lable at ScienceDirect
Journal of Stored Products Research 84 (2019) 101517Contents lists avaiJournal of Stored Products Research
journal homepage: www.elsevier .com/locate/ jsprPhysical quality of maize grain harvested and stored by smallholder
farmers in the Northern highlands of Tanzania: Effects of harvesting
and pre-storage handling practices in two marginally contrasting
agro-locations
Christopher Mutungi*, Francis Muthoni, Mateete Bekunda, Audifas Gaspar, Esther Kabula,
Adebayo Abass
International Institute of Tropical Agriculture, Plot No. 25, Mikocheni Light Industrial Area Mwenge, Coca Cola Road, Mikocheni B, P.O. Box 34441, Dar es
Salaam, Tanzaniaa r t i c l e i n f o
Article history:
Received 9 March 2019
Received in revised form
18 September 2019
Accepted 21 September 2019
Available online xxx
Keywords:
Agro-climatic conditions
Post-harvest operations
Farmer practices
Quality
Losses* Corresponding author.
E-mail address: c.mutungi@cgiar.org (C. Mutungi)
https://doi.org/10.1016/j.jspr.2019.101517
0022-474X/© 2019 The Authors. Published by Elseviea b s t r a c t
On-farm trials were conducted to investigate the effects of maize harvesting and handling practices of
smallholder farmers on the quality of the produce before, and during storage in two contrasting agro-
locations. Farmers harvested and prepared the crop according to local practices, and stored it in ordi-
nary woven polypropylene bags for 30 weeks. Grain moisture, insect populations, insect-damage, moldy/
diseased/discolored grain, rodent-damage, shriveled grain, broken grains, non-consumable grains, im-
purities, and overall losses were monitored. Moisture of the pre-stored grain ranged between 11.0 and
23.7% while the overall physical damage was 16.9± 6.2%. Late harvesting increased moldy/diseased/
discolored grain two-fold while de-husking and drying practices increased the levels in early-harvested
grain by factor of 2e3. Insect populations were >10 times higher in the cooler agro-location, and
handling practices increased them by factor of 2e10. The interaction of agro-location, harvesting time
and drying influenced the amount of grain that was unfit for human consumption. Pre-storage losses of
3.6e11.2% were determined, mainly as grade-outs. With storage, the quality of early-and late-harvested
maize did not differ. However, the majority of examined parameters were distinct by agro-location.
Moreover, secondary pests and the levels of shriveled and broken grain levels were also distinct by
drying method, while moldy/diseased/discolored grain, non-consumable grain, and overall losses were
distinct depending on whether the harvested cobs were de-husked or not de-husked before drying. The
high levels of grade-outs at the pre-storage stage suggest that sorting should be emphasized for quality
improvement at the farm gate not only for the market but also household nutrition. Cultivation of va-
rieties with superior maturing and post-harvest traits would lower the sorting losses. Agro-location and
farmer practices influenced grain quality and magnitude of losses during storage. These findings should
inform choice of intervention steps right from the pre-storage stage.
© 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).1. Introduction
At least 7.4 million farm producers cultivate maize (Zea mays L.)
for food and income in Tanzania (TNBS, 2019). The annual per capita
consumption is estimated at 73 kg, and maize alone is projected to
contribute approximately 33e40% of the total daily calorie intake,
because of the greater caloric density compared to other crops.
r Ltd. This is an open access articleconsumed by households (Minot, 2010). With a national annual
production of at least 5.7 million tons (TNBS, 2019), about
250e400 kg of each ton undergoes quality deterioration or is
completely lost along the post-production pipeline, and therefore
not unavailable to feed families or generate revenue. Improper
approaches of harvesting, preliminary processing and handling,
diminish the harvest by 20e130 kg per ton, while poor storage
further decreases the stocks by 150e250 kg per every ton stored
(Abass et al., 2014).
Insects, molds, and manual processing were identified as theunder the CC BY license (http://creativecommons.org/licenses/by/4.0/).
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 1015172main causes of grain loss in smallholder maize farming systems in
Tanzania (Abass et al., 2014). The generally warm and humid con-
ditions encourage pests and molds to attack the maturing crop in
the field, and at the pre-storage stage (Boxall, 2002). During the
latter, inappropriate methods of drying, threshing, winnowing,
sorting, and grading leave behind broken grain, dirt, impurities, and
other contaminants increasing the proneness of the produce to
further damage by pests and pathogens in the course of storage.
Consequently, post-harvest operations may affect the class of grain
produced and consumed or sold, as well as its safety and shelf-life
(Mendoza et al., 2017).
Many factors are responsible for the way farmers harvest and
handle produce when the crop is mature. In maize-based crop-
livestock systems, early harvesting is beneficial as the crop resi-
dues can be secured for forage when the quality is still high
(Tolera and Sundstøl, 1999). Early harvesting is therefore a
desirable practice. Occasionally, farmers strip the leaves and cut
off the tops of the maize crop after attaining physiological
maturity, leaving behind the cobs to dry in the field. In cooler
zones, this practice encourages drying by exposing the ears to the
sun (Subedi, 1996). Other farmers harvest early to make room for
the next crop (Chegere, 2018). Socio-economic factors such as the
size of farms, fear of pilferage, how quickly farmers want to utilize
the crop, availability of labor, and availability of post-harvest
equipment were also found to be important decision factors
(Kaaya et al., 2005). Moreover, agro-climatic conditions are also
important. In cooler agro-zones, for example, lower ambient
temperatures reduce the drying potential of air hence the pro-
duce fails to dry well or stays longer in the field and on drying
platforms.
Understanding the impact of post-harvest operations on grain
quality within the contexts of farming environments should guide
farmers on the choice of better intervention steps, if necessary, to
decrease spoilage and post-harvest losses, and ultimately
contribute to food security and safety (Mendoza et al., 2017). Maize
is cultivated across almost all agro-climatic regions of Tanzania in
the lowland, intermediate, and highland zones (Suleiman and Kurt,
2015). These agro-environments are also becoming increasingly
variable due to climate change. As a result, the conditions under
which maize is harvested, handled, and stored continue to vary
widely affecting not only the incidence and severity of loss agents
(Khaliq et al., 2014; Magan et al., 2011), but also the way farmers
respond to post-harvest challenges (Stathers et al., 2013). Tradi-
tionally, most post-harvest losses have been associated with insect
pests. However, a significant proportion of the total loss, which is
often not quantified, emanates from fungal contamination, rot and
diseases, rodent damage, mechanical injury, and other defects.
These forms of damage are linked to farming environments, but
also to the harvesting and handling practices. For this reason,
addressing harvesting and handling practices is critical for safety
and wholesomeness of the grain that ultimately becomes available
for consumption by households and for the market.
The objectives of the present study were therefore to: (i) eval-
uate the kinds and levels of defects that constitute post-production
loss at harvest stage and during pre-storage handling of maize in
contrasting agricultural environments; (ii) examine the effect of
harvesting/handling practices on the quality of the harvested pro-
duce before storage and (iii) investigate the effect of harvesting/
handling practices on grain quality and overall losses during ordi-
nary storage. The objectives were achieved using a losses assess-
ment methodology that evaluates a broad spectrum of grain
damage and quality defects. Such knowledge is important for
identifying measures that farmers, under different contexts, might
need to take in order to derive the intended benefits of improved
post-harvest technologies.2. Materials and methods
2.1. Trial sites, timing, and selection of farmers
On-farm trials were conducted in Babati district, located in
Manyara region of Tanzania. Babati is characterized by warm and
temperate climate, but is also typified by agro-climatic gradients.
The district generally experiences a bimodal rainfall pattern with
short rains peaking in NovembereDecember while long rains peak
inMarcheApril. On average, temperatures are highest in November
and lowest in July. Trials were conducted in two agro-locations:
Long village (S 4
 130 15.62''; E 35
 250 31.80''; 2162.8m.a.s.l) and
Seloto village (S 4
 150 2.48''; E 35
 310 3.70''; 1628m.a.s.l) that have
contrasting weather patterns. Weather data of the two agro-
locations were downloaded from agriculture - weather data plat-
form (aWhere). The platform (www.awhere.com) provides global
cloud-based data for historic and current observations, and fore-
casts with daily and 9 km spatial resolutions. This data was
generated using aWhere's in-house 3D curvilinear interpolation
algorithm that blends multi-source data from ground weather
stations, Doppler radar (where available), and satellite observa-
tions. Satellite data sources include the Tropical Rainfall Measuring
Mission (TRMM) (Huffman et al. 2007), Global Precipitation Mea-
surement (GPM) mission (Hou et al. 2013) and the Agricultural Re-
analysis of Precipitation Data (AREA-PD). Global assessment of
aWhere data revealed that temperature differ from ground stations
data by a mean of less than 0.67 
C and precipitation is within
10mm up-to 95% of the time. The aWhere datasets were developed
to improve agricultural advisories especially in Africa where the
density of weather observations network is low. Trials were con-
ducted from July 2016eMarch 2017, covering a whole post-harvest
cycle that usually lasts 7e9 months.
