Agriculture, Ecosystems and Environment 105 (2005) 595614

www.elsevier.com/locate/agee

Plant productivity in cassava-based mixed cropping

systems in Colombian hillside farms
G.C. Daellenbacha,b,*, P.C. Kerridgea, M.S. Wolfec,
E. Frossardb, M.R. Finckhd
a

Centro Internacional de Agricultura Tropical (CIAT), Cali, Colombia

b
Swiss Federal Institute of Technology (ETH), Zurich, Switzerland
c
Wakelyns Agroforestry, Fressingfield, Suffolk, UK
d
Faculty of Agriculture, University of Kassel, Witzenhausen, Germany

Received 4 October 2001; received in revised form 12 August 2004; accepted 26 August 2004

Abstract
In the Colombian hillsides cassava (Manihot esculenta Crantz) is cultivated because of its ability to produce high yields on
acidic soils poor in nutrients. Farmers often plant mixtures of cassava cultivars, while bush-beans or maize are traditionally
grown as cassava-intercrops. The objectives of this study were: (a) to determine if cassava or overall production can be improved
by planting cassava cultivar mixtures or intercropping, (b) to assess the influence of soil properties on the dry matter production
of cassava production systems, (c) to verify if soil cover can be increased by growing cultivar or species mixtures. On-farm trials
were conducted at four locations in typical hillside environments with slopes up to 55% in the Southwest of Colombia from 1996
to 1998. Two cassava varieties contrasting in plant architecture (early branching variety, rich in apices versus erect, late
branching variety, poor in apices) were grown as pure stands, as a variety mixture and each intercropped independently with
upland rice or Canavalia brasiliensis. Rainfall during the trial period was only 76% of the long term average due to the El nino
phenomenon. The cassava cultivars produced tuber yields of 9.0 and 7.5 t ha
1 DM when planted in cultivar pure stands.
Cassava growth and biomass production increased with increasing size of water stable aggregates and soil N content and
decreased with increasing soil bulk density. In the cassava cultivar mixture, competition changed the pattern of biomass
allocation, leading to a significantly lower harvest index compared to the mean of the pure stands (
6%). Intercropped C.
brasiliensis significantly reduced cassava harvest index (
13%; mean of cassava/C. brasiliensis mixtures compared to mean of
pure stands) as well as cassava (
53%) and total biomass production (
24%), while differences were not statistically significant
in the cassavarice systems probably because of the poor performance of rice. The strong reduction in cassava tuber yield in the
Abbreviations: ANOVA, analysis of variance; CIAT, Centro Internacional de Agricultura Tropical; coeff, coefficient; DAP, days after
planting; DM (kg), dry matter; DMRT, Duncans multiple range test; p, error probability level; P < 0.1(*), error probability level of 10%;
*
P < 0.05, error probability level of 5%; **P < 0.01, error probability level of 1%; ***P < 0.001, error probability level of 0.1%; ns, not
significant; r, coefficient of correlation (Pearson-r); R2, coefficient of determination; SBRA, stepwise backward multiple regression analysis;
stdv, standard deviation; stdcoeff, standard coefficient; WSA, water stable aggregates
* Corresponding author. Present address: Bio Inspecta AG, Ackerstrasse, CH-5070 Frick, Switzerland. Tel.: +41 62 865 63 20;
fax: +41 62 865 63 01.
E-mail address: georg.daellenbach@ipw.agrl.ethz.ch (G.C. Daellenbach).
0167-8809/$ see front matter # 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.agee.2004.08.009

596

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

cassava/C. brasiliensis systems was due to competition for water between cassava and the intercrop, aggravated by the lack of
rain. The percentage of soil cover was slightly higher in all mixed cropping systems compared to the pure stands. In contrast to
the mixture concept which seeks to increase productivity and soil cover compared to monocropping, the mixed cropping systems
used in the studies in Rio Cabuyal reduced cassava tuber yield and total biomass production of the cropping systems compared to
the cassava cultivar monocrops. When total soil cover was improved compared to the cassava cultivar pure stands it was
paralleled by reductions in terms of cassava tuber yield.
# 2004 Elsevier B.V. All rights reserved.
Keywords: Cassava; C. brasiliensis; Upland rice; Cultivar mixture; Intercropping; Soil structure stability; Bulk density; Diameter of
aggregates in air-dried soil; Soil cover

1. Introduction
Cassava (Manihot esculenta Crantz) is grown in
the Colombian hillsides on acidic soils because tuber
yield remains relatively high even under conditions
of degraded land with low inputs (Braun et al., 1993;
Olasantan et al., 1994). Erosion and nutrient
depletion, especially of K (Howeler and Cadavid,
1990), are at the same time the reason for and the
consequence of continuous cassava cultivation on
steep slopes.
In Colombia, farmers often plant mixtures of
cassava cultivars due to shortage of planting material
or relative geographic isolation (Lozano et al., 1980).
Gold and co-workers found that cultivar mixtures
increased cassava yield by reducing herbivore load
(Gold et al., 1989a, 1989b, 1990). Cassava with bush
bean and maize intercrops may be grown where
fertilizer is available for the intercrops (Zaffaroni et
al., 1991; Mutsaers et al., 1993; Olasantan et al.,
1996). Based on the principles outlined by Altieri
(1994) and Risch et al. (1983) mixed cropping systems
reduce susceptibility of crop(s) to pest (Ezulike and
Igwatu, 1993; Gold, 1993) and disease attacks and the
risk of total crop failure (Ahohuendo and Sarkar, 1995;
Fondong et al., 2002). Such cropping systems can also
improve soil cover and reduce erosion compared to
monocrops directly (Francis, 1990; Evangelio et al.,
1994; Leihner et al., 1996; Ruppenthal et al., 1997;
Howeler, 1998) or indirectly by increasing water
infiltration and earth worm activity (Hulugalle and
Ezumah, 1991; Hulugalle et al., 1994). Plants with
different aerial and subterranean architectures growing on the same fields might increase the resource use
efficiency for light, water, and nutrients (Mason et al.,
1986a, 1986b, 1986c; Ikeorgu and Odurukwe, 1990;
Sitompul et al., 1992). By suppressing weeds,

intercrops utilize resources otherwise economically

lost (Akobundu, 1981; Unamma and Ene, 1984; Zuofa
et al., 1992; Olasantan et al., 1994). Both is relevant
for cassava since it grows slowly in its initial
development phase making limited use of the
available resources (Mohankumar and Hrishi, 1978;
Mutsaers et al., 1993). Systems with intercropped
legumes additionally improve the fertility status of the
soil by N-fixation or by pumping nutrients from deeper
soil layers (Obiagwu, 1995; Leihner et al., 1996;
Lusembo et al., 1998). Mixed cropping systems can
also improve labor efficiency (Odurukwe and Ikeorgu,
1994), enhance the diversity of the farmers diet and
provide them with an additional income (Gold, 1993),
though these advantages are disputed (Benites et al.,
1993). Cassava cultivars and intercrop partners have to
be carefully selected concerning early development,
plant architecture, competitive ability and time to
maturity. Only fast maturing species may be planted
simultaneously with cassava (Kawano and Thung,
1982; Tsay et al., 1988; Cenpukdee and Fukai, 1991;
Ezumah and Lawson, 1990; Muhr et al., 1995; Okeke,
1996; Okoli et al., 1996; Polthanee and Anan, 1999;
Hernandez et al., 1999).
Little is known about the beneficial or detrimental
effects of cultivar or species mixtures for cassava
under on-farm conditions. The purpose of this study
was to evaluate different cassava cultivation systems
with respect to their effect on plant productivity. Onfarm trials were conducted using two contrasting types
of cassava cultivars intercropped with each other or
rice or Canavalia brasiliensis as intercrop partners.
The objectives were: (a) to determine if production
can be improved through cropping system diversification (i.e. by planting a mixture of cassava varieties, or
intercropping cassava with another crop), (b) to assess
the influence of soil properties on cassava production,

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

(c) to determine if soil cover can be increased by

growing more diverse systems.
Before the start of the trial, two meetings were
organized with farmers to define cassava cultivars,
intercrops for cassava and experimental fields. Besides
cassava, farmers were interested in upland rice and in
C. brasiliensis. Rice is an important component of the
farmers diet with some potential as a subsistence crop
in Rio Cabuyal. The idea of planting upland rice as
single row intercrop with cassava was to combine the
production of a subsistence crop with the establishment of life barriers analogous to those formed with
Vetiver grass (Vetiver zizanioides) which the farmers
are already familiar with (Ruppenthal et al., 1997).
The farmers were interested in C. brasiliensis due to
its supposed ability to repel leaf cutter ants (Atta sp.)
(CIAT, 1996b). C. brasiliensis establishes very fast
and provides early soil cover and suppression of weeds
(CIAT Tropical Forages Program, Robinson Mosquera, pers. commun., 1995). Its strong roots may
substantially reduce erosion during high-intensity
rainfalls and especially tillage erosion at harvest.
Since the plant may be cut back easily several times
during the cassava cycle without loosing its ability for
re-growth, competition of cassava may be controlled
(Akanvou et al., 2002). Even in low P soils Canavalia
spp. are able to fix between 22 and 133 kg ha
1 N
from the atmosphere (Cadisch et al., 1993; Wortmann
et al., 2000; Becker and Johnson, 1998).

