PLOS ONE
RESEARCH ARTICLE
Evaluating combined effects of pesticide and
crop nutrition (with N, P, K and Si) on weevil
damage in East African Highland Bananas
Hannington Bukomeko 1,2
ID *, Godfrey Taulya2, Antonius G. T. Schut1, Gerrie W. J. van de
Ven1, Jerome Kubiriba3, Ken Giller 1
ID
1 Plant Production Systems Group, Wageningen University & Research, Wageningen, The Netherlands,
2 International Institute of Tropical Agriculture, Kampala, Uganda, 3 National Agricultural Research
Laboratories, Kawanda, Uganda
a1111111111
a1111111111 * hannington.bukomeko@wur.nl, B.Hannington@cgiar.org, hbukomeko@gmail.com
a1111111111
a1111111111
a1111111111
Abstract
Banana weevil (Cosmopolites sordidus, Germar) is a major pest in East African Highland
Banana. The influence of crop nutritional status on weevil damage is poorly understood.
OPEN ACCESS Nutrient availability affects the nutritional quality of plants for weevils and may affect weevil
Citation: Bukomeko H, Taulya G, Schut AGT, van damage. Here, we evaluate the effect of insecticides alone and in combination with fertili-
de Ven GWJ, Kubiriba J, Giller K (2023) Evaluating sers (N, P, K and Si) on weevil damage using data from two experiments in central and
combined effects of pesticide and crop nutrition southwest Uganda. In the first experiment, we varied chlorpyrifos and application rates of N,
(with N, P, K and Si) on weevil damage in East
P and K. In the second experiment, we varied the application rates of K and Si. Treatment
African Highland Bananas. PLoS ONE 18(3):
e0282493. https://doi.org/10.1371/journal. effects were analysed using generalised linear mixed models with a negative binomial distri-
pone.0282493 bution. In the first experiment, chlorpyrifos reduced and N increased weevil damage, while P
Editor: Sengottayan Senthil-Nathan, and K had no significant effect. In the K or Si application rates reduced weevil damage com-
Manonmaniam Sundaranar University, INDIA pared with the control. We conclude that the combined application of chlorpyrifos with K and
Received: September 2, 2022 Si fertilisers can contribute to weevil damage control on sites with low nutrient availability
and should form part of integrated weevil management in bananas. Future studies should
Accepted: February 15, 2023
assess how much reduction in insecticide use is possible in EAHB with judicious input rates.
Published: March 10, 2023
Copyright: © 2023 Bukomeko et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original 1. Introduction
author and source are credited.
The productivity of East African Highland Bananas (EAHBs) in Uganda is 10 to 20 t ha-1 y-1
Data Availability Statement: The data for this [1], barely a third of the attainable yield of 60–70 t ha-1 y-1 [2]. Yield is mostly constrained by
paper is stored here: https://models.pps.wur.nl/ drought, nutrient limitations and pest damage [1]. The banana weevil (Cosmopolites sordidus,
effect-nitrogen-potassium-and-silicon-weevil-
Germar) is a major banana pest that can cause up to 44% yield loss by the third crop cycle [3].
damage-east-african-highland-banana.
Weevil larvae damage the corm and, hence, interfere with nutrient uptake and transport, wors-
Funding: The study was conducted within the ening nutrient shortages [4]. Sometimes, EAHBs may not even respond to fertilizers without
framework of the project “Improving Scalable
controlling weevil damage first [5].
Banana Agronomy for Small Scale Farmers in
Weevil damage control options include chemical control, cultural control practices (e.g.
Highland Banana Cropping Systems in East Africa”
led by the National Banana Research Program crop sanitation and clean planting materials) and other agronomic practices like good nutri-
(NBRP) of the National Agricultural Research tional management [4]. None of these methods is completely effective, hence the advice for
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 1 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
Organization (NARO) with financial support from integrated pest management (IPM) with a mix of crop management actions that complement
the Bill & Melinda Gates Foundation (BMGF), Grant each other to augment weevil damage control [4]. IPM reduces the need for insecticides, the
ID: OPP1134098. The funding agency had no role
risk of insecticide resistance and can limit unintended negative effects on non-target species.
in study design, data collection, preparation of the
Using a combination of fertilisers and insecticides, Kagoda et al. [6] attempted to rehabilitate a
manuscript or decision to publish. https://www.
gatesfoundation.org/about/committed-grants. heavily weevil-infested plantation but failed because the weevil control interventions started
too late (beyond the 5th crop cycle) and instead recommended replanting rather than rehabili-
Competing interests: The authors have declared
tating. It, therefore, remains to be seen if the combined application of insecticide and fertiliser
that they have no competing interests with respect
to this work. can contribute to weevil control.