2.2. Experimental details
A preliminary survey was undertaken to inventory harvesting/
handling practices of farmers. Seven procedures were identified
(see Table 1). Ten willing farmers (5 in each village) who had suf-
ficient produce (>5 acres of crop) were recruited. A section
measuring approximately 50m 30m of the farm belonging to
each of the participating farmers was demarcated. The farmers, in
each of the two locations, were randomly assigned the harvesting/
handling practice procedures (Table 1). Each farmer was assigned
2e3 treatments, which they replicated four times each. The farmers
harvested and prepared the maize accordingly, and bagged the
shelled grain in woven polypropylene (PP) which were then stored
in their own stores for 30 weeks. Early harvesting was done at the
beginning of July, and late harvesting 6 weeks later. Shelling of cobs
was done using a motorized sheller with a throughput of 0.7 tons
per hour. Tarpaulins were improved drying mats (Collapsible Drier
Cases, CDC™) from Grainpro Philippines Inc. (Subic Bay,
Philippines). Storage bags were obtained locally from a dealer in
Babati. All farmers were compensated for the maize, but vol-
unteered storage structures in the homestead. The structures were
rooms made of brick or mud wall, with earthen or concrete floor
and roofed with iron sheets. Village extension officers monitored
the trial on regular basis.
2.3. Sampling
Samples were taken at the time of storage, and at six weeks
intervals. Each bag was opened, and 1 kg of grain was drawn from
several random points by pushing a sampler from the top to the
bottom of the bag at multiple random points. The sample was
analyzed for moisture content (m.c.) using a HE lite moisture tester
Table 1
Harvesting and pre-storage handling practices.
Practice Abbreviation
Early harvesta, de-husk cobs, shell, bag EH
Early harvest, dry cobs with husks on the ground, husk cobs, shell, bag EDgH
Early harvest, dry cobs with husks on tarpaulin, husk cobs, shell, bag EDtH
Early harvest, de-husk cobs, heap on the ground to dry, shell, bag EHDg
Early harvest, de-husk cobs, dry on tarpaulin, shell, bag EHDt
Late harvestb, de-husk cobs, shell, bag LH
Late harvest, de-husk cobs, dry on tarpaulin, shell, bag LHDt
a Early July.
b Six weeks after early harvest.
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 101517 3(Pfeuffer GmbH, Kitzingen, Germany), and then sub-divided into
two sub-samples using quartering method. One sub-sample was
randomly drawn and set aside for analysis. Samples were sealed in
zip-lock bags, and preserved in a cold room maintained at 8 
C
awaiting analysis within two weeks.
2.4. Determination of insect counts, grain damage and physical loss
The aim was to quantify not only insects and insect damage but
also other forms of damage including moldy/diseased/discolored
grain, rodent damaged grain, broken grain, shriveled grain, and
impurities/foreign matter so as to have a full picture of the physical
grain quality and total losses. A sub-sample was weighed and
sieved through a 6mm mesh sieve to separate the impurities
comprising insects, insect frass, small broken grains, dust, and
other fine debris. From the trash, adult insects were separated by
species using forceps, and counted. The broken grains were also
separated. Insect counts were reported as number per kg of grain.
The large foreign matter remaining on the sieve were hand-picked
and combined with the finer impurities and the mass recorded
(Mimp) as g/kg. From the grains retained on the sieve, 1000 grains
were randomly picked and weighed (M1000). These were then each
displayed on a sorting board with 1000 slots (one slot for one
grain), and then sorted into various categories according to domi-
nant type of damage. Further, the damaged grains were separated
into consumable and non-consumable ones based on acceptance -
rejection criteria of local maize consumers. Trained technicians
were used to collect the following data: Number of insect damaged
grains (ni) and themass (mi); number of moldy/diseased/discolored
grains (nmrd) and the mass (mmrd); number of rodent damaged
grains (nr) and the mass (mr); number of broken grains (nb) and the
mass (mb); number discolored grains (nd) and the mass (md);
number of shriveled grains (ns) and the mass (ms); total number of
damaged grains, Ntd ¼ (ni þ nmrd þ nr þ nb þ nd þ ns) and the mass,
Mtd ¼ (mi þ mmrd þ mr þ mb þ md þ ms); total number of sound
grains, Nts ¼ (1000 - Ntd) and the mass,Mts ¼ (M1000 -Mtd); number
of consumable damaged grain (Ntdc) the mass (Mtdc); number non-
consumable damaged grains, Ntdnc ¼ (Ntd - Ntdc) and the mass,
Mtdnc¼ (Mtd - Mtdc). The averagemass of one damaged grain (a) was
estimated as Mtd/Ntd. The average mass of one sound grain (b) was
estimated as Mts/Nts. The mass of the damaged grains before the
damage (c) was estimated as b x Ntd. Hence, the mass of the 1000
grain sample (T), assuming no damage, was obtained as follows: T¼
(cþMts). The percent overall damage bymass was calculated as (c/T
x 100). The percent loss was calculated as (c - Mtdc)/T x 100. An
assumption was made, that on average, a damaged grain weighed
less than a non-damaged one.
2.5. Statistical analysis
Data were analyzed using GLM procedures on IBM® SPSS®
Statistics 20 (IBM Corporation, New York, USA). A series ofmultivariate analyses of variance were performed. Climate data
(Five dependent variables: daily temperature - minimum and
maximum; atmospheric relative humidity - minimum and
maximum; and precipitation) as well as grain quality data (10
dependent variables: insect population, insect damage, rodent
damage, shriveled grain, moldy/diseased/discolored grain, broken
grain, non-consumable grain, impurities, total damage, overall
losses) were first checked for the critical assumptions for multi-
variate analysis: normality, absence of univariate and multivariate
outliers, absence of multicollinearity, linear relationship among
variables, and homogeneity of covariance matrices. For the grain
quality parameters, data were not normally distributed and
therefore, were required, non-parametric Kruskal-Wallis test was
used to separate means of treatment groups. Multivariate outliers
were checked by running a multiple linear regression with the
dependent variables as the independent variables, and computing
the Mahalanobis' distance (MD) values for each case. Outlier cases
were identified by comparing the derived MD values with the
critical maximum value from chi square tables at P¼ 0.001, and
df¼ number of dependent variables). Critical value of 20.52 and
29.59 were applied for the five dependent variables related to
climate, and the 10 dependent variables related to grain quality,
respectively. All cases with MD values above the critical value were
removed. Multicollinearity of dependent variables was tested using
collinearity diagnostics on the regression function of SPSS, followed
by evaluation of the derived Variance Inflation Factors (VIF).
Absence of multicollinearity was satisfied if the derived VIF values
for all variables were <4. For the grain quality data, this assumption
was met after excluding the parameter ‘overall damage’. Thus nine
variables (rodent-damaged grains, shriveled grain, moldy/diseased/
discolored grain, insect population, broken grains, non-consumable
grains, impurities, insect-damaged grains, and losses) were
retained. The assumption of linear relationship between pairs of
dependent variables was tested using pearson's correlation. For the
grain quality data, a threshold was applied; only parameters that
had a correlation coefficient of at least 0.2, with at least three other
parameters were included in the analysis. On the basis of this cri-
terion, the variable 'impurities' was excluded. Homogeneity of
covariance matrices across groups was tested as part of multivar-
iate analysis using Box's M test; covariance matrices were pre-
sumed homogenous if the derived p-value was >0.001. This last
assumption was not met for the analyses, hence Pillai's Trace sta-
tistic has been reported as it is considered to be more robust where
the assumptions of multivariate analysis have been violated.
Analyses to assess effects of harvesting/handling practices on
quality of grain before storagewere performed on the baseline data.
A preliminary multivariate analysis of covariance (MANCOVA) with
grainmoisture as covariate evaluatedwhether the various practices
identified in Table 1 (combinations of harvesting time, de-husking
and drying) differed significantly. At this level, practice was the
independent variable, while the eight quality parameters (as a
group) were the dependent variables. After identifying a significant
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 1015174effect, an expanded MANCOVA was then performed to test the ef-
fects of the specific practice aspects. For this we considered an
experimental matrix consisting of 4 independent variables and
their categories: climate or agro-location (Long, Seloto); harvesting
time (early-harvesting, late harvesting); De-husking (de-husk
before drying, de-husk after drying); Drying (on the ground, on
tarpaulin, none).