2. Materials and methods

2.1. Experimental sites
Experiments were carried out for one cassava
cropping cycle from November 1996 to January 1998
at four locations at altitudes between 1425 and 1625 m
above sea level. Sites 1 (Julio) and 2 (Lizardo) were in
the middle-upper, sites 3 (Jeremias) and 4 (Merardo)
in the lower portion of the watershed of Rio Cabuyal,
situated in Cauca Department in Southwest Colombia
(28470 N, 768320 W). Selected characteristics of these
sites are given in Table 1. The soils in Julio and
Lizardo were classified as Typic Dystrandepts while the
soils in Jeremias and Merardo were oxic Dystropepts
(USDA, 1998). Selected physico-chemical characteristics of these soils are presented in Table 2. Available

597

phosphorus content at all sites was extremely low and

probably the most limiting nutrient for cassava
production.
Based on the classification of life zones (Holdridge,
1967), Rio Cabuyal lies in the tropical premontane
moist forest zone. The annual mean temperature and
the annual mean precipitation are 20 8C and 2066 mm,
respectively. There are two dry periods, from June to
August and from December to February (Gijsman and
Sanz, 1998). Based on its environmental and socioeconomic characteristics this watershed had been
chosen as a representative pilot area for the hillside
areas in Latin America (CIAT, 1996a).
2.2. Plant material and experimental set up
Two cassava cultivars with contrasting growth
patterns were selected for the trials. They were the
early branching CG 402-11 forming a large number of
apices and the erect, late branching SM 526-3 which
develops only a few apices. Both had produced high
tuber yields, i.e. 810 t ha
1 DM in variety trials
under hillside conditions (Jaramillo, 1994). Upland
rice var. Fofifa 62 (origin Madagascar)  Shin-Ei 3
(origin Japan) with cycle length of 123 days was sown
in a single row at a rate of 5 g m
1 (equals one quarter
of normal seed rate, i.e. 25 kg ha
1 instead of
100 kg ha
1). Based on previous results (Robinson,
1997; Jalloh et al., 1994; Dahniya et al., 1994) and
taking into account soil and climatic conditions in Rio
Cabuyal favoring cassava over rice, rice and cassava
were planted at the same time.
C. brasiliensis, CIAT-Accession No. 17009, known
to deal well with low-P conditions was sown in a
single row at a distance of 80 cm in the row (three
grains of C. brasiliensis per planting hole).
Cassava was fertilized at planting with 0.5 kg of
chicken manure containing 1.8% N, 1.4% P, and 1.8%
K (on a DM basis, CIAT analytical services) per plant
as is traditionally done in this region. The respective
amounts of nutrients on an area basis were
113 kg ha
1 N and K, and 88 kg ha
1 P. Intercrops
did not receive additional fertilizer.
There were seven treatments (Table 3) replicated
twice at each of the four locations in a randomized
complete block design. Soil preparation and length of
prior fallow periods were different between the
experimental sites (Table 1). Sixteen rows of five

598

Table 1
Characteristics and plot history of trial sites
Soil classificationa

Usual agronomic practices

and pest or disease incidence

Crops produced prior to

trial (rest of time under
spontaneous fallow); length of fallow
period before experiment

Soil preparation for the experiment

Julio (1), 2943% (37%)

Typic Dystrandept

June 1995July 1996: cassavabean

intercrop. Fallow length: 3 months

Manual hoeing to 10 cm (only area

of cassava planting)

Lizardo (2), 3757% (44%)

Typic Dystrandept

Ploughing with oxen; fertilizer,

chicken manure, 0.5 kg cassava
per plant; ant and whitefly attack
Manual hoeing; fertilizer, chicken
manure, 0.5 kg cassava per plant;
yield losses through ant damage

September 1993October 1994:

cassava; November 1994December
1995: cassava. Fallow
length: 10 months
19891991: cassava. Fallow
length: 5 years
March 1994April 1995: cassava.
Five plots were last cultivated in 1990.
Fallow length: 18 months/5 years

Ploughing with oxen to 15 cm

Jeremias (3), 1631% (24%) Oxic Dystropept

Ploughing with oxen; no fertilization

Merardo (4), 1437% (27%)

Ploughing with oxen; no fertilization

Oxic Dystropept

Ploughing with oxen to 15 cm

Ploughing with oxen to 15 cm

USDA soil taxonomy (1998).

Table 2
Soil characteristics of experimental sitesa, depth 020 cm
Parameters

Julio (1)
Typic Dystrandept

Lizardo (2)
Typic Dystrandept

Jeremias (3)
Oxic Dystropept

Merardo (4)
Oxic Dystropept

Sand (g kg
1 soil)

Clay (g kg
1 soil)
Bulk density (g cm
3)
Air-dried aggregates (<6.3 > 2 mm), % (w)
Water stable aggregates, % (w)
Total N content (g kg
1 soil)
OM content (g kg
1 soil)
P content (Bray II) (mg kg
1 soil)
K content (Bray II) (cmol kg
1)

380  30
310  28
0.76  0.05
42  5.8
91  3
4  1.03
82  12
1.94  0.8
0.27  0.08

410  26
240  13
0.69  0.03
32  5.8
87  3
4.7  0.94
116  21
0.96  0.3
0.28  0.05

110  46
660  41
1.07  0.04
71  4.6
93  3
1.9  0.27
46  7
3.1  0.8
0.19  0.05

220  80
600  92
0.95  0.08
62  2.7
95  2
3.1  0.86
72  11
2.3  0.5
0.26  0.12

Mean values  standard deviation.

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Site name, slope range

(average)

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614
Table 3
Cassava cultivars and intercropping species selected for the on-farm
trials
Treatment

Cropping system

1
2
3
4
5
6

CG 402-11 pure stand (CG)

SM 526-3 pure stand (SM)
Variety mixture of CG and SM (CG/SM)
Species mixture CG + ricea (CGrice)
Species mixture SM + rice (SMrice)
Species mixture CG + C. brasiliensisb
(CGC. brasiliensis)
Species mixture SM + C. brasiliensis
(SMC. brasiliensis)

7
a
b

Upland rice Oryza sativa var. Fofifa 62  Shin-Ei 3 was used.

Canavalia brasiliensis CIAT 17009 was used.

1520 cm long cassava cuttings (approximately 10

nodes) were planted on contour at a distance of
0.8 m  1.0 m in plots of 4.5 m  17 m. At the same
time intercropping species were manually sown in
contour rows halfway between the cassava rows.
2.3. Methods
2.3.1. Climate data
Rainfall and temperature data were collected on a
daily basis at a CIAT experimental site in Pescador in
the central part of Rio Cabuyal watershed at a distance
of 2 and 4 km from the experimental locations in the
middle-upper and in the lower part of the watershed,
respectively (Jorge Rubiano, CIAT, pers. commun.).
2.3.2. Cassava growth analysis
Commencing at 90 days after planting (DAP),
cassava plant height, number of apices per plant and
rate of leaf production per apex were measured biweekly on 10 plants per variety and plot over the
whole cassava growing cycle, omitting the border
rows.
2.3.3. Biomass production of the different
system components
Cassava was harvested 15 months after planting to
determine fresh tuber yield and biomass per plot and
per-cultivar. Dry matter production of cassava tuber
and stem was measured on 3 kg subsamples after
drying at 60 8C in an oven. At site Julio cassava
yields were extrapolated based on yields of
completely harvestable rows. At this site, between

599

4 and 28 plants per plot (12 plants on an average) of a

total of 45 plants relevant for biomass measurement
had been lost due to a massive pest attack (white grub,
Scarabaeidae sp.).
C. brasiliensis was cut back twice before cassava
harvest to reduce competition to cassava growth. Plant
residues were left in the plots. Biomass fresh weight of
C. brasiliensis was determined from samples from
2.5 m  3 m, 3 and 6 months after planting, and from
the whole plots at cassava harvest. Dry weight was
determined after drying at 60 8C in an oven, on the
basis of the sample from 2.5 m  3 m for the
intermediate cuts and on a subsample of 5 kg at
harvest. Rice panicles were harvested 140 DAP.
2.3.4. Chemical soil properties
Three bulk soil samples, 020 cm depth, were
collected 2 weeks after planting along the diagonal of
each plot (4.5 m  17 m) in the interrows to avoid the
fertilized zone. Soil samples were air-dried, sieved at
2 mm, and analysed to determine organic matter
content (Nelson and Sommers, 1982), total N (Krom,
1980), 0.1 M HCl + 0.03 M NH4F (Bray II method)
extractable P and K (Salinas and Garcia, 1985), 1 M
KCl extractable Ca, Mg, Al (Thomas, 1982), 0.05 M
HCl + 0.0125 M H2SO4 extractable zinc (Martens
and Lindsay, 1990), soil texture (Bouyoucos, 1962)
and pHH2O (Salinas and Garcia, 1985). Analyses
were performed on each sample and averaged for each
plot.
2.3.5. Physical soil properties
Undisturbed soil cores, 5 cm in diameter and 2.5 or
5 cm in length, were taken 14 months after planting
cassava at the four sites for the determination of soil
physical characteristics. The following parameters
were determined using the methodologies indicated in
brackets: (1) compressibility (Klute, 1986; Culley,
1993), (2) hydraulic conductivity (Reynolds, 1993),
(3) air permeability (Weeks, 1978), (4) total porosity
and pore size distribution (Topp et al., 1993; Klute,
1986), (5) permeability ka was indirectly determined
with a DIK-5001 permeameter (Daiki Rika Kogyo
Co., Tokyo, Japan).
Additional bulk soil samples, of 1 kg fresh soil,
were taken 14 months after planting cassava from 05
and 520 cm depths with two replications per
experimental plot. Samples were air-dried for 3 days