Fertilizer applications and water management affect pest damage by altering the nutritional
quality of plants to pests. For example, drought stress enhances pest survival among boring
insects but deters free-living chewing insects [7]. High nitrogen (N) intake can promote pest
damage by increasing the concentration of primary metabolites, such as amino acids which is
a nutritional resource for insects. It makes the plant more palatable, nutritious, and digestible
[8]. Conversely, silicon (Si) can suppress damage physically by fortifying cell walls or biochem-
ically by inducing resistance [9,10]. Similarly, potassium (K) can reduce insect damage because
of its role in metabolic pathways, some of which upregulate defence mechanisms or promote
the synthesis of secondary metabolites that make plants less palatable to insect pests [11].
In EAHB, previous studies on weevils and nutrition showed that NPK fertilizer use does
not improve productivity in weevil-infested plants [5] nor affect weevil damage [12]. The wee-
vils attacked vigorously growing plants just as much as drought and nutrient-stressed plants
[13]. These studies, however, applied low rates of fertilizers and combined nutrient rates in a
way that masks individual nutrient effects. For example, [12] combined equal amounts of N
and K at a rate of 50 kg ha-1 y-1. This rate is low and lacks variation in rates of individual nutri-
ents, making it impossible to segregate N and K effects. We are also yet to understand the
effects of water or Si on weevil damage. Si alleviates other biotic stresses in bananas like
Xanthomonas wilt disease [14] in EAHBs caused by Xanthomonas campestris pv. musacearum,
Fusarium wilt disease [15] caused by Fusarium oxysporum and, Black sigatoka [16] in Grand
Nain bananas caused by Mycosphaerella fijiensis (Morelet). This study aimed to evaluate the
effect of the most used insecticide chlorpyrifos in combination with water, N, K and Si on wee-
vil damage in EAHBs. This knowledge can inform best practices for integrated weevil
management.
2. Materials and methods
2.1. Study sites
The first field trial (referred to below as the Nutrient Omission Trial) was established on land
without a history of EAHB cropping in two study areas: Ntungamo (0˚540 S, 30˚150 E, 1405 m.
a.s.l) in south-western Uganda and Kawanda (0˚250 N, 32˚310 E, 1156 m.a.s.l) in central
Uganda. The trial was planted between October and December 2004 and monitored until
2009. A second trial (referred to as the Potassium Response Trial) was established at Kawanda
in December 2018 and monitored until September 2021. The soil type in Ntungamo is a Lixic
Ferralsol while the soil in Kawanda is a Haplic Ferralsol. The soils were generally of low fertility
(Table 1). Rainfall patterns are bimodal with dry spells from June to August and December to
February. Rainfall in Ntungamo ranges from 935 to 1380 mm while rainfall in Kawanda ran-
ged from 1034 to 1663 mm [17]. The climate is typical for much of the EAHB growing areas in
the mid-altitude East African highlands with a mean daily minimum and maximum tempera-
ture that ranges from 13 to 17˚C and 26 to 27˚C, respectively [18,19].
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 2 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
Table 1. Chemical properties of the top 30 cm of soils in the experimental sites.
Soil Location
Chemical properties Kawanda§ Ntungamo§ Kawanda†
Range (Mean) Class Range (Mean) Class Range (Mean) Class
pH (1:2.5) 4.9–6.2 (5.5) Strongly acidic 4.6–5.6 (4.8) Strongly acidic 5.3–6.3 Moderately acidic
(5.8)
Organic matter 1.0–4.6 (2.6) Medium 0.14–1.9 (0.7) Very Low 0.82–4.7 (2.19) Medium
(%)
Nitrogen 0.005–0.2 (0.1) Low 0.04–0.14 (0.07) Low 0.077–0.20 (0.11) Low
(%)
Extractable P 0.7–8.6 (1.8) Low 0.61–38.0 (3.52) Very Low <0.05 Very Low
(mg kg-1)
Exchangeable K (cmolc kg-1) 0.04–1.0 (0.4) Medium 0.02–0.36 (0.12) Low 0.054–0.351 (0.19) Low
Exchangeable Ca (cmolc kg-1) 2.2–8.6 Low 0.47–7.4 Low 2.08–5.462 Low
(4.5) (1.7) (3.6)
Exchangeable Mg (cmolc kg-1) 0.9–2.9 (1.48) Medium 0.01–1.6 (0.45) Low 0.897–1.893 (1.34) Medium
§ Nutrition Omission Trial.
