulis in 20 weeks. Litter of

I. edulis and I. samanensis can be classified as ‘low’ and

‘medium-quality’ mulch, respectively, in comparison to

other species used in tropical agroforestry. In other

incubation studies, the percentage of initial N content

mineralized was 14% for Calliandra calothyrsus Meisn.,

32% for L. leucocephala and 40% Sesbania sesban (L)

Merr. in 12 weeks (Oglesby and Fownes, 1992); 5% for

Acacia auriculiformis A. Cunn. ex Benth., 30% for I. edulis,

52% for Cassia siamea Lam, and 72% for G. sepium in 16

weeks (Constantinides and Fownes, 1994).

If we consider an average application of 145 kg [N] haK1

in green mulch, the N release in 20 weeks from I. samanensis

and I. edulis (Fig. 4) would fulfill 2/3 and half, respectively,

of the N-fertilization recommendation of 100 kg [N] haK1

by the Costa Rican Ministry of Agriculture and Animal

Husbandry for the study zone (MAG, 1991). However, the

residual effect of the green mulch that slowly releases N

could result in long-term accumulation of soil N (Kass et al.,

1997).

In organic agriculture where nutrients cannot be added to

the system applying chemical fertilizers, the main task is to

maintain long-term soil fertility by enhancing internal

nutrient cycling. Long-term experiments in Europe have

shown that the effect of the organic amendments increase

with time due to the residual effect combined with the effect

of the amendment applied during the cropping cycle

(Benzing, 2001). Thus, the medium- and low-quality

mulch of I. samanensis and I. edulis that are expected to

slowly increase soil organic matter and N content could be

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358 357

adequate amendments for long-term soil improvement in

organic agriculture.

5. Conclusion

In spite of the favorable environmental conditions, the

leaf decomposition rate of both Inga spp. was relatively

slow in comparison to many tree species used in tropical

agroforestry; 59 and 67% of initial litter mass was retained

over the 20-week incubation period. Although I. edulis litter

contained more labile hemicellulose and less recalcitrant

lignin than I. samanensis, the latter decomposed faster. This

may indicate an important inhibitory effect of the high

polyphenol content of I. edulis on microbial decay.

Hemicellulose disappeared almost completely in 20

weeks. A small but significant fraction of cellulose also

decayed. Because of the relatively large recalcitrant

fraction, green mulch of the two Inga spp. is likely to

sequester C to soil. Nitrogen release rate was greater from

mulch of I. samanensis than from I. edulis. The N release

from a typical green mulch application of both Inga spp.

would only partially fulfill the N-fertilization recommen-

dations for maize during a cropping season, but the slow

release of N may contribute to long-term N enrichment of

the soil.

The MM and SEI models used for analyzing the decay

process satisfactorily described the decomposition

dynamics of these 2 litters with a relatively large recalcitrant

factor during a maize cropping cycle.

Acknowledgements

This research was supported by the Missouri Agricultural

Experiment Station and the EARTH University Research

Committee.

References

Aihou, K., Sanginga, N., Vanlauwe, B., Lyasse, O., Diels, J., Merckx, R.,

1999. Alley cropping in the moist savanna of West-Africa. I.

Restoration and maintenance of soil fertility on ‘terre de barre’ soils

in Benin republic. Agroforestry Systems 42, 213–227.

Anderson, J.M., Ingram, J.S.I., 1998. Tropical Soil Biology and Fertility. A

Handbook of Methods, second ed. CAB International, England. 221 pp.

Benzing, A., 2001. Agricultura organica. Fundamentos para la region

andina. Neckar, Villingen-Schwenningen, Germany. 682 pp.

Bolanos, R.A., Watson, V.C., 1993. Mapa ecologico de Costa Rica, segun

el sistema de clasificacion de zonas de vida del mundo de L.R Holdrige.

Centro Cientıfico Tropical, Costa Rica.

Bremner, J.M., 1996. Nitrogen-total. In: Bartels, J.M., Bigham, J.M. (Eds.),

Methods of Soil Analysis. Part 3. Chemical Methods. SSSA and ASA,

Madison, WI, pp. 1085–1121.

Cadisch, G., Giller, K.E., 1997. Driven by Nature; Plant Litter Quality and

Decomposition. Oxford University Press, UK. 409 pp.

Cochran, W.G., Cox, G.M., 1992. Experimental Designs. Wiley, USA. 611

pp.

Constantinides, M., Fownes, J.H., 1994. Nitrogen mineralization from

leaves and litter of tropical plants: relationship to nitrogen, lignin, and

soluble polyphenol concentrations. Soil Biology & Biochemistry 26,

49–55.

FAO, 2002. Organic Agriculture, Environment and Food Security. In: El-

Hage Scialabba, N., Hattam, C. (Eds.), http://www.fao.org/sd/2003/

EN0102_en.htm.

Geier, B., 2003. An overview and facts or worldwide organic agriculture,

organic trade a growing reality. In: FAO, IFOAM, EARTH Net

fundation (Eds.), Seminar on Production and Export of Organic Fruits

and Vegetables in Asia. FAO, IFOAM, EARTH Net fundation,

Bangkok, Thailand, pp. 9–10.