600

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

for the aggregate analysis. Seven aggregate size

classes were separated using an electric Sieve Shaker
(CSC Scientific Company, Fairfax, VA, USA).
Aggregates with a diameter between 2 and 4 mm
were subsequently used to determine the net weight of
water stable aggregates (WSA) and the geometric
mean diameter (GMD) of the WSA with the humid
sieving technique (E. Amezquita, CIAT, pers. commun.) applying an apparatus similar to that described
by Bourget and Kemp (1957).
2.3.6. Soil cover and soil erosion
Soil cover was assessed using a sighting frame as
described by Elwell and Stocking (1974). Cover was
expressed as a percentage of covered sights of the
total number of 351 observations (i.e. nine frame
lengths) per plot. Cover measurements were carried
out bi-weekly until 200 DAP when the height of
cassava plants made the use of the frame impossible.
Total soil cover over the observation period, i.e. from
planting until 200 DAP was estimated based on the
area under the curve.
Eroded soil and runoff water were collected in a
plastic-lined channel at the bottom of the plots. While
the sites Julio and Lizardo (middle-upper watershed)
were furnished completely with collection channels
(i.e. one channel/plot, 14 channels per site), channels
were dug only for one repetition per site (seven
channels per site) at sites Merardo and Jeremias in the
lower part. Excess water was pumped off when
necessary and dry weight of eroded soil was
determined (a) for the first 90 days and (b) for the
total length of the cropping season. At the Julio site,
soil cover and erosion measurements were discarded
90 DAP due to a massive pest attack (white grub,
Scarabaeidae sp.) which caused the total loss of a
substantial number of cassava plants in five plots at
this site.
2.4. Data processing and statistical analysis
Relative values of plant growth parameters for all
treatments were calculated on a per site basis by
dividing the observed value by the value for the
respective cultivar grown in pure stand. The use of
relative values allowed direct comparison of effects of
competition on both cultivars in the different
treatments.

Cassava top growth was assessed by repeatedly

measuring the parameters plant height, number of
apices per plant and leaf formation rate over the whole
cultivation cycle. These data were depicted as graphs
of the following function
y f timeDAP

(1)

where y is the plant height, number of apices per plant

or leaf formation rate of cassava.
In order to convert repeated measurement values
into a single number, the area under the curve, i.e. the
area between the x-axis and the graph of the respective
function, was calculated for each of the three growth
parameters. Relative cassava top growth represents
the mean value of the relative areas under the curve
(i.e. value for the respective pure stand was set = 1) for
the parameters plant height, number of apices per
plant, and leaf formation rate.
Data were analysed using the Statistical Analysis
Systems (SAS, 1988) and SYSTAT (SPSS, 1998)
software. The former was applied to assess the effect
of treatments and sites on cassava growth, biomass
production of the cropping system components (i.e.
cassava, rice, and C. brasiliensis), and soil cover.
SYSTAT was used to determine the effect of soil
physico-chemical properties on cassava growth and on
biomass production of the cropping system components.
Firstly, ANOVAs, Duncans multiple range tests
and linear contrasts were performed separately for
each of the four experimental sites (Gomez and
Gomez, 1984) to assess the total treatment effect and
to compare growth, production, and soil cover of the
different treatments. Subsequently, the same statistical
tools were applied carrying out a combined analysis
over all sites for the growth and production data and
over three sites for soil cover data (discarded at site
Julio) to investigate differences between the experimental locations and site  treatment interactions.
Site was considered as a fixed variable.
In the final step of data analysis, stepwise backward
multiple regression analysis (SBRA) was used to
simultaneously estimate the effect of selected soil
parameters on growth and production of the tested
cropping systems. The SBRA was carried out to
replace the categoric variable replication in the
ANOVA by continuous variables of agronomic
relevance.

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

In the SBRA model, the observed values for the

dependent parameters are explained by k independent
parameters and described by Eq. (2):
yobs csite1
4 ctreatment1
7
soil parameter1  c1
soil parameter2  c2
soil parameterk  ck errorresidual (2)
where csite, ctreatment, and ck are the category-specific
and parameter-specific constants, respectively.
In order to pre-select the variables entered into the
SBRA, data were subjected to a discriminant analysis.
Thirty-four of 35 original soil parameters could be
represented by 11 proxies based on significant pooled
within-class correlations (jrj > 0.27, P < 0.05) (data not
shown). Variables were only replaced if the per-site
correlations were consistently positive or negative at all
four experimental sites. Eleven proxies and the
parameter share of waterstable aggregates, 520 cm
depth, which had no significant correlation (P < 0.05) to
any of the other parameters, i.e. 12 independent
parameters in total were fed into the SBRA. Only
two chemical properties were used as proxies, soil N and
Ca content. The remaining main (P and K) and
secondary plant nutrients (Mg and Zn) are highly
correlated either to soil bulk density (520 cm) or to Ca
content (data not shown). Original values of soil
chemical parameters and the residual values of soil
physical parameters were used as independent variables
in the SBRA. Physical parameters had been measured at
the end of the cropping cycle and were therefore
corrected by subtracting the treatment effect from the
original values, i.e. calculating the residuals from the
ANOVA. Original values of growth and production
parameters of the cropping systems were used as
dependent variables. Since cassava and cropping system
growth and production were not only affected by the
parameters used in the SBRA, but also by the parameters
replaced by proxies (data not shown), the correlations
which were relevant for the identification of the latter
were taken into account for the discussion.
To simplify the discussion about the effect of the soil
parameters replaced by proxies on production, parameter pairs not varying significantly between depth
levels were merged into one variable per pair. The same
was done where correlations between parameters were
significant on both depth levels 05 and 520 cm, i.e.

601

compressibility, % share of aggregates <1 mm in airdried soil, mean diameter of WSA, % shares of macro,
meso, and micropores and bulk density.
SBRA defined the significance and the relevance of the effect of simultaneous changes of
various soil parameters on cassava and cropping
system growth and production (Hair et al., 1992;
Jambu, 1991). Significance measures the change of
y caused by a one unit change of x and is indicated in
the form of a coefficient (i.e. regression slope) per
parameter. Relevance is indicated in the form of a
standard coefficient and measures the change of y
relative to its standard deviation (stdvy) caused by a
change of x by one standard deviation (stdvx) (Eq. (3))
Dy
Dx

standard coefficient  error term

stdvy stdvx
(3)

3. Results and discussion

3.1. Weather conditions and soil erosion
Climate in Rio Cabuyal during the trial period was
influenced by the El nin o phenomenon with rainfall
being less than normal. While the average air
temperature (19.7 8C) was close to the long term
average of 20 8C, total rainfall during the cassava
cropping cycle (November 1996 to January 1998) was
2800 mm, 76% of the long term average of 3700 mm
for the same 15-month period (Fraiture et al., 1997).
Only 5 mm of rain, 3% of the long term average, fell
between July and August 1997 (between 240 and 300
DAP cassava; Jorge Rubiano, CIAT, pers. commun.).
Leaf formation was reduced greatly at all sites after 240
DAP but increased again after 300 DAP as rainfall
resumed (Fig. 1). Beside the reduction of total rainfall,
the climate during the cropping cycle reported here was
characterized by the absence of high-intensity rains
(>100 mm h
1). The estimated rates of total soil loss
during the cropping cycle, i.e. 1.9 t ha
1 dry soil in the
upper watershed (based on site Lizardo) and 0.6 t ha
1
dry soil in the lower watershed, respectively were about
10 times lower compared to those measured by other
workers in the same area under similar soil conditions
(Reining, 1992). Erosion data were not further analysed
due to the lack of relevant rainfall events during the

602

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Fig. 1. Daily rainfall and rate of cassava leaf formation (average over all treatments) at the four sites. Reduction of leaf formation at site Lizardo
prior to 160 DAP as caused by ants (Linepithema sp.).

observation period and due to the pest attack leading to

total cassava plant loss at site Julio.
3.2. Crop performance
Over all sites (including site Julio, where yields of
plots with missing plants were extrapolated based on

yields of complete rows (see Section 2.3.3), mean

cassava tuber yields in the cultivar pure stands were
9.0 and 7.5 t ha
1 DM for the cultivars CG and SM,
respectively (Fig. 2). These yields were larger than the
average tuber yield of a wide range of improved
cultivars (5.1 t ha
1 DM) grown under similar soil and
fertilization conditions (Jaramillo, 1994). This sug-

Fig. 2. Performance of cassava cropping systems over all trial sites: component and total biomass production; harvest indices of cropping
systems. Columns and column segments of the same colour and pattern marked with different letters are of different size at P < 0.05 (Duncans
MRT).