Analyses to evaluate effects of storage duration was achieved
using one-way repeated measures ANOVA (RM-ANOVA) with
Greehouse-Geisser correction. Individual quality parameters were
the subjects. Time was the ‘within-subject’ factor, in which 0, 6, 12,
18, 24, and 30 week sampling points were the factor levels. Agro-
location, harvesting time, de-husking, and drying were incorpo-
rated as ‘between-subject’ factors. From the output, ‘within-sub-
ject’ effects revealed whether storage time influenced significantly
the quality parameter in question, whereas ‘between-subjects' ef-
fects revealed whether the outcomes of the different levels of agro-
location, harvesting time, de-husking or drying differed signifi-
cantly during the storage span. All graphs were plotted using Sig-
maPlot® for Windows Version 11.0 (Systat Software GMBH Erkrath,
Germany).
3. Results
3.1. Weather patterns of trial sites
From multivariate analysis, the trial sites were significantly
different in terms of the prevailing conditions (Pillais Trace¼ .653, F
5, 716¼ 270.053, P< 0.001, np2¼ 0.653). The significance was on
temperature and atmospheric r.h. (minimum temperature:
P< 0.001; maximum temperature: P< 0.001; minimum r.h.:
P< 0.001; maximum r.h.: P¼ 0.003), but not on precipitation
(P¼ 0.560). Paired t-tests showed than Long village was cooler and
less humid than Seloto (see Fig. 1). The maximum as well as the
minimum temperature profiles were significantly different
(maximum: t¼ 19.815, df¼ 728, P< 0.001; minimum: t¼30.70,
df¼ 728, P< 0.001). The maximum and minimum temperatures in
Seloto averaged 27.74± 0.10 and 15.92± 0.08 
C, respectively
compared to 24.99± 0.09 and 12.43± 0.08 in Long. The relative
humidity (r.h.) profiles were also significantly different (MaximumFig. 1. Profiles of precipitation (Green), temperature (red) and relative humidity (blue) in L
profiles are represented by the lower- and upper-line plots, respectively. Data are daily meas
is referred to the Web version of this article.)r.h: t¼ 3.080, df¼ 728, P¼ 0.002; Minimum r.h: t¼ 5.932, df¼ 728,
P< 0.001). The maximum r.h. averaged 89.34± 0.36% in Seloto and
87.44± 0.50% in Long, whereasminimum r.h. averaged 43.50± 0.54
and 38.74± 0.59, respectively.
3.2. Harvesting and pre-storage handling practices of farmers
From the preliminary survey undertaken to inventory farmer
harvesting/handling practices, about two thirds of farmers (62%;
n¼ 129) harvested early, when themaize showed signs of maturity.
Harvesting was done by opening the husks and removing the cobs
from the plant. The cobs were then transported to the homestead
for drying, shelling, and bagging for storage. Dryingwas done either
directly on the ground or on tarpaulins. A significant 16% of farmers
cut the entire plant and piled the crop in the field for 2e3 weeks to
dry, and then de-husked the cobs which they dropped in the field
(on the ground or tarpaulin) for a few days to dry further, before
transporting to the homestead for shelling and bagging. Some
farmers also shelled the maize on the farm and bagged or sold right
away. Another 8% of farmers left the crop in the field to dry (late
harvesting), removed the cobs from the plant and then transported
to homestead for shelling and bagging. About a quarter (24%) of
farmers did not dry the maize further after field drying. Prior to
shelling the cobs were sorted (85% of farmers) on the basis of mold
damage (71.7%), cob rot (46.2%), insect damage (21%), rodent
damage (12.3%), maturity (11.2%), color (2.8%), cob size (2.8%),
sprout damage (1.9%), grain size (1.0%), and damage by birds (0.9%).
Half of farmers (52.8%) used mechanical threshers for shelling,
while the rest applied manual methods. The foreign matter (chaff,
soil, dust, broken cobs) in shelled grainwas removed bywinnowing
(88%) and hand picking (10.1%). A few farmers (5.8%) relied on the
thresher to blow away foreign matter, and therefore did not carry
out further cleaning or winnowing.
3.3. Grain quality prior to storage
An overview of the levels of the various quality defects on
shelled grain as measured before storage is presented in Fig. 2. The
levels were generally spread across a wide range (grain moisture:
11e23.7%; moldy/diseased/discolored grain: 1.7e21%; total insectong and Seloto during the trial period. Minimum and maximum temperature and r.h.
urements. (For interpretation of the references to color in this figure legend, the reader
E
H
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t
N
um
be
r p
er
 k
g
E
H
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t
E
H
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t
M
as
s,
 k
g 
pe
r 1
00
kg
0
10
20
30
a
b b b
b b
c
b b b
a
bbb
a a
b b
Moisture content Insect populationMoldy/ diseased/discolored 
grains
c b
c
c
a a abab b bb
a b c
E
H
ED
gH
E
D
tH
EH
D
g
E
HD
t
LH
LH
D
t
Total losses
c
a
a
bc
ab
a
bc
i
a a a
a
a
a
a
EH
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t
M
as
s,
 k
g 
pe
r 1
00
kg
0.0
0.5
1.0
1.5
2.0
a
b
b b
c c
bc
Insect damaged grains
aa a
a
a
a a
g
E
H
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t
M
as
s,
 k
g 
pe
r 1
00
kg
0
2
4
Broken grains
E
H
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t
Shriveled grains
E
H
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t
Rodent damaged grains
a
b
aa
bb
a
b
a
b
c cc
bb
c
b
ab
b
a
b
bc
a
b
c
ab a
b
ab ab
a
bb
b
bb
b
a a
a
a
ab ab
d e f
Practice
13.5
a
b
ab
ab ab
bb
a
b
bc
d
bc cd e
d
E
H
E
D
gH
E
D
tH
E
HD
g
E
HD
t
LH
LH
D
t0
5
10
15
Non-consumable grains
b
a ab ab ab
ab
ab
c
a
a
ab a
b
a
b
h
PracticePractice
Fig. 2. Quality defects on shelled maize grain in Long (black) and Seloto (grey). EH: cobs harvested early, de-husked, shelled, grain bagged; EDgH: cobs harvested early, dried on
bare ground, de-husked, shelled, grain bagged; EDtH: cobs harvested early, dried on tarpaulin, de-husked, shelled, grain bagged; EHDg: cobs harvested early, de-husked, dried on
bare ground, shelled, grain bagged; EHDt: cobs harvested early, de-husked, dried on tarpaulin, shelled, grain bagged; LH: cobs harvested late, de-husked, shelled, grain bagged;
LHDt: cobs harvested late, de-husked, dried on tarpaulin, shelled, grain bagged. Regular letters compare means within practices in Long; italic letters compare means within
practices in Seloto.
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 101517 5population: 0e22.4 counts/kg; broken grains: 0.0e3.7%; shriveled
grain: 0.0e4.2%; rodent damaged grain: 0.0e4.0%; insect damaged
grain: 0.0e1.3%; non-consumable grain: 0.2e9.7%; overall losses
1.4e14.8%.
Preliminary MANCOVA was performed to examine whether the
various practices as described in Table 1 influenced grain quality as
a whole, i.e. with the various quality parameters analyzed together
as a group. Grain moisture was the covariate. The effect of practice
was statistically significant (Pillai's Trace¼ 1.469; F 48, 492¼ 3.324;
P¼ 0.005; np2¼ 0.239). The covariate was also significant (Pillai's
Trace¼ 0.357; F 8, 77¼ 5.355; P< 0.001; np2¼ 0.357), meaning that
grain moisture influenced the observations on some of the pa-
rameters. Specifically, grain moisture was significant on insect
population (P< 0.001, np2 0.206), insect damage (P< 0.001, np2 0.178),
moldy/diseased/discolored grain (P< 0.001; np2 0.125), non-
consumable grain (P< 0.001, np2 0.169), and losses (P¼ 0.003,
np2¼ 0.097). The significance of practice was on all parameters
except insect damage (broken grain: P< 0.001, np2¼ 0.463; rodent
damage: P< 0.001, np2.290; moldy/diseased/discolored grain:
P< 0.017; np2 0.267; shriveled grain: P¼ 0.025, np2¼ 0.154; non-
consumable grain: P< 0.001; np2¼ 0.413; insect population(P¼ 0.015; np2¼ 0.168; losses: P< 0.001; np2¼ 0.278). The separation
of means on the basis of practice is given in Fig. 2. A separate
preliminary MANCOVA examined the effect of agro-location. The
covariate (grainmoisture)was not significant (Pillai's Trace¼ 0.166;
F 8, 82¼ 2.042; P< 0.051; np2¼ 0.166), whereas agro-location was
significant (Pillai's Trace¼ 0.593; F 8, 82¼14.96; P< 0.001;
np2¼ 0.593) on insect population (P< 0.001, np2¼ 0.262), insect
damaged grain (P< 0.001, np2¼ 0.137), shriveled grain (P< 0.001,
np2¼ 0.243), moldy/diseased/discolored grain (P< 0.001,
np2¼ 0.254) and non-consumable grain (P¼ 0.003, np2¼ 0.092).