† Potassium Response Trial.
https://doi.org/10.1371/journal.pone.0282493.t001
2.2 Experimental designs and data collection
2.2.1. Nutrient omission trial (2004–2009). A randomized complete block design was
used with four blocks across the slope. Each block had 10 treatments (Table 2) and each treat-
ment consisted of 35 mats laid out in a 5 × 7 arrangement occupying an area of 315 m2. The
inner 3 × 5 mats were sampled. EAHBs of the variety Kisansa were used–a variety susceptible
to weevil damage. The primary nutrients N-P-K-Mg were applied using the mineral fertilizers
urea (CH4N2O), muriate of potash (KCl), triple superphosphate (Ca (H2PO4)2�H2O), and kie-
serite (MgSO4) respectively. Micro-nutrients were applied using sodium molybdate
(Na2MoO4), borax (Na₂ [B₄O₅ (OH) ₄] �8H₂O) and zinc sulphate (ZnSO4). The nutrient rates
in this trial were selected to enable QUEFTS modelling and quantify banana yield response to
nutrient fertilisers. For treatments 1, 5, 8 and 10 (Table 2) with the highest rates of fertilizer, N
and K fertilizers were applied in four splits, two per rainy season. Fertilizers for all other treat-
ments were applied in two splits, one at the start of each rainy season. Weevils were controlled
using chlorpyrifos insecticide of the brand Dursban [20]–sprayed at a recommended rate of
1.03 g per mat per month. Micro-bunds (soil heaped up to 30 cm high around the plot area)
were installed between plots to prevent runoff/run-on.
Table 2. Treatments applied in the nutrient omission trial.
Treatments
Application 1 2 3 4 5 6 7 8 9 10
N (kg ha-1 y-1) 400 - - 150 400 400 400 400 - 400
P (kg ha-1 y-1) 50 - 50 50 - 50 50 50 - 50
K (kg ha-1 y-1) 600 - 600 600 600 - 250 600 - 600
Other nutrients 1 - 1 1 1 1 1 - - 1
Pesticide 1 1 1 1 1 1 1 1 - -
Treatments 1–7 were also used in Nyombi [19] and treatments 1–4 and 6–7 were also used in Taulya [17]. Other nutrients included magnesium, zinc, boron and
molybdenum.
https://doi.org/10.1371/journal.pone.0282493.t002
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 3 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
Weevil damage was assessed on freshly harvested corms of EAHBs [21] for every crop
cycle. A banana plant, is composed of an underground stem called a "corm" whose tissues can
be distinguished into a central cylinder surrounded by an outer cortex. Out of the corm, multi-
ple aerial shoots, known as pseudo-stems emerge sporadically and together they constitute a
banana mat. The reproductive growth phase starts when an inflorescence emerges from the
top of the pseudo stem and develops into a bunch, which at maturity (banana fingers start to
ripen) culminates in the death and decay of the pseudo stem and the portion of corm attached
to it. The cohort of pseudo stems in a given field that emerge from their respective corms
within the same season constitutes a banana crop cycle even though these may be harvested at
different periods of time since their growth and development rates can be different. Weevil
assessment was done on part of the corm directly attached to the harvested pseudo-stem. To
assess weevil damage, two cross-sectional cuts were made through the corm at the collar, i.e.,
at the junction of the pseudo-stem and corm, and 5 cm below the collar. For each cross-sec-
tion, the percentage area of tissue consumed by weevil larvae in the central cylinder and the
cortex were estimated, giving two damage estimates per cross-section. Overall weevil damage
was determined as the mean of these four estimates.
Nyombi [19] used data from this nutrient omission trial to describe the biomass growth
response to fertilizer inputs, while Taulya [22] used it to study the effect of nutrients on
drought tolerance of EAHB. We used data from the nutrient omission trial to examine the
additional effect of fertilizers on weevil damage on top of pesticide use. The setup of a nutrient
omission trial was however not optimal for assessing the effect of potassium on weevil damage
because it lacked sufficient variation in potassium levels with the low/moderate nitrogen rate.
For this, we considered the potassium response trial where potassium was varied while keeping
a moderate rate of nitrogen.
2.2.2. Potassium response trial (2018–2021). The potassium response trial was used to
examine the contribution of K and Si to weevil damage control. This trial had a similar layout
as the nutrient omission trial but with only three blocks and had mixed varieties of EAHBs
randomly distributed in the trial–all susceptible to weevil damage. Each block had 16 treat-
ment plots and each treatment plot included 15 mats at the start of the experiment. In each
block, eight treatment plots were rain-fed, and the other eight plots were drip-irrigated with a
pressure-compensating pump. The irrigation was only done during the dry season and each
irrigation event supplied 30 litres of water per mat within five hours. It was not applied fre-
quently enough to fully prevent water limitation. The primary nutrients N, P and K were
applied using mineral fertilizers urea (CO (NH2)2), muriate of potash (KCl) and triple super-
phosphate (Ca (H2PO4)2�H2O). The rate of nitrogen used in this trial was considered moder-
ate while potassium varied from lowest to maximum plausible for bananas. These rates were
selected to test the effect of varying K without the likely masking effect of high N. The N was
applied in 4 splits (2 times per rainy season, 25 kg N ha-1 per application), adding to a total of
100 kg N ha-1 y-1. P was applied twice a year at the rate of 25 kg P ha-1 at the start of each rainy
season, adding to a total of 50 kg P ha-1 y-1. Varying amounts of K (Table 3) were applied in
four splits. Si was provided as Elkem B–a Si fertilizer containing 45% Si in the form of SiO4 –at
the manufacturer’s recommended rate of 300 kg Si ha-1 y-1 and applied in two splits. Weevils
were controlled with the insecticide chlorpyrifos, sprayed monthly. Weevil damage was
assessed similarly to section 3.2.1 following Gold et al. [21] starting from December 2019 to
September 2021. The weevil damage assessments were done on four of the 15 mats. These four
were chosen randomly but the same four mats were assessed throughout the assessment
period.