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

gests that the lower rainfall in this cropping cycle had

no negative influence on the tuber yield of the cultivars
CG and SM.
On the average over all sites, the total biomass
production of rice was only 0.1 t ha
1 DM. In
contrast, rice grain yields for the same variety in
intensively managed monocrop plots, grown in Rio
Cabuyal at the same period in an earlier season and on
soils comparable with the sites Julio and Lizardo, were
between 1.6 and 2.3 t ha
1 DM (Quiros, 1996). We
observed that rice grew well at site Julio with the
lowest clay and the highest OM content among the
experimental fields. The rice biomass production at
site Julio was 200 kg ha
1on average over the four
cassava/rice plots, reaching 400 kg ha
1 in the best
plot which is still 812 times lower compared to the
rice variety trials described by Quiros (1996). The fact
that the rice had been sown at a lower seed rate, i.e.
25 kg ha
1 instead of 100 kg ha
1, cannot explain the
very low production. Since our results showed a
reduced cassava production in the cassava/rice
intercrop plots (see Section 3.4), we supposed a
strong negative competition effect of cassava on rice.
Conditions for rice growth were suboptimal in most
experimental plots since the nutrient status was low,
there was no additional fertilizer for the intercrops and
clay content in the lower watershed was too high.
Germination of rice was supposedly hindered by
insufficient contact of the seeds with the soil at the
sites where plots were ploughed with oxen (see Table
1). In addition, rice growth was severely hampered by
repeated bird attacks at each of the four sites. The

603

results showed that with low input practices typical for

cassava plots most soil characteristics-soil preparation
combinations allow for a negligible rice intercrop
production only leading to grain yields 1050 times
lower compared to intensely managed monocrop plots.
C. brasiliensis developed rapidly and covered the
soil between the cassava rows. The legume performed
best at site Julio with a P-concentration of only
1.9 mg kg
1 soil. Mean biomass production over all
sites in Rio Cabuyal was 3.6 t ha
1 DM. There is no
literature on C. brasiliensis production on soils
comparable to Rio Cabuyal, but C. brasiliensis pure
stands on the Cerrado soils in Brazil, when not cut
back, produced 5 t ha
1 DM (Lathwell, 1990).
3.3. Site effect on cassava performance and soil
cover over all cropping systems
Dry matter production of cassava shoots and tubers,
relative cassava top growth and total biomass were
very variable at each of the four sites. Cassava harvest
index (average over all cropping systems) at site
Lizardo was significantly lower compared to the sites
Jeremias and Julio (Table 4). Per-cultivar analyses for
CG and SM showed that at site Lizardo, SM tuber
yield and harvest index were significantly lower
compared to the other sites (Table 4), while the
performance of CG did not differ significantly
between the sites. This was due to heavy attacks of
small ants (Linepithema sp.) on cultivar SM which
reduced the rate of formation of leaves until 200 DAP
at site Lizardo (Fig. 1). The ants acted as a competing

Table 4
Mean cassava performancea at different experimental sites (mean effect over both cassava cultivars CG and SM (per plot) and on each of two
varieties (CG and SM) separately)
Site

Julio
Lizardo
Jeremias
Merardo
a

Cassava shoots
(t ha
1 DM)

Cassava tubers
(t ha
1 DM)

Harvest indexb (excl.

fallen leaves)

Total biomass of cropping

system (t ha
1 DM)

Relative cassava
top growthc

Plot

Plot

Variety

Plot

Variety

Plot

Plot

Variety

CG

SM

CG

SM

CG

SM

7.2a
4.1a
7.9a
7.2a

8.2a
5.2a
7.5a
7.5a

6.4a
3.1b
8.5a
7.4a

0.70a
0.50b
0.71a
0.61ba

0.71a
0.58a
0.70a
0.57a

0.69a
0.42b
0.72a
0.65a

11.6a
7.7a
12.0a
12.4a

0.92a
1.03a
1.04a
0.91a

0.87a
1.09a
1.08a
0.84a

0.98a
0.99a
1.03a
1.02a

2.7ad
3.6ad
3.2ad
4.5ad

Variety
CG

SM

2.9a
3.6a
3.4a
5.6a

2.6a
3.8a
3.3a
4.0a

Averaged across all treatments.

Tuber yield/cassava biomass.
c
Relative area under curve, see Section 2.4 for explications; no difference between pure stands by definition.
d
Numbers within a column followed by different letters are significantly different (differences between experimental sites, P < 0.05,
DMRT).
b

Based on linear contrast analysis.

Excluding fallen cassava leaves.
Area under curve.
Relative area under curve, see Section 2.4 for explications; no difference between pure stands by definition of relative value.
Mean of the pure stands (expected value).
+ means A > B;
means A < B.
*
, (*), **, ***: contrast significant at P < 0.05, P < 0.1, P < 0.01, P < 0.001, respectively.
b

+f
+
+(*)
+
+***
+***
Mean p.s.e
CG
CG
SM
CG
SM

f
+
+
+
+*
+
B

Cultivar mix
SM
CGrice
SMrice
CGC. brasiliensis
SMC. brasiliensis

+***
+


+**
+


+
+
+
+*
+*
+(*)g
+
+


+*
+***

+
+
+
+*
+

Cassava
number of
apices per plantc
Total biomass
crop. system
(t ha
1 DM)
Harvest
indexb
Cassava tubers
(t ha
1 DM)
Cassava shoots
(t ha
1 DM)

Table 5
Differencesa of performance (A minus B) between cropping systems across four experimental sites

Cassava
plant
heightc

3.4. Treatment effect on cassava and cropping system

performance and soil cover across the four sites
As expected from the choice of cassava cultivars,
cultivar CG in pure stand produced significantly more
apices (P < 0.001) compared to cultivar SM (Table 5).
Conversely, production of leaves was significantly
higher (P < 0.05) in SM compared to CG. However,
production parameters at harvest (tuber and shoot dry
matter) did not differ significantly between CG and
SM (Table 5, Fig. 2).
In the cultivar mixture, tuber and shoot dry matter
production was lower and higher, respectively as
compared to the mean of the pure stands (Table 5, Fig.
2). Linear contrast analysis showed a significant shift
in biomass allocation from tubers to shoots, which was
reflected in a lower harvest index for the cultivar
mixture (P < 0.10) compared to the mean of the pure
stands (Table 5).
Separate analyses based on DMRT for the cultivars
CG and SM presented in Table 6 showed different
reaction patterns of the cultivars when planted
together in cultivar mixture. For cultivar CG, all
parameters except for shoot biomass were slightly
reduced, while cultivar SM increased shoot and total
biomass production, formed more apices and grew
taller compared to the respective pure stands (Table 6).
The differences found when contrasting the cultivar
mixture versus the mean of the pure stands (Table 5)
were thus due to changes in SM shoot growth. The
growth and production of aerial parts in cultivar CG
planted in mixture with cultivar SM were more
resilient compared to SM to the changed patterns of

*
+
+
+(*)
+**

Cassava leaf
formation ratec

nutrient sink without affecting cassava shoot production, as previously observed by (Gold, 1993).
At site Lizardo, total soil cover over the observation
period was significantly reduced compared to the sites
Jeremias and Merardo. Four out of seven cover
measurements between 100 and 200 DAP showed the
same ranking of the locations (DMRT, P < 0.05, data
not shown). Slower establishment and growth of C.
brasiliensis at site Lizardo compared to Jeremias and
Merardo explained these findings. At site Julio, cover
measurements had been discarded 90 days DAP
because of massive loss of cassava plants in several
experimental plots.

Not analysed
+
+
+(*)
+(*)

Relative
cassava
top growthd

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Cropping systems

604

CG-based systems
CG pure stand
CG in cultivar mix
CG/rice
CG/C. brasiliensis
SM based systems
SM pure stand
SM in cultivar mix
SM/rice
SM/C. brasiliensis
Cultivar mixturee
a
b
c
d
e

Cassava shoots
(t ha
1 DM)

Cassava tubers
(t ha
1 DM)

Harvest
indexa

Total
biomass for
cropping system
(t ha
1 DM)

Cassava plant
heightb

Cassava number
of apices per
plantb

Cassava leaf
formation rateb

Relative
cassava top
growthc

4.3ad
4.7a
3.6a
2.8a

9.0a
7.9a
7.2a
4.3b

0.67a
0.63a
0.67a
0.60a

13.3a
12.6a
11.0a
9.7a

2191a
2171a
2111a
1842a

220a
210ab
202ab
162b

642a
636ab
633ab
597b

1.00a
1.01a
0.99a
0.87a

3.6ab
4.4a
3.3ab
2.5b

7.5a
7.5a
6.9a
3.5b

0.65a
0.61ab
0.67a
0.55b

11.1ab
11.9a
10.3ab
8.9b

2000ab
2247a
1856b
1672b

50ab
68a
52ab
41b

711a
696a
670a
623b

1.00ab
1.17a
0.99b
0.87b

4.5ns

7.7ns

0.62(*)

12.2ns

2209ns

139ns

666ns

1.09ns

Excluding fallen cassava leaves.