Thus agro-location influenced all the examined quality parameters
except rodent damage, amount of broken grain, and the overall
losses.
An expanded was MANCOVA further explored the effects of the
individual operations or practice aspects. Agro-location was
incorporated in the analysis to examine the potential interaction of
specific operational actionswith climatic conditions. The results are
presented in Table 2. The covariate (grain moisture) was significant,
and explained 34% of the variability. Similarly, agro-location, har-
vesting time, and drying were significant, whereas de-husking was
not, although it accounted for 16% of the observed variability in
Table 2
MANCOVA output for quality of maize prior to storage.
Effect Pillai's Trace F hypothesis df, error df P-value Effect size (np2)
Intercept 0.234 2.676 8,70 0.013 0.234
Grain moisture content (covariate) 0.343 4.569 8,70 0.000 0.343
Agro-location 0.510 9.106 8,70 0.000 0.510
Harvesting time 0.355 4.824 8,70 0.000 0.355
De-husking 0.161 1.682 8,70 0.118 (NS) 0.161
Drying 0.652 4.296 16,142 0.000 0.326
Agro-location * harvesting time 0.256 3.009 8,70 0.006 0.256
Agro-location *De-husking 0.254 2.985 8,70 0.006 0.254
Agro-location * Drying 0.486 2.848 16,142 0.000 0.243
Harvesting time * Drying 0.548 10.596 8,70 0.000 0.548
De-husking *Drying 0.178 1.896 8,70 0.074 (NS) 0.178
Agro-location * Harvesting time * Drying 0.302 3.793 8,70 0.001 0.302
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 1015176grain quality. Agro-location was the most significant main effect
accounting for 51% of the variability. Harvesting time, and drying
explained 36 and 33% of the variability, respectively. Several
interaction effects were significant and these are also presented in
Table 2. Harvesting time * drying was the most significant two-way
effect; the interaction explained 55% of observed variability. A
three-way effect: agro-location * harvesting time * drying was also
significant and explained 30% of observed variability. The actual
effects on specific quality parameters are summarized in Table 3.
Insect population was explained by grain moisture and the in-
teractions agro-location * de-husking and agro-location * drying.
The insect damage levels were also explained by grainmoisture and
agro location * de-husking, as well as drying. Insect counts were
>10 times higher in Long village (6.4 ± 0.9 adults/kg) compared to
Seloto (0.6 ± 0.2 adults/kg). Drying the maize cobs with the husks
whether on the ground or on tarpaulin (practices EDtH, EDgH)
resulted in higher insect infestation. The insect damage levels were
also higher by factors of 1.8 and 2.2, respectively, when cobs were
dried with the husks compared to when the cobs were de-husked
and then dried (practices EHDt, EHDg, LHDt). This effect was
observed in Long and not in Seloto where insect populations were
rather low.
The levels of moldy/diseased/discolored grain were explainedTable 3
Significance of main effects and interaction effects on quality parameters of harvested g
Effect Insect
population
Insect
damage
Moldy/diseased/disco
grain
Moisture 0.011 (0.063)1 0.041
(0.053)
NS
Agro-location 0.000 (0.163) 0.001
(0.130)
0.001 (0.130)
Harvesting time NS2 0.006 (0.094)
De-husking NS NS NS
Drying NS 0.009
(0.114)
0.007 (0.122)
Agro-location * harvesting time NS NS NS
Agro-location * de-husking 0.032 (0.056) 0.005
(0.093)
NS
Agro-location *drying 0.048 (0.076) NS NS
Harvesting time *drying NS NS 0.024 (0.065)
De-husking *drying NS NS NS
Agro-location * harvesting time
*drying
NS NS NS
1P -value (Effect size).
2NS¼ not significant.by agro-location, and harvesting time * drying. The levels were two
times higher in Long (13.4 ± 0.8%) than Seloto (7.5 ± 0.5%). Also, in
both agro-locations, late-harvested maize (Long: 14.5%, Seloto:
7.9%) had twice the amount of moldy/diseased/discolored grain
than early-harvested maize (Long: 6.5%, Seloto: 3.8%). However, the
levels increased two-to three-fold when the early-harvest maize
was dried. Drying on the ground (practices EDgH, EHDg) resulted in
higher levels of moldy/diseased/discolored grain by factor of 1.2
compared to drying on tarpaulin (practice EHDt). Moreover, drying
the sheathed cobs on ground (practice EDgH) resulted in the
highest amounts of moldy/diseased/discolored grain (Long 20.1%,
Seloto 9.5%).
The amount of broken grain was dependent mainly on har-
vesting time * drying. This interaction explained 23% of the vari-
ability in levels of broken grain. Late-harvested maize dried on
tarpaulin (practice LHDt), and early-harvested maize that was not
dried at all (practice EH) had higher amounts of broken grains. For
shriveled grain, harvesting time * drying was the most significant
effect (15% of the variability) although other interactions (agro-
location * harvesting time; agro-location * de-husking) were also
significant, accounting for 10% and 5% of the variability, respec-
tively. The levels of shriveled maize were 4 times higher in Seloto
than Long, and the early-harvestedmaize in Seloto had significantlyrain prior to storage.
lored Broken
grain
Shriveled
grain
Rodent
damage
Non- consumable
grain
Loss
NS 0.070 (0.040) NS 0.000 (0.179) 0.000
(0.243)
NS 0.000 (0.206) NS NS 0.060
(0.045)
NS NS NS 0.000 (0.251) 0.000
(0.246)
0.006
(0.093)
NS NS NS NS
0.006
(0.125)
NS 0.000 (0.236) 0.001 (0.161) 0.014
(0.105)
0.044
(0.052)
0.006 (0.095) NS 0.001 (0.146) 0.001
(0.124)
0.031
(0.057)
0.037 (0.053) 0.003 (0.105) NS NS
NS NS 0.000 (.206) NS 0.040
(0.080)
0.000
(0.230)
0.000 (.151) 0.000 (.212) 0.000 (0.405) 0.000
(0.301)
NS .024 (.064) NS NS 0.024
(0.065)
NS NS NS 0.000 (0.240) 0.000
(0.157)
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 101517 7higher amounts, an indication of poor filling. Drying increased the
levels further particularly when the maize was dried on tarpaulin,
probably due to grain shrinkage. Rodent damage was explained by
agro-location * de-husking (effect size: 11%), agro-location * drying
(effect size: 21%), and harvesting time *drying (effect size: 21%).
There was higher rodent damage in Seloto, and the magnitude was
higher by factor of 1.5 when harvested cobs were dried on the
ground. Furthermore, the maize dried with the husks had higher
rodent damage levels in Long. Late-harvested maize (practice LH)
had higher rodent damage compared to early-harvested maize
(practice EH) but drying also increased rodent damage levels in the
early-harvested cobs probably because of longer exposure periods
that were need to achieve sufficient drying, alongside other factors
such as better camouflage of rodents within the sheathed cobs.
Agro-location * harvesting time * drying was significant for the
amount of grains that were unfit for human consumption. In
addition, de-husking was significant as a main effect, accounting
for 16% of the variability, compared to 24% accounted for by the
three-way effect. Analogously, these effects were significant for
losses, which averaged 7.48 ± 3.28%. The loss levels did not differ
across agro-locations for the early-harvested maize but were
higher in Long (8.4%) than Seloto (6.4%) for the late-harvested
maize, which was probably because of the higher insect and
mold incidence levels in Long. Also, the early-harvested cobs that
were de-husked and then dried on tarpaulins (practice EHDt)
exhibited lower loss levels by up to 5 percentage points in Long
compared to the other practices. In Seloto, the loss levels did not
vary with the various practices.
3.4. Changes in grain quality during storage
A preliminary repeated measures ANOVA (RM-ANOVA) with
practice as the between-subject factor tested whether practice (see
Table 1) had significant effect on selected parameters. After con-
firming a significant effect, a further expanded RM-ANOVA tested
the effects of the individual aspects of practice (harvesting time, de-
husking, drying) and agro-location, as well as the interactions of
these with time. The statistical outputs are summarized in Table 4.