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 4 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
Table 3. Treatments applied in the potassium response trial.
Treatments Water Si (kg ha-1 y-1) K (kg ha-1 y-1)
1 Irrigated 0 0
2 Irrigated 300 0
3 Irrigated 0 75
4 Irrigated 300 75
5 Irrigated 0 150
6 Irrigated 300 150
7 Irrigated 0 250
8 Irrigated 0 600
9 Rain-fed 0 0
10 Rain-fed 300 0
11 Rain-fed 0 75
12 Rain-fed 300 75
13 Rain-fed 0 150
14 Rain-fed 300 150
15 Rain-fed 0 250
16 Rain-fed 0 600
https://doi.org/10.1371/journal.pone.0282493.t003
2.3. Data analysis
We visualized the raw data in both trials using a cumulative distribution function of the pro-
portion of weevil damage in the corm for each treatment. To test the effect of predictors on
weevil damage, we fitted generalized linear mixed models (GLMM). In the nutrient omission
trial, predictor variables were binary in the case of chlorpyrifos use, application of phosphorus
(P) and “other nutrients” (magnesium, zinc, boron, molybdenum). N and K were each applied
at three rates. There were four crop cycles. In the potassium response trial, the predictor vari-
ables were binary in the case of irrigation and Si. K was applied at five rates. There were three
crop cycles. The predictor variables were used as fixed factors in the model. The random vari-
ables were mats nested in plots, which in turn were nested in blocks. The GLMM used an
unstructured variance-covariance matrix where it estimates each variance and covariance
directly from the data without constraints [23]. We fitted the GLMM using a negative binomial
distribution with a log-link function instead of the Poisson model which was over-dispersed
[24]. The negative binomial has a dispersion parameter that relaxes the strict Poisson assump-
tion that mean equals variance [24]. Model diagnostic tests like tests for over dispersion, zero
inflation, outliers and patterns in residuals were performed. These tests indicated that the
selected model fitted the data well.
For each trial, we compared various combinations of predictors with and without interac-
tions. Models with the interaction between crop cycle and treatments were not significant and
we instead considered models with main effects of crop cycle plus the various combination of
treatments. Additionally, we considered models specified with the crop cycle as a fixed predic-
tor or as part of the dispersion model and, models specifying nutrient application rates with
more than two rates as either categorical or continuous variables. We selected models with the
lowest value of Akaike information criteria (AIC) and when AIC was not different, we choose
the simpler model [25]. During comparisons, model parameters were estimated using maxi-
mum likelihood with Laplace approximation which gives reliable fit statistics but biased vari-
ance parameter estimates. After model selection, the final models (Model 1 for the nutrient
omission trial & model 2 the for potassium response trial), were refitted with restricted maxi-
mum likelihood with Laplace approximation which gives unbiased variance parameter
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 5 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
estimates.
Weevil damage � N þ P þ K þ Insecticideþ Other nutrient þ ð1jBlock=Plot=MatÞ ð1Þ
Weevil damage � Crop cycleþWater þ K þ Siþ ð1jBlock=Plot=MatÞ ð2Þ
In both models, a restricted maximum likelihood was used in combination with a negative
binomial distribution (family was set to “nbinom2”). In Model 1, N, K and crop cycle were
continuous variables while the other variables were categorical with crop cycle specified as part
of the dispersion model allowing the dispersion parameter to vary with the crop cycle [26]. In
Model 2, all variables were categorical. We used Tukey’s post hoc test to compare contrasts
among K application rates in Model 2.
In the tables, the estimate is either positive to indicate an increase or negative to indicate a
decrease in the response variable due to the predictor variable associated with the estimate. We
back-transformed the estimates from the natural log scale and calculated percentage change
according to Eq 3:
Percentage change ¼ 100� ðeestimate  1Þ ð3Þ
We performed these analyses in R [27] with packages: “ggplot2” [28] for plotting,
“glmmTMB” [26,29] for model fitting, “bblme” [30] for AIC comparisons, “DHARMa” [31]
for model diagnostic tests, and “multcomp” [32] for post hoc testing.