Area under curve.
Relative area under curve, see Section 2.4 for explications for explications.
Treatments belonging to the same cultivar group followed by the same character are not significantly different.
The cultivar mixture was compared to the mean of the pure stands: (*): linear contrast significant at P < 0.1.

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Table 6
Per-cultivar analysis: comparisons among the four cropping systems containing the same cultivar (for cultivars CG and SM; mean across four sites)

605

606

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

shading and competition for aerial space relative to the

pure stands.
While competition for light in the cultivar mixture
led to increased shoot production in both cultivars, the
net effect was negative concerning total tuber yield.
The harvest indices in both cultivars were decreased as
compared to the pure stands. While tuber yield of
cultivar SM grown together with CG was the same
compared to the SM pure stand, the mentioned shift in
assimilate distribution led to a reduction of tuber yield
in cultivar CG as compared to the CG pure stand and to
a reduction of total tuber yield in the cultivar mixture
compared to the mean of the pure stands. The different
reaction patterns of CG and SM confirmed the results
of Kawano and Thung (1982). They found that the
essential part of competition in cassava was for light.
While competitive ability and spacing response (e.g.
occupation of free space through formation of new
apices and branches like in cultivar SM) were found to
be strongly positively correlated, the correlation
between competitive ability and harvest index was
negative (Kawano and Thung, 1982). Our results
concerning the mixture of two cassava cultivars with
different competitive abilities may be explained in two
ways. Firstly the majority of the total amount of
assimilates produced by both cultivars was drained
to the cultivar with the higher competitive ability and
the lower harvest index, the productivity in terms of
tuber yield decreased. Secondly there was a shift in
assimilate distribution in both cultivars from the tubers
to the shoots which was triggered in the stronger
cultivar by the spacing response and in the weaker
by the stress induced by the competing cultivar.
In all interspecies mixtures (i.e. cassava with C.
brasiliensis or upland rice), cassava growth and
production parameters were reduced compared to
the pure stands (Tables 5 and 6). In combination with
C. brasiliensis, all parameters in both cultivars were
reduced compared to the respective pure stands, with
significant differences for cassava shoot and tuber
production, harvest index, total biomass production of
the cultivation system, cassava plant height (P <
0.05), leaf formation rate and relative top growth (P <
0.10; Table 5). In addition, the number of cassava
apices per plant in CG was significantly reduced in the
presence of C. brasiliensis (Tables 5 and 6).
Linear contrasts showed that despite the low rice
production, cassava tuber production (P < 0.10) and

harvest index (ns) of cultivar CG intercropped with

rice were lower compared to the pure stand (Table 5).
There were no such effects in the SM/rice system,
possibly due to the more erect growth habit of SM
compared to CG which reduced competition from the
low growing rice. Generally spoken, reductions of
cassava growth and production in the interspecific
mixtures with rice or C. brasiliensis were more severe
in cultivar CG compared to SM.
The decrease of total biomass production due to
intercropping C. brasiliensis was likely due to strong
interspecific competition for water induced by the
higher evapo-transpiration rates of these systems
compared to the cassava monocrop and aggravated by
the dry summer of 1997. Our findings concerning the
shift in assimilate distribution in the cassava plant
support the results reported by Gold et al. (1989a,
1989b) and by Cock et al. (1979a), who found that
under stress cassava favored top growth, first fulfilling
the requirements of the energy producing structures
before filling the storage tubers.
While this mechanism may reduce the production
in terms of tuber yield, the plasticity of its above
ground organs enables cassava to quickly respond to
biotic and abiotic stresses such as pests and diseases,
drought, and competition by weeds or intercrop
partners.
Since the biomass production per unit nutrient and
water uptake in cassava is higher compared to other
crops (Howeler and Cadavid, 1990), the intercrop
biomass was not sufficient to make up for the loss in
total cassava biomass, not even in the case of the
competitive C. brasiliensis.
Concerning soil cover at 100 DAP values were in
between 45 and 63% depending on the cropping
system (Fig. 3). Soil cover in both cassava/rice
systems was significantly higher compared to the
respective pure stands (linear contrasts, P < 0.05).
Before the second cover measurement fields were
weeded and C. brasiliensis was cut back, causing
decreases in soil cover of between 13 and 35%,
depending on the cropping system. Soil cover
decrease after cutting the intercrop and increase
through re-growth were most drastic in the CG/C.
brasiliensis system, as the CG pure stand provided the
least soil cover of all cropping systems.
Despite its erect growth, the SM pure stand
provided better soil protection compared to the CG

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

607

Fig. 3. Percent of soil cover in different cassava cropping systems, 100200 days after cassava planting (average over three experimental sites).
(1) Significant differences between treatments based on linear contrasts (*P < 0.05, **P < 0.01). (2) Significant differences between treatments
based on DMRT (*P < 0.05).

pure stand at 150 and 194 DAP. This could be

explained by its higher leaf production rate and larger
individual leaf size. Due to the increased soil cover in
the SM pure stand compared to the CG pure stand,
differences in soil cover between the intercropping
systems and the respective pure stands were larger and
more frequent in the CG-based systems.
Based on the area under the curve, soil cover over
the whole observation period was slightly higher in all
mixed cropping systems compared to the respective
pure stands and higher in the SM pure stand compared
to the CG pure stand. However, the only significant
difference measured was between CG/rice (57% total
soil cover) and the CG pure stand (44% total soil
cover) (linear contrast, P < 0.05, data not shown).
The gain in total soil cover in the CG/rice system
compared to the CG pure stand was linked to a
significant cassava tuber yield loss of 20% (linear
contrast, P < 0.10) (Table 6). In all other mixed
cropping systems insignificant soil cover increases of
between 10 and 13% were paralleled by cassava tuber
yield losses of between 7 and 54% compared to the
respective pure stands. Yield losses were highly
significant for both cassava/C. brasiliensis systems
(Tables 5 and 6). The trade-off of reduced cassava
tuber yield for increased soil cover, as observed in the
mixed cropping systems is highly disadvantageous for
farmers, especially for the cassava/rice and cassava/C.
brasiliensis systems. Besides the yield loss in cassava,
the additional costs and time for seed and more careful

soil preparation and weeding have to be taken into

account. In contrast to the results presented here,
Howeler (1998), working with farmers in Asia, has
demonstrated that a combination of promoting more
rapid growth of cassava through efficient fertilization,
short-term cash intercrops, that mature before canopy
closure and competition for moisture, e.g. peanuts,
and erosion barriers lead to high yielding sustainable
cassava-based systems.
3.5. Effects of soil physical and chemical
characteristics on growth and production of cassava
cropping systems
In a first approach, the stepwise backward multiple
regression analysis (SBRA) had shown highly significant negative correlations between the dependent
parameters and the clay content. The relevance of clay
content had been up to three times higher compared to
the second most important variable. Plotting of the
residuals which were the basis for the regression
calculations (original data reduced by site and treatment
effects), i.e. dependent parameters versus clay content,
revealed that these regressions were dominantly
influenced by data from the experimental plots of site
4 (Merardo) (Fig. 4). Five plots at site Merardo had been
under fallow for a longer period of time compared to the
other plots of the same site before the start of the
experiment (Table 1). These five plots had considerably
lower clay contents than all others and cassava yields

608

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Fig. 4. Regression lines over four and three (without site Merardo) experimental sites, respectively of clay content of soil (020 cm depth) on
cassava stick production. x = 0 and y = 0, respectively are the means across all observations (56 and 42, respectively).

were much higher than in other plots (Fig. 5). For this
reason, the Merardo site was subsequently excluded
from this analysis and the results of the SBRA are based
on three sites and 42 observations (Fig. 3, Table 7). The
SBRA showed that the N content of the 020 cm
horizon, the mean diameter of WSA of the 05 cm
horizon, and the bulk density of the 05 and 520 cm
horizons, respectively, had a significant effect on one or

more of the dependent variables and that eight

regressions of soil parameters on dependent variables
were statistically significant (P < 0.05) (Table 7). All
these correlations showed similar trends at the three sites
and were therefore further considered for the interpretation.
Soil parameters explained between 7 and 21%
(measured as % share of the total sum of squares) of the

Fig. 5. Cassava tuber yield (residual, effect of cropping system subtracted) and clay content in plots with long (5 years, A) and short (18 months,
B) fallow at experimental site Merardo (lower watershed).