Generally there was no difference between early-and late-harvest
maize on all the quality parameters examined. Strikingly, however,
the levels of all parameters except grain moisture, rodent damage,
and Tribolium spp counts (secondary insect pest) remained distinct
by agro-location. The levels of moldy/diseased/discolored grain,
non-consumable grain and overall losses were distinct depending
on whether cobs were de-husked or not during drying, whereas
grain moisture, Tribolium counts, shriveled grain, and broken grain
levels were distinct based on the drying approach applied (drying
on the ground, drying on tarpauling or not drying at all). Profile
plots showing estimated marginal means of some of the parameterTable 4
Grain quality during storage as influenced by harvesting time, de-husking, drying, and a
Effect of time
F hypothesis df, error df P-value (np2)
Grain moisture 95.06 2.92, 122.44 0.000 (0.694)
Moldy/diseased/discolored grain 1.93 3.12, 122.20 NS
Sitophillus spp 61.11 3.058, 128.43 0.000 (0.593)
Tribolium spp 16.37 1.85, 77.80 0.000 (0.280)
Insect damaged grain 112.80 2.92, 122.72 0.000 (0.729)
Rodent damage grain 1.81 2.37, 99.43 NS
Shriveled grain 11.70 2.15, 96.74 0.000 (0.226)
Broken grain 30.67 3.85, 157.66 0.000 (0.428)
Non-consumable grain 20.08 2.82, 118.62 0.000 (0.323)
Losses 91.71 2.80, 86.92 0.000 (0.750)
(np2): Effect size; NS: not significant.s as a function of time are presented in Fig. 3, Fig. 4, and Fig. 5.
3.4.1. Grain moisture
Grain moisture declined steadily. From the preliminary RM-
ANOVA, effect of practice was significant as well as the interac-
tion of time and practice (F19.88, 141.97¼4.104; P ¼< .000;
np2¼ 0.334). The expanded RM-ANOVA showed that the significant
effect of practice was on drying (Table 4). The maize dried on
tarpaulin maintained significantly lower moisture levels for nearly
18 weeks (Fig. 3c). Early- and late-harvested maize lots were not
different in terms of the moisture level during storage. Similarly,
moisture contents of grain shelled frommaize cobs that were dried
with and without the husks did not differ. However, the interaction
effects of the three practice aspects with time were significant (see
Fig. 3aec) suggesting differences in the in-situ drying dynamics of
stored grain depending on the initial moisture levels, as a result of
the different harvesting and handling practices. Compared to the
drier grain, thewet grain continued to losemoisture, apprentlly at a
faster rate evenutally attaining a a lower moisture level. Moreover,
moisture contents of grain stored in the two agro-locations were
not distinct (P¼ 0.697), and significant interaction effect between
time and agro-location was observed (see Fig. 3d). The higher grain
moisture in Seloto after 12 weeks of storage may be explained by
the higher atmospheric r.h., which might have resulted in lower
rate of in-situ drying of the stored maize.
3.4.2. Moldy, diseased, discolored grain
The effect of time was not significant for the levels of moldy/
diseased/discolored grain. Also, preliminary RM-ANOVA showed
that practice (Table 1) was not significant (P¼ 0.406; np2¼ 0.125).
The expanded test, however, showed that de-husking was signifi-
cant (Table 4); grain from the maize cobs dried with the husks
retained higher levels of moldy/diseased/discolored grain
throughout (see Fig. 3f). Indeed, early harvested-cobs that were
dried on tarpaulin after de-husking (practice EHDT) had the lowest
levels of moldy, discolored, and diseased grain); the levels were 1.7
times lower compared to the early-harvested maize cobs dried on
the ground with the husks (practice EDgH). The levels of moldy/
diseased/discolored grain in the cooler location were consistently
higher by factor of 1.6. The interaction of agro-location with time
was also significant (P¼ 0.022; Fig. 3h).
3.4.3. Insect population and damage
Sitophilus and Tribolium spp were identified during storage. The
effect of practice on Sitophilus counts was not significant (P¼ 0.883)
and follow-up tests showed that the population were not different
as a result of harvesting time, de-husking, or drying (Fig. 4aec). The
effect of agro-location was, however, significant; Sitophilus spp
counts were consistently higher in Long village through the entiregro-location.
Between-factor effects (P-value (np2))
Harvesting time De-husking Drying Agro-location
NS NS 0.012 (0.189) NS
NS 0.033 (.117) NS 0.000 (0.317)
NS NS NS 0.015 (0.133)
NS NS 0.012 (0.190) NS
NS NS NS 0.000 (0.729)
NS NS NS NS
NS NS 0.034 (0.156) 0.000 (0.436)
NS NS 0.001 (0.302) 0.010 (0.151)
0.088 (0.079) 0.036 (0.116) NS 0.000 (0.302)
NS 0.030 (0.156) 0.054 (0.171) 0.002 (0.356)
Agro-locationHarvesting time De-husking Drying 
M
oi
st
ur
e 
co
nt
en
t
(%
 w
/w
)
0 12 240 12 24
6
12
18
0 12 24 0 12 24
Long 
Seloto 
12
16
Early 
Late 
AŌer drying 
Before drying 
None 
On the ground 
On tarpaul in 
***
Time (weeks)
**
****** *
** **
M
ol
dy
/ d
is
ea
se
d/
  
di
sc
ol
or
ed
 g
ra
in
 (%
 w
/w
)
**
***
**
a b c d
e f g h
Fig. 3. Profile plots for marginal means of grain moisture, and moldy/diseased/discolored grain by harvesting time, de-husking, drying, and agro-location as categorical variables.
Blue asterisk: depicted categories of the particular variable are significantly different; Green asterisk: there is significant interaction with time. ***P < 0.001; **P < 0.05; *P< 0.1. (For
interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
Long 
Seloto 
0
120
240
Early 
Late 
AŌer drying 
Before drying  
No drying 
On the ground 
On Tarpaulin 
0
25
50
0 12 240 12 24
0
40
80
0 12 24 0 12 24
In
se
ct
 d
am
ag
e
(%
 w
/w
)
Agro-locationHarvesting time De-husking Drying 
***
**
Time (weeks)
Tr
ib
ol
iu
m
 s
pp
 s
pp
. 
(c
ou
nt
s 
/k
g)
S
ito
ph
ilu
s 
sp
p 
sp
p.
 
(c
ou
nt
s 
/k
g)
**
***
**
**
a b c d
e f g h
i j k l
Fig. 4. Profile plots for marginal means of insect counts and insect damage by harvesting time, de-husking, drying and agro-location as categorical variables. Blue asterisk: depicted
categories of the particular variable are significantly different; Green asterisk: there is significant interaction with time. ***P < 0.001; **P < 0.05; *P< 0.1. (For interpretation of the
references to color in this figure legend, the reader is referred to the Web version of this article.)
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 1015178
Long 
Seloto 
4
8
12
Early 
Late 
AŌer drying 
Before drying  
No drying 
On the ground 
On Tarpaulin 
0 12 240 12 24
0
16
32
0 12 24 0 12 24
Agro-locationHarvesting time De-husking Drying 
Time (weeks)
***
**
***
** *
* ***
** * ***
G
ra
in
 u
nf
it 
fo
r 
co
ns
um
pt
io
n 
(%
w
/w
)
Lo
ss
es
 (%
 w
/w
)
a b c d
e f g h
Fig. 5. Profile plots of marginal means for proportion of grain unfit for consumption and overall losses by harvesting time, de-husking, drying, and agro-location as categorical
variables. Blue asterisk: depicted categories are significantly different; Green asterisk: there is significant interaction with time. ***P < 0.001; **P < 0.05; *P< 0.1. (For interpretation
of the references to color in this figure legend, the reader is referred to the Web version of this article.)
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 101517 9storage period (Fig. 4d). For Tribolium spp, practice was significant
(P¼ 0.029) and follow-up analysis showed that the significance
was on drying (Fig. 4g). The interaction of drying method with time
was also significant (Fig. 4g); the maize dried on the ground
(practices: EDgH and EHDg) had significantly higher Tribolium spp.
counts from 18 weeks onwards suggesting that infestations were
picked from the soil. Furthermore, the maize dried with the husks
had comparatively higher Tribolium counts (Fig. 4f) although the
effect was not significant. Unlike Sitophilus, the Tribolium counts did
not differ with agro-location (P¼ 0.247; Fig. 4h).