3. Results
3.1. Effect of insecticide and NPK on weevil damage in EAHBs
In the nutrient omission trial, applying the insecticide chlorpyrifos and N affected weevil dam-
age in EAHBs. For any given level of weevil damage, the proportion of the plant population
affected was consistently less in plots sprayed with chlorpyrifos (sprayed but no fertilizer appli-
cation) than in non-sprayed plots (Fig 1, panel A). This reduction in weevil damage was
Fig 1. The cumulative distribution function for weevil damage in EAHBs with and without spraying chlorpyrifos (A)
and at different N application rates in sprayed treatments (B) in the nutrient omission trial.
https://doi.org/10.1371/journal.pone.0282493.g001
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 6 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
Table 4. Estimates, standard errors (SE), back-transformed estimates and per cent change in weevil damage as a function of insecticide and fertiliser application to
EAHBs in the nutrient omission trial using a GLMM with a negative binomial distribution, log link function and Laplace approximation (n = 1370).
Term Natural log scale Back-transformed % Change P value
estimate
Conditional model
Fixed effects Estimate ± SE
Intercept 1.4775 ± 0.14142 4.3819 0.000
Insecticide -0.8553 ± 0.0987 0.42512 - 57 0.000
N (kg ha-1 y-1) 0.0008 ± 0.0003 1.0008 0.08 0.003
P50 (kg ha-1 y-1) -0.1262 ± 0.1096 0.8815 0.250
K (kg ha-1 y-1) 0.0001 ± 0.0002 1.000 0.688
Other nutrients (kg ha-1 y-1) -0.0747 ± 0.1405 0.9280 0.595
Dispersion model
Intercept -1.4879 ± 0.1886 0.2258 0.000
Crop cycle 0.6225 ± 0.0870 1.8637 0.000
Random effects standard deviation
Mat: Plot: Block 0.3261
Plot: Block 0.0687
Block 0.1740
The reference category was zero for the categorical variables insecticide application, P, Other nutrients applied. Nitrogen and Potassium were treated as continuous
variables.
https://doi.org/10.1371/journal.pone.0282493.t004
strongly significant (p = 0.000). The sprayed plants had 57% less damage than plants that were
not sprayed (Table 4). The proportion of the plant population sprayed with chlorpyrifos and
affected by weevil damage was significantly higher among plants that received 400 kg N ha-1 y-
1; every kg increase in N application per ha per year was associated with a 0.08% increase in
weevil damage (Table 4). Applying K, P and “other nutrients” did not significantly affect weevil
damage.
3.2. Effect of Si, K and irrigation on weevil damage in EAHBs
In the potassium response trial, higher rates of Si and K were associated with lower weevil
damage among plants sprayed with chlorpyrifos (Fig 2). Applying 300 kg Si ha-1 y-1 was associ-
ated with a 45% decrease in weevil damage. Among plants that did not receive Si, the propor-
tion of the plant population affected by weevil damage was generally smaller among plants
treated with high K rates such as 250 and 600 kg ha-1 y-1 than those that received less K. This
difference in weevil damage was significant (p< 0.01). When compared to 0 kg K ha-1 y-1, 250
kg K ha-1 y-1 was associated with a 67% decrease in weevil damage and 600 kg K ha-1 y-1 was
associated with a 57% decrease in weevil damage (Table 5). These high rates (250 and 600 kg
ha-1 y-1) did not differ significantly from each other (p> 0.05). The effect of irrigation was not
significant (Table 5).
4. Discussion
The insecticide chlorpyrifos significantly reduced weevil damage in EAHBs as expected [20].
Chlorpyrifos is a contact insecticide that inhibits nervous-system messaging leading to a ner-
vous-system breakdown that kills the pest. It is, however, not 100% effective because weevils
spend a significant time of their life cycle protected inside the banana plant. In the nutrient
omission trial, pesticides alone reduced weevil damage by 57%. This study, therefore, com-
bined chemical control with fertiliser use.
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 7 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
Fig 2. The cumulative distribution function of weevil damage in EAHBs under different water and nutrient
treatments. All treatments were sprayed with chlorpyrifos.
https://doi.org/10.1371/journal.pone.0282493.g002
Our data show that weevil damage was reduced by 67% with a large rate of K in the potas-
sium response trial where K was combined with moderate rates of N. When high application
rates of K were combined with high rates of N–in the nutrient omission trial–the effect of K
was not significant. This suggests that the observed effect of K is counteracted by the availabil-
ity of N, which could explain why previous work [12] did not find a significant effect of NPK
on weevil damage in EAHBs when the same amount of K and N were applied. Ssali et al. [12]
applied a much lower rate of K (50 kg ha-1 y-1) compared with that applied in our experiments
Table 5. Estimates, standard errors (SE), back-transformed estimates and per cent change in weevil damage as a function of pesticide application combined with
irrigation or K or Si fertilizer in the potassium response trial analysed using a GLMM with a negative binomial distribution, log link function and Laplace approxi-
mation (n = 449). Pesticide and 100 kg N ha y-1 were blankets applied to all treatments shown here.