Independent (soil) parameters

N content, 020 cm
(g kg
1 soil)

Diameter of WSA,
Bulk density,
depth d = 05 cm (mm) d = 05 cm
(g cm
3)

Bulk density,
d = 520 cm
(g cm
3)

Regression
Beta (P)c Regression
Beta (P) Regression
Beta (P) Regression
coefficient (b)b
coefficient (b)
coefficient (b)
coefficient (b)
Cassava shoots (kg ha
1 DM);
R2 = 0.38/0.21d
Total biomass of cult. system
(kg ha
1 DM); R2 = 0.50/0.10d
Cassava plant height (cm); R2 = 0.34/0.19d
Cassava, number of apices P/plant;
17.3  7
R2 = 0.88/0.07d
Cassava leaf formation ratee;
R2 = 0.50/0.14d
a

4479  3529 0.35*

Percentage share of
aggregates < 1 mm
in air-dried soil,
d = 520 cma
Beta (P) Regression
Beta (P)
coefficient (b)

6565  5845
1.10*

11241  8553 0.31*

1037  968
0.30*

0.30*

1917  1587
1.22*

458  141

0.87**

418  118

1.60***

On a w/w basis.
Coefficient: measures the significance, i.e. the change of the target variable caused by a change of the independent variable by one unit (mathematically: Dy = Dx  coefficient).
c
Beta: measures relative importance of the independent variables (Beta = b  Sx/Sy). Absolute values have to be taken. For example
1.10 is more important than 0.35. Values are
to be compared within a dependent variable only. Confidence level: *P <0.05; **P < 0.01; ***P < 0.001.
d
R2 value for model including categoric variables (site and treatment)/R2 value for model with soil parameters only.
e
New leaves per apex/21d.
b

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Table 7
Relevance and significance of effect of soil parameters that significantly affected cassava and cropping system performance based on data of three experimental sites (without site
Merardo)

609

610

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

total variability in growth and production parameters in

cassava cropping systems (Table 7). Cassava tuber yield
was not affected by the soil parameters at all.
Mean diameter of WSA of the top 5 cm of the soil
was positively correlated to the production of cassava
shoots, total biomass production of the cropping
system, and cassava plant height.
Soil bulk density (520 cm depth) was negatively
correlated to cassava shoot production, plant height
and leaf production rate, while bulk density (05 cm
depth) was negatively correlated to the number of
apices per plant.
An increased total N content enhanced the number
of apices per cassava plant.
Combining the results of the discriminant analysis
(i.e. selection of the 11 proxies) with those of the
SBRA suggested that soils with a low bulk density, a
small size of aggregates in air-dried soil, a high
percentage of macropores, a high water conductivity, a
coarse texture, a high percentage of large WSA, high
levels of pH, organic matter, N, Ca, Mg, and K
increased plant growth and production compared to
soils with the opposite characteristics.
The importance of aggregate stability for soil
fertility and crop production has already been shown
in a number of trials from the temperate and tropical
zones (Karlen et al., 1992; Reining, 1992). These
aggregates provide good growing conditions during
phases of increased rainfall by preventing soil
puddling and reducing soil erosion (Balesdent, 1997).
Soils with low bulk density and a high macropore/
micropore ratio exert little physical resistance on
tubers and increase cassava above ground growth and
production due to a feedback mechanism described by
Masle (1999) which reduces above ground growth as
soon as root growth gets impeded.
Since bulk density was negatively correlated to
both soil macroporosity and water conductivity (data
not shown) the negative correlations observed
between soil bulk density and cassava top growth
parameters underlined the importance for cassava
growth of both good aeration and rapid drainage of
excess water as observed by Rehm (1991). The
positive response of cassava above ground growth to
increased N (increased number of apices with higher
N level; Table 7) has been observed before by Howeler
and Cadavid (1983), Olasantan et al. (1994, 1997) and
in various field trials at CIAT (CIAT, 1992).

While our trial showed no significant correlation

between cassava tuber yield and soil parameters, the
farmers apply considerable amounts of nutrients in
form of chicken manure (see Section 2), supposedly to
increase tuber yield. In contrast to the Rio Cabuyal
watershed where constant N supply is provided for by
the high organic matter content of the soils, cassava
yield was significantly improved by N applications in
other areas poor in organic matter (Howeler and
Cadavid, 1990). With respect to P and K, there are
several reasons, why the regular application of
fertilizer is important. K removal at cassava harvest
in this experiment was between 100 and 150 kg ha
1
depending on cultivar and yield, so the amount of K
fertilizer applied at planting (113 kg ha
1) was hardly
enough to make up for the loss at harvest. If cassava is
planted for several crop cycles, K becomes the most
limiting element (CIAT, 1992; Howeler and Cadavid,
1990). The application of P (88 kg ha
1) was much
higher than the removal (1015 kg ha
1), but the soil
P content at all locations was below the critical level
allowing for 95% of the maximum yield (Howeler and
Cadavid, 1990), which fully justified the high P doses
applied by the farmers.
In addition, the chicken manure used by the farmers
had very high organic matter content (75% of DM).
On the long run, this will help to improve aggregate
stability and to reduce soil bulk density, which in turn
will have a positive effect on crop growth and
production and will decrease susceptibility to erosion.

4. Conclusions
Mixed cropping systems used in the studies in Rio
Cabuyal reduced cassava tuber yield and total
biomass production of the cropping systems compared to the cassava cultivar monocrops. When total
soil cover was improved compared to the cassava
cultivar pure stands it was paralleled by reductions in
terms of cassava tuber yield. These findings were in
contrast to the mixture concept which seeks to
increase productivity and soil cover compared to
monocropping by improving the use of natural
resources and farmer inputs (Mutsaers et al., 1993).
The results indicated that in terms of tuber production
and soil cover plant-to-plant cassava cultivar mixtures are in fact no improvement compared to cultivar

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

monocrops. They raised questions concerning the

ideal characteristics of a cassava cultivar and its
partner crop planted in species mixtures. Based on our
findings a cassava cultivar with low competition
ability and high harvest index should be combined
with a fast maturing intercrop. The plants morphology and cassava branching type have to be taken into
account. Though the contrasting morphologies of
cassava cultivar SM and upland rice seemed to
combine well in soils providing good growth
conditions for rice, our experiment clearly showed
that for the vast majority of soil characteristics-soil
preparation combinations rice is not a suitable
intercrop partner for cassava under the low input
practices applied in Rio Cabuyal. Fast growing and
highly competitive species with late maturity, e.g. C.
brasiliensis should not be planted together with
cassava right from the start of the cropping cycle.
Other cropping systems different from cassava
cultivar mixtures and intercropping systems need to
be developed. Farmers in Rio Cabuyal are experimenting with cassava cultivated in strips between
perennial barriers of grass and blackberry planted on
the contour in order to reduce soil erosion (Mu llerSa mann, 1997). In addition, cassava plantings
consisting of different blocks of cassava varieties,
would provide some of the potential advantages of
more diverse cropping systems, but avoid the negative
effects of strong intergenotype and interspecies
competition observed in these experiments. Our
results showed that not depending on a specific
cropping system production is increased by the
cultivation of improved and locally adapted cassava
cultivars with adequate fertilization.
Nevertheless, it will not be possible to resolve all
problems related to cassava monocropping with more
diverse cassava cropping systems. In the Cauca
department, improving of the fallow is another
obvious option. Instead of intercropping C. brasiliensis with cassava, the legume could be cultivated in
pure stands during the fallow period or it could be
sown in fields going to the fallow period after the next
cassava harvest as soon as the cassava crop is fully
established. C. brasiliensis could play an important
role as nutrient pump producing large amounts of
green manure. These studies showed that taking
advantage of the theoretical potential of species
mixtures needs careful selection of the mixing

611

partners for maximum complementarity. Cassava

cultivars for species mixture plantings are not easily
available and need to be selected under the same
micro-environmental conditions (i.e. in species mixtures) they are thought to be cultivated in practice.
Different management options and elements of
production systems need to be combined and tested
to improve overall productivity at the farm and at the
landscape level. These studies also demonstrated that
there is a need to test cropping systems at different
locations with varying environmental conditions over
an extended period of time.

Acknowledgements
We thank both the Direktion fu r Entwicklung und
Zusammenarbeit (DEZA) and the Zentrum fu r
Internationale Landwirtschaft (ZIL) for the initial
financing of this project. Additional financial support
for successful conclusion of field and laboratory work
in Colombia was provided by the HochstrasserStiftung at the Swiss Federal Institute of Technology
(ETH). The Centro Internacional de Agricultura
Tropical (CIAT) provided logistical support during
the practical phase of this study from 1996 to 1999.
We thank the Drs. J.I. Sanz, B. Knapp and K. Mu llerSaemann at CIAT for many fruitful discussions and
helpful hints. We are indebted to the hillside farmers
Julio Hurtado, Lizardo Campo, Jeremias Chocue,
Merardo Mun oz, and Luis Guastumal in Pescador,
Cauca, Colombia who offered the experimental fields
and shared valuable knowledge and experience with
us.
Special thanks go to Dr. Carlos Moreno of
Cenican a in Palmira, Valle del Cauca, Colombia for
statistical support.

References
Ahohuendo, B.C., Sarkar, S., 1995. Partial control of the spread of
African cassava mosaic virus in Benin by intercropping. Zeitschrift-fuer-Pflanzenkrankheiten-und-Pflanzenschutz 102 (3),
249256.
Akanvou, R., Kropff, M.J., Bastiaans, L., Becker, M., 2002. Evaluating the use of two contrasting legume species as relay
intercrop in upland rice cropping systems. Field Crops Res.
74 (1), 2336.