Insect damage increased steadily reaching 69% and 94% at 30
weeks in Seloto and Long, respectively, and the effect of practice
was not significant (P¼ 0.588). Interaction of practice and timewas
also not significant (P¼ 0.834). Moreover, harvesting time, de-
husking and drying approaches did not result in distinct insect
damage levels (Fig. 4iek). The effect of agro-locationwas significant
(Fig. 4l). Interaction of agro-location and time was also significant
(Fig. 4h). The upsurge in insect damage commenced earlier in Long
(6 weeks after storage), than Seloto (12 weeks) following the same
trends observed for Sitophilus spp. populations.3.4.4. Rodent damaged, shriveled and broken grains
The amounts of rodent damaged, shriveled and broken grains
remained generally low (<2%). There was no significant effect of
time, and the effects of agro-location, harvesting time, de-husking
and drying were not significant. For shriveled and broken grain,
however, effect of time was significant; the amounts decreased
steadily to <0.5%. Practice was significant on broken (P< 0.000) but
not on shriveled grain (P¼ 0.171). The follow-up RM-ANOVA
showed that drying was the significant aspect of practice (Table 4);
the maize dried on tarpaulin, consistently retained had higher
levels of broken and shriveled grains. The effect of agro-location
was significant as well. The amounts of shriveled and broken
grain decreased significantly (to <0.4%) but the levels remainedconsistently higher in Seloto than Long (shriveled grain: P< 0.000;
broken grain: P¼ 0.010).3.4.5. Non-consumable grain, and overall losses
Part of the damaged grain was unfit for human consumption. A
marked increase in this fraction occurred from 12 weeks onwards
(Fig. 5). The preliminary RM-ANOVA showed that practice was not
significant on non-consumable grain (P¼ 0.299). However, the
expanded analysis revealed that de-husking was significant
(P¼ 0.025); the maize dried with husks had higher amounts of non-
consumable grain in the early and advanced stages of storage
(Fig. 5b). Harvesting time was also significant at P< 0.1 (P¼ 0.088);
the early-harvested maize contained lower amounts on non-
consumable grains in the earlier stages of storage (Fig. 5a) but not in
the later stages probably because of advanced insect damage. Effect of
agro-locationwassignificantand the interactionof agro-locationwith
timewasalsosignificant(Fig.5d);higheramountsofnon-consumable
grain were determined in Long and the levels increased at an
increasing rate while they increased at a decreasing rate in Seloto.
The effect of time on losses was likewise significant, and the
effect of de-husking as well (Table 4). Interaction of de-husking
with time was also significant (P¼ 0.044; Fig. 5f). The maize
dried without the husks had significantly lower loss levels as
storage progressed, which coincidedwell with lower insect damage
levels (though not significant), and lower levels of moldy/diseased/
discolored grain. The effect of drying was also significant at P< 0.1
(P¼ 0.054) whereby maize dried on tarpaulin exhibited lower loss
levels from 12 weeks onwards (Fig. 5g). Effect of agro-location was
significant on level of losses, and the interaction with time was
highly significant (Fig. 5f). On average, the losses increased five-fold
(from 7.7% to 36.3%) in Long, and three-fold (from 7.5% to 25.3%) in
Seloto, over the storage period. Thus the rate of accumulation of
losses was higher in Long. Furthermore the build-up of losses also
commenced 6 weeks earlier compared to Seloto.
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 101517104. Discussion
4.1. Effect of agro-location and harvesting/handling on quality of
the pre-stored grain
The results of the present study show that there were significant
differences in the quality of maize grain stored in the two agro-
locations. The differences emanated from the interaction between
climatic conditions and the harvesting and grain handling practices
applied by farmers, and reflected on almost all quality parameters.
A key quality parameter of harvested grain is moisture. There were
conspicuous differences in the m.c of early- and late-harvest maize
in the two agro-locations. Moisture content of 22e25% is consid-
ered ideal for efficient harvesting (Nielsen, 2018) although field
drying to ~18% m.c. is sometimes considered sensible to reduce
drying costs. Early harvesting was donewithin the proper timing in
the cooler agro-location, but was somewhat late in the warmer
agro-location because of rapid dry-down. Maize kernels are
considered physiologically mature when the m.c. is 30e35%
(Brooking, 1990). At this stage, grain filling ceases, and so maturity
is noticeable with the apparent blackening of the kernel's placental
region. Depending on atmospheric conditions, and the ear and husk
characteristics of the maize variety (e.g. number and thickness of
husk leaves, husk coverage of the ear, tightness of husk leaves, ear
angle, and properties of the kernel pericarp), the crop would
require a post-maturity dry-down period of 2e4 weeks to reach the
right moisture for harvesting (Cross and Kabir, 1989; Nielsen, 2018).
An ideal waiting period of 3e4 weeks was suggested for humid
zones in East Africa (Kaaya et al., 2005; Alakonya et al., 2008) and
West Africa (Borgemeister et al., 1998). Ordinarily, however,
farmers may recognize the ideal harvest stage when the husks turn
brown and the cobs droop, or when the kernels are hard, glassy,
and resistant to scratching with the thumbnail (De Lucia and
Assennato, 1994).
Despite allowing the crop to wait longer for late harvest (6
weeks after early harvest), grain moisture did not reach the rec-
ommended level for safe storage in the cooler location (average
atmospheric temperature 18.7 
C; r.h. 63%). Similar findings were
reported in Uganda (Kaaya et al., 2005). Thus, in some regions,
maize could potentially enter storage with higher than the rec-
ommended moisture if keenness to ensure effective post-harvest
drying is not observed. The practice of drying the de-husked cobs
on the improved tarpaulin was able to lower grain moisture to the
safe level for both early and late harvest maize in both locations.
This indicates that the practice of de-husking followed by improved
drying would contribute to more uniform grain lots with regard to
moisture content at the time of storage, which is desirable. Other
dying practices applied on early-harvested maize, specifically,
drying the sheathed cobs (practices EDgH, EDtH), or drying the
maize cobs on the ground (practice EHDg, EDgH) did not perform
well. Drying with the husks prevents moisture migration off the
surface of the grain, while drying on the ground may allow the
produce to absorb moisture from the soil.
The present results also show that there was significant differ-
ence in the quality of pre-stored grain with regard to insect pop-
ulations, the amounts of insect damaged grain, moldy/diseased/
discolored grain, shriveled grain, and non-consumable grain in the
two agro-locations. Form the weather data, the two agro-locations
differed in terms of atmospheric temperature and r.h. by 2e3 
C
and 3e5%, respectively. Temperature and r.h. differences within
this range can affect incidence and multiplication of insects (Khaliq
et al., 2014), molds (Magan et al., 2011) and rodents (Edoh-
Ognakossan et al., 2016). The higher Sitophilus spp counts
observed on the pre-stored grain in the cooler location could be
related to the effect of interaction of temperature and r.h. onintrinsic growth rates and population development. The maize
harvested in cooler location had >10 times higher insect counts.
Throne (1994) reported the optimal temperature and r.h. for
growth of maize weevil to be 30 
C and 75%, respectively. But
Okelana and Osuji (1985) also observed that the development of
Sitophilus spp. from egg to adult was best at 25 
C, with variations at
different r.h. levels; low r.h. (e. g. 30e50% at 30 
C) could retard
progeny development and survival. At the time of harvesting
(JulyeAugust), the average conditions in the warmer location
(20 
C; r.h. 60%) would have favored insect multiplication
compared to the average conditions (17 
C; r.h. 55%) in the cooler
location. Thus other factors such as the cultivated varieties (Haines,
1991) and farm practices may have contributed to the difference
observed on insect populations in the two localities. Drying and de-
husking practices contributed to higher levels of insect infestation.
In particular, insect counts and insect damage increased during
drying especially when harvested cobs were dried with the husks.
Drying on bare ground increased insect counts by factor of 3e4
especially in the cooler location, indicating that new infestations
were picked from the soil.
The levels of visibly moldy, diseased and discolored grains were
two times higher in the late-harvested maize, and drying practices
applied on the early-harvested maize in the cooler location
increased the levels by factor of 2e3. These findings show that
proliferations of fungi and bacteria were encouraged by allowing
the crop to stand longer than necessary in the field, and by the
drying actions applied. Delayed harvesting encourages ear rot dis-
ease and molds (Alakonya et al. (2008). In fact, Kaaya et al. (2005)
reported increased levels of Aspergillus, Fusarium, Penicillium and
other mold species on maize when harvesting was delayed. The
increase was more pronounced for Aspergillus spp, reaching ~20
times higher on maize harvested 4 weeks after physiological
maturity. Consequently, late harvesting and some post-harvest
handling practices, could also compromise the nutritional quality
and safety of the produce even before storage (Smith and DiMenna,
2007).
With respect to rodent damage, the interaction of harvesting
time and drying, as well as the interaction of agro-location and de-
husking/drying were significant. Delayed harvesting is associated
with crop exposure to damage by pests including rodents (Abass
et al., 2014). Some rodent species attack maize in the field, and
then migrate to granaries at the end of the harvest season when
food in the field becomes scarce (Edoh-Ognakossan et al., 2016).
However, in the warmer agro-location, drying of the early-harvest
maize on bare ground as opposed to drying on tarpaulin
increased rodent damage two-fold compared to delayed-harvest
maize. Higher diversities and populations of rodents are found in
warmer environments (Edoh-Ognakossan et al., 2016). Thus while
early harvesting is recommended for better harvest quality,
increased crop loss due to rodent damage may occur if subsequent
drying steps do not protect the produce against rodent attack.