Natural log scale Back-transformed % Change P value
estimate
Fixed effects Estimate ± SE
Intercept 2.1928 ± 0.2775 8.9599 0.000
Crop cycle 2 -1.2333 ± 0.1712 0.2913 0.000
Crop cycle 3 0.2309 ± 0.1821 1.2598 0.205
Irrigated 0.0436 ± 0.1432 1.0445 0.761
Si 300 (kg ha-1 y-1) -0.6057± 0.1983 0.5457 - 45 0.002
K 75 (kg ha-1 y-1) -0.3795 ± 0.2366 0.6842 0.109
K 150 (kg ha-1 y-1) -0.4196 ± 0.2227 0.6573 0.059
K 250 (kg ha-1 y-1) -0.9609± 0.2921 0.3825 - 67 0.001
K 600 (kg ha-1 y-1) -0.8363± 0.2960 0.4333 - 57 0.005
Random effects standard deviation
Mat: Plot: Block (intercept) 0.64274
Plot: Block (intercept) 0.06756
Block (intercept) 0.00005
Dispersion parameter = 0.76.
The reference category is “Crop cycle 1” for Crop cycle, Rain fed for Irrigated and zero for Si and K application rates.
https://doi.org/10.1371/journal.pone.0282493.t005
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 8 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
(up to 600 kg K ha-1 y-1). Lower rates of K application did not significantly reduce weevil dam-
age in our experiment either. The effect of high rates of K on weevil damage in sites that have
low K (e.g. in the potassium response trial) is likely because K enhances the assimilation of car-
bohydrates into structural material, reducing excess sugars and free proteins in cells hence
making them less palatable to weevil larvae. K also facilitates the production of secondary
metabolites like phenolic compounds [33] which have been shown to deter weevil-larvae feed-
ing in the resistant dessert banana variety Yagambi-Km5. K deficiency is one of the main pro-
duction constraints in EAHB in Uganda [1].
In the potassium response trial, we found that plants fertilized with Si had less weevil dam-
age than plants without Si, concurring with findings for other plant-pest interactions [34]. A
stronger mechanical barrier [35] and induced resistance [10] may explain the role of Si
although Coskun et al. [36] argue that the apoplastic obstruction hypothesis is more likely. The
premise is that insects release effectors–insect proteins released into the plant to aid insect
attack–into the apoplast [37] where effectors manipulate plant defences [38] and the plant fails
to mobilize relevant defence [37,39]. For example, oral secretions of Colorado potato beetle
larvae contained bacteria that served as a microbial decoy. The decoy induced the salicylic acid
(SA) signalling pathway and, through cross-talk, suppressed Jasmonic acid (JA) mediated
defences, which enhanced larval growth [38]. Si, taken up as silicic acid (Si(OH)4) and present
in the apoplast, obstructs effectors from reaching their targets such that they do not compro-
mise plant defence [36].
In EAHB, Bakaze et al. [40] showed that when weevil larvae fed on resistant varieties, they
triggered greater production of phenolics and, greater deposition of lignin and suberin around
the damaged area. This response was lacking in the susceptible EAHB variety Mbwazirume
until it was artificially supplied with methyl Jasmonate. Following the logic of the apoplastic
obstruction hypothesis [36], pest effectors can successfully block the susceptible plants from
activating methyl Jasmonate pathways for defence but fail in the resistant variety. Applying Si
to susceptible EAHBs may obstruct pest effectors from their targets and allow otherwise sus-
ceptible EAHBs, to activate the methyl Jasmonate pathway for defence. To confirm this
hypothesis, more experiments are needed that explore the biochemical responses of EAHBs to
weevil damage under different fertilizer regimes.
Weevil damage generally increased with N, similar to N effects on other pests including
stem borers in rice [41]. These observations concur with the plant vigour hypothesis that sug-
gests that pests prefer to feed on vigorously growing plants [42]. We found that weevil damage
increased with N supply most likely because of the high concentration of soluble N-based com-
pounds and free amino acids associated with a high nitrogen supply. A higher concentration
of these compounds leads to more pest damage because they make the plant more nutritious
and easier to digest for the pest [8]. The bunch yields of EAHB in our experiment did not
respond to N applications [22], although drought and impaired uptake due to root constraints
may have played a role [17]. Regardless, large N applications were in excess which may have
affected the observed increase in weevil damage. The precise (optimal) N application beyond
which these negative effects start is still not known.