612

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Akobundu, I.O., 1981. Weed control in maize-cassava intercrop. Les

mauvaises herbes dans les cultures mixtes de mais et de manioc.
In: Terry, E.R., Oduro, K.A., Caveness, F. (Eds.), 1. Triennial
Root Crops Symposium of the International Society for Tropical
Root Crops. Tropical Root Crops: Research Strategies for the
1980s. Plantes-racines tropicales: strategies de recherches pour
les annees 1980. IDRC, Ottawa, Canada, Eng. ed. 124128;
Fr.ed. 131136.
Altieri, M.A., 1994. Biodiversity and Pest Management in Agroecosystems. Food Products Press, New York, USA.
Balesdent, J., 1997. Considerations about soil organic matter.
Chambres dAgriculture 856, 1722.
Becker, M., Johnson, D.E., 1998. Legumes as dry season fallow in
upland rice-based systems of West Africa. Biol. Fertil. Soils 27
(4), 358367.
Benites, J.R., McCollum, R.E., Naderman, G.C., 1993. Production
efficiency of intercrops relative to sequentially planted sole
crops in a humid tropical environment. Field Crops Res. 31 (1
2), 118.
Bourget, S.J., Kemp, J.G., 1957. Wet sieving apparatus for stability
analysis of soil aggregates. Can. J. Soil Sci. 37, 60.
Bouyoucos, G.J., 1962. Hydrometer method improved for making
particle size analysis of soils. Agron. J. 54, 464465.
Braun, A.R., Bellotti, A.C., Lozano, J.C., 1993. Implementation of
IPM for small-scale cassava farmers. In: Altieri, M.A. (Ed.),
Crop Protection Strategies for Small Farmers. Westview Press,
London, pp. 103115.
Cadisch, G., Sylvesterbradley, R., Boller, B.C., Nosberger, J., 1993.
Effects of phosphorus and potassium on N-2 fixation (N-15dilution) of field-grown Centrosema acutifolium and C. macrocarpum. Field Crops Res. 31 (34), 329340.
Cenpukdee, U., Fukai, S., 1991. Effects of nitrogen supply on
cassava-pigeonpea intercropping with three contrasting cassava
cultivars. Fertil. Res. 29 (3), 275280.
CIAT, 1992. Cassava Program Annual Report. Centro Internacional
de Agricultura Tropical (CIAT), Cali, Colombia.
CIAT, 1996a. CIAT Hillsides Program Annual Report 19941995,
Cali, Colombia.
CIAT, 1996b. Manejo de recursos naturales. Microcuenca Dos
zanjones, Rio Cabuyal. Centro Internacional de Agricultura
Tropical, Cali, Colombia.
Cock, J.H., Franklin, D., Sandoval, G., Juri, P., 1979a. The ideal
cassava plant for maximal yield. Crop Sci. 19, 271279.
Culley, J.L.B., 1993. Carter, M.R. (Ed.), Density and Compressibility. CRC Press, Boca Raton, FL, pp. 529539.
Dahniya, M.T., Jalloh, A., Ezumah, H.C., 1994. Intercrop yields of
cassava and rice as influenced by cassava variety and population.
Acta Hortic. 380, 194199.
Elwell, H.A., Stocking, M.A., 1974. Rainfall parameters and a cover
model to predict runoff and soil loss from grazing trials in the
Rhodesian sandveld. In: Proceedings of the Grassland Society
of South Africa. pp. 157164.
Evangelio, F.A., Villamayor Jr., F.G., Ecleo, D.G., Howeler, R.H.,
1994. Erosion control strategies for cassava grown in sloping
lands. Philippine J. Crop Sci. (Philippines) 19 (Suppl. 1).
Ezulike, T.O., Igwatu, R.I., 1993. Effects of intercropping cassava
and pigeon pea on green spider mite Mononychellus tanajoa

(Bondar) infestation and on yields of the associated crops.

Discovery-and-Innovation 5 (4), 355359.
Ezumah, H.C., Lawson, T.L., 1990. Cassava and maize intercropping systems the effects of varieties and plant-populations. J.
Agron. Crop Sci.-Zeitschrift Fur Acker Und Pflanzenbau 164
(5), 334342.
Fondong, V.N., Thresh, J.M., Zok, S., 2002. Spatial and temporal
spread of cassava mosaic virus disease in cassava grown alone
and when intercropped with maize and/or cowpea. Phytopath-Z
150 (7), 365374.
Fraiture, C.D., Rubiano, J., Alvarez, C., 1997. Actual Versus Potential Wateruse in Cabuyal Watershed, Colombia. Centro Internacional de Agricultura Tropical (CIAT).
Francis, C.A., 1990. Potential of multiple cropping systems. In:
Altieri, M.A., Hecht, S.B. (Eds.), Agroecology and Small Farm
Development. CRC Press, Boca Raton, FL, p. 262.
Gijsman, A.J., Sanz, J.I., 1998. Soil organic matter pools in a
volcanic-ash soil under fallow or cultivation with applied
chicken manure. Euro. J. Soil Sci. 49 (3), 427436.
Gold, C.S., 1993. Effects of cassava intercropping and varietal
mixtures on herbivore load, plant growth, and yields: applications for small farmers in Latin America. In: Altieri, M.A. (Ed.),
Crop Protection Strategies for Small Farmers. Westview Press,
London, pp. 117141.
Gold, C.S., Altieri, M.A., Bellotti, A.C., 1989a. Effects of cassava
varietal mixtures on the whiteflies Aleurotrachelus socialis and
Trialeurodes variabilis in Colombia. Entomol. Exp. Appl. 53,
195202.
Gold, C.S., Altieri, M.A., Bellotti, A.C., 1989b. The effects of
intercropping and mixed varieties of predators and parasitoids
of cassava whiteflies (Hemiptera: Aleyrodidae) in Colombia.
Bull. Ent. Res. 79, 115121.
Gold, C.S., Altieri, M.A., Bellotti, A.C., 1990. Effects of intercropping and varietal mixtures on cassava hornworm, Erinnyis ello
L. (Lepidoptera: Sphingidae), and the stemborer, Chilomima
clarkei (Amsel) (Lepidoptera: Pyralidae), in Colombia. Trop.
Pest Manage. 34 (4), 362367.
Gomez, K.A., Gomez, A.A., 1984. Statistical Procedures for Agricultural Research. Wiley, New York, USA.
Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.C., 1992.
Multivariate Data Analysis. Prentice-Hall International,
London.
Hernandez, A., Ramos, R., Sanchez, J., 1999. Spacing and timing of
intercropping cassava and beans: land equivalent ratio. Agronomia-Mesoamericana 10 (1), 6366.
Holdridge, L.R., 1967. Life Zone Ecology. San Jose , Costa Rica.
Howeler, R.H., 1998. Cassava agronomy research in Asia an
overview 19931996. In: Howeler, R.H. (Ed.), Cassava Breeding, Agronomy and Farmer Participatory Research in Asia.
Proceedings of the Fifth Regional Workshop, November 38,
1996. CIAT, Bangkok Regional Cassava Office.
Howeler, R.H., Cadavid, L.F., 1983. Accumulation and distribution
of dry matter during a 12-month growth cycle of cassava. Field
Crops Res. 7, 123139.
Howeler, R.H., Cadavid, L.F., 1990. Short- and long-term fertility
trials in Colombia to determine the nutrient requirements of
cassava. Fertil. Res. 26, 6180.

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614
Hulugalle, N.R., Ezumah, H.C., 1991. Effects of cassava-based
cropping systems on physico-chemical properties of soil and
earthworm casts in a tropical Alfisol. Agric. Ecosyst. Environ. 35
(1), 5564.
Hulugalle, N.R., Ezumah, H.C., Leyman, T., 1994. Changes in
surface soil properties of a no-tilled tropical Alfisol due to
intercropping maize, cassava and egusi melon. Field Crops
Res. 36, 191200.
Ikeorgu, J.E.G., Odurukwe, S.O., 1990. Increasing the productivity
of cassava/maize intercrops with groundnuts (Arachis hypogaea
L.). Trop. Agric. (Trinidad and Tobago) 67 (2), 164168.
Jalloh, A., Dahniya, M.T., Ezumah, H.C., 1994. Production of
cassavarice intercrop as influenced by cassava variety and time
of introducing rice into cassava. Acta Hortic. 380, 200204.
Jambu, M., 1991. Exploratory and Multivariate Data Analysis.
Academic Press, San Diego, CA.
Jaramillo, G., 1994. Primer Informe sobre resultados de evaluaciones de clones de yuca en el norte del Cauca. Centro Internacional de Agricultura Tropical (CIAT), Cali, Colombia.
Karlen, D.L., Nash, N.S., Unger, P.W., 1992. Soil and crop management effects on soil indicators. Am. J. Alternative Agric. 7, 48
55.
Kawano, K., Thung, M.D., 1982. Intergenotypic competition and
competition with associated crops in cassava. Crop Sci. 22 (1),
5963.
Klute, A., 1986. Methods of Soil Analysis. Part 1. Physical and
Mineralogical Methods. American Society of Agronomy, Inc./
Soil Science Society of America, Inc., Madison, WI, USA.
Krom, M., 1980. Spectrophotometric determination of ammonia: a
study of modified Berthlot reaction using salicylate and dichloroisocyanate. The Analyst 105, 305316.
Lathwell, D.J., 1990. Legume green manures. Principles for management based on recent research. Tropsoils Bulletin. Soil
Management Collaborative Research Support Program, 90-01,
30.
Leihner, D.E., Ruppenthal, M., Hilger, T.H., Castillo, J.A., 1996.
Soil conservation effectiveness and crop productivity of
forage legume intercropping, contour grass barriers and contour
ridging in cassava on Andean hillsides. Exp. Agric. 32 (3), 327
338.
Lozano, J.D., Byrne, D., Bellotti, A., 1980. Cassava/ecosystem
relationships and their influence on breeding strategy. Trop.
Pest Manage. 26, 180187.
Lusembo, P., Ebong, C., Sabiiti, E.N., 1998. Integration of cassava
tuber and forage legume seed production for sustained soil
fertility. Trop. Agric. 75 (12), 1820.
Martens, D.C., Lindsay, W.L., 1990. Testing soil for Cu, Fe, Mn and
Zn. In: Westermann, R.L. (Ed.), Soil Testing and Plant Analysis,
USA.
Masle, J., 1999. Root impedance: sensing, signaling and physiological effects. In: Lerner, H.R. (Ed.), Plant Responses to Environmental Stresses: from Phytohormones to Genome
Organization. Marcel Dekker, New York.
Mason, S.C., Leihner, D.E., Vorst, J.J., 1986a. Cassavacowpea and
cassavapeanut intercropping. 1. Yield and land-use efficiency.
Agron. J. 78 (1), 4346.