The amount of broken grains averaged 0.5e3%. Large amounts of
broken grain encourage insects and micro-organisms, and are
therefore not desirable on grain lots intended for long-term stor-
age. Results of this study showed that the levels of broken grain
were dependent on the interaction of harvesting and drying ac-
tions, as well as the interaction of agro-location and de-husking/
drying. Proper drying of maize before shelling is recommended
for easy and complete stripping of cobs (Srison et al., 2016), and a
direct relationship between grain moisture (range: 14e28%) and
mechanical damage was reported (Nalbant, 1990; Srison et al.,
2016). In the present case, interaction of harvesting time and dry-
ing was the most significant effect explaining the amount of broken
grains. Drying of the late harvest maize on improved tarpaulin
increased the levels by factor of 2e6, especially in the warmer
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 101517 11location. The late-harvest maize dried on the improved tarpaulin
had m.c < 12% which possibly made the grains brittle and easy to
break when placed on the motorized sheller. Other factors such as
physical and morphological characteristics of the cobs, as well as
sheller design (Akubuo, 2002; Nalbant, 1990; Petkevi
cius et al.,
2008; Srison et al., 2016) may contribute to grain breakage. The
former connote how strongly grains are attached to the cob, kernels
ability to deform, kernel size, shape, hardness, and the size and
strength of cobs (Akubuo, 2002). Regarding shriveled grain, higher
levels were determined in the warmer location. This was probably
related to heat stress during maturation, leading to poorer filling of
the endosperm (Begcy and Walia, 2015). High levels of shriveled
grain are undesirable because of poor processing quality; they have
low flour yield, are difficult to de-hull, and the nutritional value is
inferior (Gaines et al., 1997). However, just like the broken grains,
our findings show that multiple interactions were responsible for
the levels of shriveled grain measured on the grain at the pre-
storage stage.
The levels of non-consumable grain and the overall losses were
a function of the interaction of agro-location with harvesting time
and drying. Early harvesting followed by drying the sheathed cobs
on the ground or on tarpaulin, as well as late harvesting, were
characterized by higher levels of grain that were unfit for human
consumption. These were largely the moldy and diseased grains.
Consumers often discard grains that are visibly moldy but such
grain are also fed to poultry and livestock, and are likely to pass
undetected in processed products when unscrupulous processors
blend them with good grain. Hence, these could present a public
health risk if effective separation is not achieved on the farm. The
overall losses were significant; they exceeded 5% (6.9± 3.0%). In the
cooler location, however, early harvesting followed by de-husking
and drying the cobs on improved tarpaulin before shelling could
reduce the pre-storage losses by up to 5 percentage points.
4.2. Effects of agro-location and harvesting/handling practices on
quality during storage
Understanding how the grain quality parameters change during
storage is useful because it provides knowledge of when destruc-
tive damage levels are likely to occur, and can therefore help
farmers to plan better for effective and economical deployment of
interventions. It is also useful for better extension and advisory on
successful use of improved storage technologies. On the overall, the
levels of rodent-damaged, andmoldy/diseased/discolored grain did
not change with time, suggesting that these were not very critical
parameters deserving stringent monitoring in the trial locations,
under ordinary storage. However, the consistently higher levels of
moldy/diseased/discolored grain observed on the maize grain
threshed from cobs that were dried with the husks point to the
importance of good practice. Similarly, the higher levels of moldy/
diseased/discolored grain determined in the cooler agro-location
throughout the storage period suggest that right interventions
would be to address the factors responsible for this defect at the
time of harvesting and preparation of the grain for storage, or even
earlier. Specifically, the interaction between harvesting time and
drying was significant. Early harvesting potentially decreased the
amount of moldy/diseased/discolored grain. However, de-husking
and drying the early-harvested cobs on improved tarpaulin (best
practice) doubled the defect even though it achieved better results
compared to drying the sheathed cobs on bare-ground. This
observation suggests that, in zones similar to the cooler agro-
location, attention should also be given to better drying tech-
niques. Moldy/diseased/discolored grains are indicators of inferior
food quality and inadequate safety. The proliferation of molds
during storage is associated with mycotoxin production, andenvironments that encourage other pests such as mites, as well as
cause loss of nutrients, reduced germination, and unpleasant taste
of food (Gwinner et al., 1996).
As expected, insect population and grain damage levels
increased with storage time. The amounts of broken and shriveled
grain in turn decreased due to attack by insects. While Sitophilus
populations remained distinct by agro-location (also at the onset of
storage) the Tribolium populations were not. The cooler agro-
location thus continued to favor multiplication of Sitophilus spp.
and the attendant insect damage levels, which were equally
distinct by agro-location but not distinct by harvesting, de-husking
and drying approaches. Significant Tribolium spp populations were
noticed after about 12 weeks of storage, which coincided well with
noticeable insect damage of the grain. Another notable observation
was that higher Tribolium counts occurred in the maize that was
dried on the ground. Tribolium spp. prefer environments that are
littered with fine debris, and grain that has not been screened of
fine materials and broken kernels is particularly susceptible to
attack by the pest, that can also survive extremely dry conditions
(Gwinner et al., 1996).
Insect damage levels were high despite the fact that Sitophilus
spp. was the sole primary pest identified (others are Prostephanus
truncatus (Horn) and Rhyzopertha dominica (Fabricius)). Similar to
findings of other recent studies in East Africa (Abass et al., 2018;
Ng'ang'a et al., 2016), we did not encounter P. truncatus. Tradi-
tionally, Sitophilus spp. is understood to cause lower damage and
weight loss than P. truncatus as the feeding habit is less ferocious.
Other researchers also reported enormous grain damage by
S. zeamais in semi-arid agro ecologies in Kenya (Ng'ang'a et al.,
2016) and Tanzania (Abass et al., 2018). The high damage levels
could be attributed to a number of factors, among them susceptible
maize varieties (Khakata et al., 2018) and other conditions favoring
rapid multiplication of grain weevils. Generally, however, harvest-
ing, de-husking and drying operations did not result in statistically
different insect damage levels.
The levels of non-consumable grain as well as the overall losses
increased with time. The effects of post-harvest practices were
apparent as higher levels of non-consumable grainwere associated
with late harvesting and drying cobs with the husks, and higher
losses occurred on the maize that was dried with the husks, or on
the ground. Moreover the non-consumable grain and overall losses
were >2 times higher in the cooler location. Non-consumable grain
connote grade-outs that are separated through sorting during
preparation for consumption or sale. In central Benin, Borgemeister
et al. (1998) reported that losses by insects were more severe
during the storage of late-harvested maize, while early-harvested
maize had a high levels of mold damage. Moreover, the authors
reported higher economic value when the stored maize had been
harvested 3 weeks after physiological maturity. Our results
corroborate these findings, and confirm that proper practice of pre-
storage operations can significantly improve storage. The obser-
vation that higher and conspicuous losses occurred six weeks
earlier in the cooler agro-location was significant. The pattern is
also seen on the build-up of insect populations and insect damage.
Other studies in East Africa showed that dramatic increase in insect
damage levels occurred after 3e4 months of storage, when no in-
terventions are applied (Ng'ang'a et al., 2016). From the present
findings, farmers would need to implement preventive strategies
much earlier in some agro-locations. Nonetheless, it would be
helpful to confirm these findings by sampling from farmer gra-
naries as well.
In the context of market access, our results have implications for
grain trade in the region. The baseline condition of grain prior to
storage was characterized by 4.7e31.8% defective grain. Such is the
quality of maize that farmer would be selling to the market early in
C. Mutungi et al. / Journal of Stored Products Research 84 (2019) 10151712postharvest season, and would require buyers including traders,
processors and consumers to apply quality improvement actions
(e.g. cleaning, sorting and grading). According to East African maize
standards, total defective grain should not exceed 3.2, 7.0 and 8.5%
for grades I, II, III, respectively after adjusting the total damage to a
tolerance factor of 0.7 (EAC, 2013). The high levels of quality defects
potentially deprive farmers of the opportunity to access markets
and better prices. With best practice of early harvesting and drying
the husked cobs on improved tarpaulins, sorting losses would be
minimized to 6e7%. Since sorting is part of quality improvement,
and encouraging farmers to undertake the practice is often part of
the education and training given by extension workers, integrating
pre-harvest approaches to lower these sorting losses would be
judicious. Such approaches include selection of varieties with su-
perior maturing and postharvest traits (such as ear rot resistance,
closed ear tips, drooping ears, hard-to-lodge stems), and better
matching of varieties to their optimal agro-ecologic conditions.
Conflict of interest and authorship conformation form
None.