Though mineral fertiliser use in EAHB is still sparse, efforts to promote fertilisers are pick-
ing up along with efforts to intensify banana production. Caution should be taken not to apply
very high rates (e.g., 400 kg ha-1 y-1) of N as this will likely expose EAHBs to higher weevil
damage and risks leaking N into the environment and polluting. It is unclear what the optimal
ratio and application rates of N and K should be to maximise production and minimize weevil
damage. On the other hand, K fertilisers applied for yield gain will come with the added advan-
tage of reducing weevil damage if applied at high rates. For Si, however, its protective role is
documented in many studies and now also in EAHBs against weevils but its contribution to
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 9 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
yield is not known. Further studies should quantify whether silicon’s protective role translates
into yield gains that can cover the cost of Si fertiliser. Given that our results were based on
experiments at one site per research question, conducting similar experiments in other sites
would confirm whether our results have broader applicability. Filling these knowledge gaps
will move us closer to harnessing silicon’s protective role in EAHB.
5. Conclusions
We showed that combining K and Si fertiliser use with insecticide can contribute to weevil
damage control. Good nutritional management is therefore a key component of the integrated
management of weevils in EAHB which might reduce the need for insecticide application. Fur-
ther studies should investigate how far insecticide use can be reduced in EAHB given good
nutritional management.
Acknowledgments
We acknowledge Elkem ASA, Silicon Materials (Elkem) for supplying the product Elkem B
used as the source of silicon in the potassium response trial. We appreciate contributions from
Teddy Mbabazi Mutesi, Zaharah Najjuma and Florence Nakamanya for data collection and
trial supervision.
Author Contributions
Conceptualization: Hannington Bukomeko, Godfrey Taulya.
Data curation: Hannington Bukomeko, Godfrey Taulya.
Formal analysis: Hannington Bukomeko, Antonius G. T. Schut, Gerrie W. J. van de Ven.
Funding acquisition: Jerome Kubiriba, Ken Giller.
Investigation: Hannington Bukomeko.
Methodology: Hannington Bukomeko.
Project administration: Godfrey Taulya, Gerrie W. J. van de Ven, Jerome Kubiriba, Ken
Giller.
Resources: Godfrey Taulya, Jerome Kubiriba, Ken Giller.
Supervision: Godfrey Taulya, Antonius G. T. Schut, Gerrie W. J. van de Ven, Ken Giller.
Visualization: Hannington Bukomeko, Antonius G. T. Schut.
Writing – original draft: Hannington Bukomeko.
Writing – review & editing: Hannington Bukomeko, Godfrey Taulya, Antonius G. T. Schut,
Gerrie W. J. van de Ven, Ken Giller.
References
1. Wairegi LWII, van Asten PJAA, Tenywa MM, Bekunda MA, Asten PJA van, Tenywa MM, et al. Abiotic
constraints override biotic constraints in East African highland banana systems. Fld Crop Res. 2010;
117(1):146–53. Available from: http://dx.doi.org/10.1016/j.fcr.2010.02.010.
2. van Asten P, Gold CC, Wendt J, De Waele D, Okech S, Ssali H, et al. The contribution of soil quality to
banana yield problems and its relation with other banana yield loss factors. In: Proceedings of the work-
shop on farmer-participatory testing of IPM options for sustainable banana production in Eastern Africa.
Montpellier: CGSpace; 2005. p. 110–5. Available from: https://cgspace.cgiar.org/handle/10568/91841?
show=full.
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 10 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
3. Rukazambuga NDTM, Gold CS, Gowen SR. Yield loss in East African highland banana (Musa spp.,
AAA-EA group) caused by the banana weevil, Cosmopolites sordidus Germar. Crop Prot. 1998; 17
(7):581–9.
4. Gold CS, Pena JE, Karamura EB. Biology and integrated pest management for the banana weevil Cos-
mopolites sordidus (Germar) (Coleoptera: Curculionidae). Integr Pest Manag Rev. 2001; 6(2):79–155.
5. Smithson PCC, Mcintyre BDD, Gold CS, Ssali H, Kashaija INN. Nitrogen and potassium fertilizer vs.
nematode and weevil effects on yield and foliar nutrient status of banana in Uganda. Nutr Cycl Agroeco-
systems. 2001 [cited 2017 Dec 11]; 59(3):239–50. Available from: http://link.springer.com/10.1023/
A:1014462923539.
6. Kagoda F, Rubaihayo PRR, Tenywa MMM. The potential of cultural and chemical control practices for
enhancing productivity of banana ratoons. African Crop Sci J. 2005; 13(1):71–81. Available from: http://
www.ajol.info/viewarticle.php?jid=176&id=15737&layout=abstract.
7. Huberty AF, Denno RF. Plant water stress and its consequences for herbivorous insects: A new synthe-
sis. Ecology. 2004; 85(5):1383–98.