613

Mason, S.C., Leihner, D.E., Vorst, J.J., 1986c. Cassavacowpea and

cassavapeanut intercropping. 3. Nutrient concentrations and
removal. Agron. J. 78 (3), 441444.
Mason, S.C., Leihner, D.E., Vorst, J.J., Salazar, E., 1986b. Cassava
cowpea and cassavapeanut intercropping. 2. Leaf-area index
and dry-matter accumulation. Agron. J. 78 (1), 4753.
Mohankumar, C.R., Hrishi, N., 1978. Intercropping systems with
cassava in Kerala State, India. In: Weber, E., Nestel, B., Campell,
M. (Eds.), Intercropping with Cassava.
Muhr, L., Leihner, D.E., Hilger, T.H., Mueller-Saemann, K.M.,
1995. Intercropping of cassava with herbaceous legumes. I.
Rooting patterns and their potential importance for belowground
competition. II. Yields as affected by below-ground competition.
Angewandte-Botanik 69 (12), 1726.
Mu ller-Sa mann, K., 1997. Documento de trabajo No. 3. Proyecto de
conservacion de suelos. CIAT, Cali, Colombia.
Mutsaers, H.J.W., Ezumah, H.C., Osiru, D.S.O., 1993. Cassavabased intercropping: a review. Field Crops Res. (Netherlands) 34
(34), 431457.
Nelson, D.W., Sommers, L.E., 1982. Total carbon, organic carbon
and organic matter. In: Page, A.L. (Ed.), Methods of Soil
Analysis. American Society of Agronomy, Madison, WI, USA.
Obiagwu, C.J., 1995. Estimated yield and nutrient contributions of
legume cover crops intercropped with yam, cassava, and maize
in the Benue River Basins of Nigeria. J.-Plant-Nutr. 18 (12),
27752782.
Odurukwe, S.O., Ikeorgu, J.E.G., 1994. Effects of fertilizer and time
of introduction of cassava in yam/maize/cassava intercrop on
component yield. Acta Hortic. 380, 7277.
Okeke, J.E., 1996. Effects of spatial arrangement and relative time of
planting on intercropped maize and cassava. Trop. Sci. 36 (1), 6
10.
Okoli, O.O., Hossain, M.A., Kissiedu, A.F.K., Asare Bediako, A.,
1996. Effect of planting dates and growth habits of cassava and
cowpea on their yield and compatibility. Trop. Agric. 73 (3),
169174.
Olasantan, F.O., Ezumah, H.C., Lucas, E.O., 1996. Effects of
intercropping with maize on the micro-environment, growth
and yield of cassava. Agric. Ecosyst. Environ. 57 (23), 149
158.
Olasantan, F.O., Ezumah, H.C., Lucas, E.O., 1997. Response of
cassava and maize to fertilizer application, and a comparison of
the factors affecting their growth during intercropping. Nutr.
Cycling Agroecosyst. 46, 215223.
Olasantan, F.O., Lucas, E.O., Ezumah, H.C., 1994. Effects of
intercropping and fertilizer application on weed control and
performance of cassava and maize. Field Crops Res. 39, 6369.
Polthanee, A., Anan, P., 1999. Studies on cassavapeanut intercropping patterns in an acid soil in northeast Thailand. Thai J. Agric.
Sci. 32 (1), 127134.
Quiros, C., 1996. Evaluacion de materiales de arroz en la cuenca del
Rio Cabuyal, Cauca, Colombia. Centro Internacional de Agricultura Tropical (CIAT), Cali, Colombia.
Rehm, S., 1991. The Cultivated Plants of the Tropics and Subtropics:
Cultivation, Economic Value, Utilization. Josef Margraf, Weikersheim, Germany.

614

G.C. Daellenbach et al. / Agriculture, Ecosystems and Environment 105 (2005) 595614

Reining, L., 1992. Erosion in Andean hillside farming: characterization and reduction of soil erosion by water in small scale
cassava cropping systems in the southern Cordilera Central of
Colombia. Margraf, Weikersheim.
Reynolds, W.D., 1993. Saturated hydraulic conductivity: laboratory
measurement. In: Carter, M.R. (Ed.), Soil Sampling and Methods of Analysis. CRC Press, Boca Raton, FL, pp. 599613.
Risch, S.J., Andow, D., Altieri, M.A., 1983. Agroecosystem diversity and pest-control data, tentative conclusions, and new
research directions. Environ. Entomol. 12 (3), 625629.
Robinson, J., 1997. Intercropping upland rice (Oryza sativa L.) and
groundnut (Arachis hypogaea L.) with cassava (Manihot esculenta Crantz) in southern Sudan. Trop. Agric. 74 (1), 711.
Ruppenthal, M., Leihner, D.E., Steinmuller, N., Elsharkawy, M.A.,
1997. Losses of organic matter and nutrients by water erosion
in cassava-based cropping systems. Exp. Agric. 33 (4),
487498.
Salinas, J.G., Garcia, R., 1985. Me todos quimicos para el a nalisis de
suelos a cidos y plantas forrajeras. Centro Internacional de
Agricultura Tropical (CIAT), Cali, Colombia.
SAS, 1988. SAS Procedures Guide, Release 6.03 ed. SAS Institute
Inc., Cary, NC, USA.
Sitompul, S.M., Setijono, S., Heide, J.V.S., Noordwijk, M.V., Van
der Heide, J., Van Noordwijk, M., 1992. Crop yields and
sustainability of cassava-based cropping systems on an Ultisol
in Lampung. Agrivita, Special issue: N-management in sustainable cropping systems on an Ultisol 15 (1), 1928.

SPSS, 1998. SYSTAT 8.0 Statistics. SPSS Inc., Chicago, IL, USA.
Thomas, G.W., 1982. Exchangeable cations. In: Page, A.L. (Ed.),
Methods of Soil Analysis. American Society of Agronomy,
Madison, WI, USA.
Topp, G.C., Galganov, Y.T., Ball, B.C., Crater, M.R., 1993. Soil
water desorption curves. In: Carter, M.R. (Ed.), Soil Sampling
and Methods of Analysis. CRC Press, Boca Raton, FL, pp. 569
580.
Tsay, J.S., Fukai, S., Wilson, G.L., 1988. Intercropping cassava with
soybean cultivars of varying maturities. Field Crops Res. 19 (3),
211225.
Unamma, R.P.A., Ene, L.S.O., 1984. Weed interference in cassava
maize intercrop in the rain forest of Nigeria. In: Terry, E.R., et al.
(Eds.), Tropical Root Crops: Production and Uses in Africa.
International Development Research Centre, Ottawa, Ont.
Weeks, E.P., 1978. Field determination of vertical permeability to air
in the unsaturated zone. Geol. Survey Prof. Paper 1051, 41.
Wortmann, C.S., McIntyre, B.D., Kaizzi, C.K., 2000. Annual soil
improving legumes: agronomic effectiveness, nutrient uptake,
nitrogen fixation and water use. Field Crops Res. 68 (1), 7583.
Zaffaroni, E., Vasconcelos, A.F.M., Lopes, E.B., 1991. Evaluation of
intercropping cassava/corn/beans (Phaseolus vulgaris L.) in
northeast Brazil. J. Agron. Crop Sci. (Germany, F.R.) 167 (3),
207212.
Zuofa, K., Tariah, N.M., Isirimah, N.O., 1992. Effects of groundnut,
cowpea and melon on weed control and yields of intercropped
cassava and maize. Field Crops Res. 28 (4), 309314.