Handayanto, E., Cadisch, G., Giller, K.E., 1994. Nitrogen release from

prunings of legume hedgerow trees in relation to quality of the prunings

and incubation method. Plant & Soil 160, 248–337.

Hands, R.M., 1998. The use of Inga in the acid soils of the rainforest zone:

alley-cropping sustainability and soil-regeneration. In:

Pennington, T.D., Fernandez, E.C.M. (Eds.), The Genus Inga

Utilization. The Royal Botanic Gardens, Kew, London, England,

pp. 53–86.

Holmberg, A., 1981. On the practical identificability of growth models

containing Michaelis–Menten type nonlinearities. Helsinki Univer-

sity of Technology, Systems Theory Laboratory, Report B 63, 24

pp.

Kass, D.C.L., Sylvester-Bradley, R., Nygren, P., 1997. The role of nitrogen

fixation and nutrient supply in some agroforestry systems of the

Americas. Soil Biology & Biochemistry 29, 775–785.

Lal, R., 2004. Soil carbon sequestration impacts on global climate change

and food security. Science 304, 1623–1629.

Lopez, F., Rodrıguez, G., Kass, M., 1992. Manual de metodos

rutinarios. Laboratorio de nutricion animal. CATIE, Turrialba,

Costa Rica. 52 pp.

Mafongoya, P.L., Nair, P.K., 1997. Multipurpose tree prunings as source of

nitrogen to maize under semiarid conditions in Zimbabwe I. Nitrogen-

recovery rates in relation to pruning quality and method of application.

Agroforestry Systems 35, 31–46.

Mafongoya, P.L., Nair, P.K., Dzowela, B.H., 1997. Multipurpose tree

pruning as a source of nitrogen to maize under semiarid condition of

Zimbabwe. 2. Nitrogen-recovery rates and crop growth as influenced by

mixtures and pruning. Agroforestry Systems 35, 47–57.

MAG, 1991. Aspectos tecnicos sobre cuarenta y cinco cultivos agrıcolas de

Costa Rica. Ministerio de Agricultura y Ganaderıa. Direccion general

de investigacion y extension agrıcola, Costa Rica. 560 pp.

Mead, R., Curnow, R.N., Hastel, A.M., 2002. Statistical Methods in

Agriculture and Experimental Biology. Chapman and Hall, London.

488 pp.

Nygren, P., Lorenzo, A., Cruz, P., 2000. Decomposition of woody legume

nodules in two tree/grass associations under contrasting environmental

conditions. Agroforestry Systems 48, 229–244.

Oglesby, K.A., Fownes, J.A., 1992. Effects of chemical composition on

nitrogen mineralization from green manures of seven tropical legumes

trees. Plant & Soil 143, 127–132.

Palm, C.A., Sanchez, P.A., 1990. Decomposition and nutrient

release patterns of the leaves of tree tropical legumes. Biotropica

22, 330–338.

Palm, C.A., Gachengo, C.N., Delve, R.J., Cadisch, G., Giller, K.E., 2001.

Organic inputs for soil fertility management in tropical agroecosystems:

application of an organic resource database. Agriculture Ecosystems &

Environment 83, 27–42.

Paul, E.A., Clark, F.E., 1996. Soil Microbiology and Biochemistry.

Academic Press, USA. 273 pp.

Resh, S.C., Binkley, S.C., Parrotta, J.A., 2002. Greater soil carbon

sequestration under nitrogen-fixing trees compared with Eucalyptus

species. Ecosystems 5, 217–231.

H.A. Leblanc et al. / Soil Biology & Biochemistry 38 (2006) 349–358358

SAS Institute, 1999. SAS/STAT User’s Guide: Statistics. V.8.02 ed. SAS

Institute, Cary, NC. 846 pp.

Thornley, J.H.N., Johnson, I.R., 1990. Plant and Crop Modeling. A

Mathematical Approach to Plant and Crop Physiology. Clarendon

press, Oxford, UK. 669 pp.

Tossah, B.K., Zamba, D.K., Vanlauwe, B., Sanguinga, N., Lyasse, O.,

Diels, J., Merckx, R., 1999. Alley cropping in the moist savanna of

West-Africa. II. Impact on soil productivity in north-to-south transect in

Tongo. Agroforestry Systems 42, 229–244.

Vanlauwe, B., Sanginga, N., Merckx, R., 1997. Decomposition or four

Leucaena and Senna prunings in alley cropping systems under sub-

humid tropical conditions: the process and its modifiers. Soil Biology &

Biochemistry 29, 131–137.

Van Soest, P.J., 1963. Use of detergents in the analysis of fibrous feeds. II.

A rapid method for the determination of fiber and lignin. Journal of the

Association of Official Analytical Chemists 46, 829–835.

Zamora, N., Pennington, T.D., 2001. Guabas y cuajiniquiles de Costa Rica.

INBIO, Costa Rica. 200 pp.