Acknowledgements
The study was funded by United States Agency for International
Development (Contract number: AID-BFS-G-11-00002) within the
frame of Africa RISING (Research in Sustainable Intensification for
the Next Generation) program. The authors are grateful for the
financial support.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jspr.2019.101517.
References
Abass, A.B., Fischler, M., Schneider, K., Daudi, S., Gaspar, A., Rüst, J., Kabula, E.,
Ndunguru, G., Madulu, D., Msola, D., 2018. On-farm comparison of different
postharvest storage technologies in a maize farming system of Tanzania Central
Corridor. J. Stored Prod. Res. 77, 55e65.
Abass, A.B., Ndunguru, G., Mamiro, P., Alenkhe, B., Mlingi, N., Bekunda, M., 2014.
Post-harvest food losses in a maize-based farming system of semi-arid
savannah area of Tanzania. J. Stored Prod. Res. 57, 49e57.
Akubuo, C.O., 2002. Performance evaluation of a local maize sheller. Biosyst. Eng.
83, 77e83.
Alakonya, A.E., Monda, E.O., Ajanga, S., 2008. Effect of delayed harvesting on maize
ear rot in Western Kenya. Am.-Eurasian J. Agric. Environ. Sci. 4, 372e380, 2008.
Begcy, K., Walia, H., 2015. Drought stress delays endosperm development and
misregulates genes associated with cytoskeleton organization and grain quality
proteins in developing wheat seeds. Plant Sci. 240, 109e119.
Borgemeister, C., Adda, C., Setamou, M., Hell, K., Djomamou, B., Markham, R.,
Cardwell, K., 1998. Timing of harvest in maize: effects on post-harvest losses
due to insects and fungi in central Benin, with particular reference to Proste-
phanus truncatus Horn coleoptera: Bostrichidae. Agric. Ecosyst. Environ. 693,
233e242.
Boxall, R.A., 2002. Damage and loss caused by the larger grain borer Prostephanus
truncatus. Integr. Pest Manag. Rev. 7, 105e121.
Brooking, I.R., 1990. Maize ear moisture during grain-filling, and its relation to
physiological maturity and grain-drying. Field Crop. Res. 23, 55e68.
Chegere, M.J., 2018. Post-harvest losses reduction by small-scale maize farmers: the
role of handling practices. Food Policy 77, 103e115.
Cross, H.Z., Kabir, K.M., 1989. Evaluation of field dry-down rates in early maize. Crop
Sci. 29, 54e58.
De Lucia, M., Assennato, D., 1994. Agricultural Engineering in Development: Post-
harvest Operations and Management of Food Grains, vol. 93. FAO AgriculturalServices Bulletin No., Rome.
EAC (East African Community), 2013. East African Standard: Maize Grains d
Specification. EAS 2:2013, third ed., pp. 1e14
Edoh-Ognakossan, K., Affognon, H.D., Mutungi, C.M., Sila, D.N., Soul-Kifouly, G.,
Midingoyi, S.-K., Owino, W.O., 2016. On-farm maize storage systems and rodent
postharvest losses in six maize growing agro-ecological zones of Kenya. Food
Security 8, 1169e1184.
Gaines, C.S., Finney, P.L., Andrews, L.C., 1997. Influence of kernel size and shriveling
on soft wheat milling and baking quality. Cereal Chem. 74, 700e704.
Gwinner, J., Harnisch, R., Muck, O., 1996. Manual of the Prevention of Post-harvest
Grain Losses. Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ)
GmbH, p. 338.
Haines, C.P., 1991. Insects and Arachnids of Tropical StoredeProducts: Their Biology
and Identification: a Training Manual. Natural Resources Institute UK.
Hou, A.Y., Kakar, R.K., Neeck, S., Azarbarzin, A.A., Kummerow, C.D., Kojima, M.,
Oki, R., Nakamura, K., Iguchi, T., 2013. The global precipitation measurement
mission. Bull. Am. Meteorol. Soc. 95, 701e722.
Huffman, G.J., Bolvin, D.T., Nelkin, E.J., Wolff, D.B., Adler, R.F., Gu, G., Hong, Y.,
Bowman, K.P., Stocker, E.F., 2007. The TRMMmultisatellite precipitation analysis
(TMPA): quasi-global, multiyear, combined-sensor precipitation estimates at
fine scales. J. Hydrometeorol. 8, 38e55.
Kaaya, A., Warren, L.H., Kyamanywa, S., Kyamuhangire, W., 2005. The effect of
delayed harvest on moisture content, insect damage, molds and aflatoxin
contamination of maize in Mayuge district of Uganda. J. Sci. Food Agric. 85,
2595e2599.
Khakata, S., Mbute, F.N., Chemining’wa, G., Mwimali, M., Karanja, J., Harvey, J.,
Mwololo, J.K., 2018. Post-harvest evaluation of selected inbred lines to maize
weevil Sitophilus zeamais resistance. J. Plant Breed Crop Sci. 10, 105e114.
Khaliq, A., Javed, M., Sohail, M., Sagheer, M., 2014. Environmental effects on insects
and their population dynamics. J. Entomol. Zool. Stud. 2, 1e7.
Magan, N., Medina, A., Aldred, D., 2011. Possible climate change effects on myco-
toxin contamination of food crops pre- and post-harvest. Plant Pathol. 60,
150e163.
Mendoza, J.R., Sabillon, L., Martinez, W., Campabadal, C., Hallen-Adams, H.E.,
Bianchini, A., 2017. Traditional maize post-harvest management practices
amongst smallholder farmers in Guatemala. J. Stored Prod. Res. 71, 14e21.
Minot, N., 2010. Staple food prices in Tanzania. In: Contributed Paper Prepared for
the COMESA Policy Seminar Maputo, Mozambique, 25e26 January.
Nalbant, Y.I.O., 1990. Mechanical damage of corn kernel in shelling corn ear, 21.
Agricultural Mechanization in Asia, Africa & Latin America, pp. 37e40.
Ng'ang'a, J., Mutungi, C., Imathiu, S.M., Affognon, H., 2016. Low-permeability triple-
layer plastic bags prevent losses of maize caused by insects in rural on- farm
stores. Food Security 8, 621e633.
(Bob) Nielsen, R.L., 2018. Field Dry-Down of Mature Corn Grain. Corny News
Network, Purdue Univ. Online at. https://www.agry.purdue.edu/ext/corn/news/
timeless/graindrying.html. (Accessed 30 January 2019).
Okelana, F.A., Osuji, F.N.C., 1985. Influence of relative humidity at 30
C on the
oviposition, development and mortality of Sitophilus zeamais Motsch. (Cole-
optera: Curculionidae) in maize kernels. J. Stored Prod. Res. 21, 13e19.
Petkevi
cius, S., 
Spokas, L., Steponavi
cius, D., 2008. Substantiation of technological
parameters of wet maize ear threshing. Agron. Res. 6, 271e280.
Smith, W.A., Di Menna, M.E., 2007. Influence of harvest date and hybrid on the
mycotoxin content of maize (Zea mays) grain grown in New Zealand. N. Z. J.
Crop Hortic. Sci. 35, 331e340.
Srison, W., Chuan-Udom, S., Saengprachatanaraka, K., 2016. Effects of operating
factors for an axial-flow corn shelling unit on losses and power consumption.
Agric. Nat. Res 50, 421e425.
Stathers, T., Lamboll, R., Mvumi, B.M., 2013. Postharvest agriculture in changing
climates: its importance to African smallholder farmers. Food Security 53,
361e392.
Subedi, K., 1996. Effect of leaf stripping, de-tasseling and topping of maize on the
yield of maize and relay intercropped finger millet. Exp. Agric. 32, 57e61.
Suleiman, R.A., Kurt, R.A., 2015. Current maize production, postharvest losses and
the risk of mycotoxins contamination in Tanzania. In: 2015 ASABE Annual In-
ternational Meeting. American Society of Agricultural and Biological Engineers.
https://lib.dr.iastate.edu/abe_eng_conf/442/. (Accessed 16 February 2019).
Throne, J.E., 1994. Life history of immature maize weevils (Coleoptera: Curculioni-
dae) on corn stored at constant temperatures and relative humidities in the
laboratory. Environ. Entomol. 23, 1459e1471.
TNBS (Tanzania National Bureau of Statistics), 2019. 2016/17 Annual Agriculture
Sample Survey (Crop and Livestock) Final Report. https://www.nbs.go.tz/nbstz/
index.php/english/statistics-by-subject/agriculture-statistics. (Accessed 17
February 2019).
Tolera, A., Sundstøl, F., 1999. Morphological fractions of maize stover harvested at
different stages of grain maturity and nutritive value of different fractions of the
stover. Anim. Feed Sci. Technol. 81, 1e16.