8. Rashid MM, Jahan M, Islam KS. Impact of nitrogen, phosphorus and potassium on brown planthopper
and tolerance of its host rice plants. Rice Sci. 2016; 23(3):119–31.
9. Bakhat HF, Bibi N, Zia Z, Abbas S, Hammad HM, Fahad S, et al. Silicon mitigates biotic stresses in crop
plants: A review. Crop Prot. 2018; 104:21–34.
10. Fawe A, Menzies JG, Chérif M, Bélanger RR. Silicon and disease resistance in dicotyledons. In: Datnoff
LE, Synder GH, Korndorfer GH, editors. Silicon in Agriculture [Internet]. Elsevier Science B.V; 2001. p.
159–69. Available from: https://www.sciencedirect.com/science/article/pii/S0928342001800136.
11. Amtmann A, Troufflard S, Armengaud P. The effect of potassium nutrition on pest and disease resis-
tance in plants. Physiol Plant [Internet]. 2008; 133(4):682–91. Available from: https://pubmed.ncbi.nlm.
nih.gov/18331404/. https://doi.org/10.1111/j.1399-3054.2008.01075.x PMID: 18331404
12. Ssali H, McIntyre BD, Gold CS, Kashaija IN, Kizito F, Ssali. Effects of mulch and mineral fertilizer on
crop, weevil and soil quality parameters in highland banana. Nutr Cycl Agroecosystems. 2003; 65
(2):141–50. Available from: https://www.scopus.com/inward/record.uri?eid=2-s2.0-0037281812&doi=
10.1023%2FA%3A1022184927506&partnerID=40&md5=2e4811d8a02a9f038669ffda6dc591de.
13. Rukazambuga NDTM, Gold CS, Gowen SR, Ragama P. The influence of crop management on banana
weevil, Cosmopolites sordidus (Coleoptera: Curculionidae) populations and yield of highland cooking
banana (cv. Atwalira) in Uganda. Bull Entomol Res. 2002; 92(5):413–21. Available from: https://www.
cambridge.org/core/product/identifier/S0007485302000482/type/journal_article. https://doi.org/10.
1079/BER2002182 PMID: 12241566
14. Mburu K, Oduor R, Mgutu A, Tripathi L. Silicon application enhances resistance to xanthomonas wilt
disease in banana. Plant Pathol. 2016; 65(5):807–18. Available from: http://doi.wiley.com/10.1111/ppa.
12468.
15. Fortunato AA, Rodrigues FÁ, Baroni JCP, Soares GCB, Rodriguez MAD, Pereira OL. Silicon Sup-
presses Fusarium Wilt Development in Banana Plants. J Phytopathol. 2012; 160(11–12):674–9.
16. Kablan L, Lagauche A, Delvaux B, Legr‘ve A. Silicon reduces black sigatoka development in banana.
Plant Dis. 2012; 96(2):273–8. Available from: http://apsjournals.apsnet.org/doi/10.1094/PDIS-04-11-
0274.
17. Taulya G. Kyosimba Onanya: Understanding productivity of East African highland banana. PhD Thesis.
Wageningen University; 2015.
18. Okech SH, Gold CS, Ssali H. Effects of potassium deficiency, drought and weevils on banana yield and
economic performance in Mbarara, Uganda. Uganda J Agric Sci. 2004; 9:511–9.
19. Nyombi K. Understanding growth of East Africa highland banana: experiments and simulation. PhD
Thesis. Wageningen University; 2010.
20. Corteva. Dursban insect control Insecticide. 2021 [cited 2021 Apr 3]. Available from: https://www.
corteva.in/products-and-solutions/crop-protection/dursban.html#:~:text=Dursban%20is%20based%
20on%20the,phosphorothioate%20group%20of%20organophosphorus%20pesticides.&text=No%
20reported%20incidences%20of%20resistance%20development%20against%20Dursban%20in%
20the%20target%20pests.
21. Gold CS, Speijer P.R., Karamura EB, Tushemereirwe WK, Kashaija I. Survey methodologies for
banana weevil and nematode damage assessment in Uganda. African Crop Sci J [Internet]. 1994; 2(3);
39–21. Available from: https://agris.fao.org/agris-search/search.do?recordID=UG19960123654.
22. Taulya G. East African highland bananas (Musa spp. AAA-EA) “worry” more about potassium defi-
ciency than drought stress. Fld Crop Res. 2013; 151:45–55. Available from: https://www.scopus.com/
inward/record.uri?eid=2-s2.0-84882646158&doi=10.1016%2Fj.fcr.2013.07.010&partnerID=40&md5=
39e89d9d91aaddc64234d76a8a33b5ae.
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 11 / 12
PLOS ONE Combining fertilisers and pesticides suppresses banana weevil damage
23. Kristensen K, McGillycuddy M. Covariance structures with glmmTMB [Internet]. CRAN. 2021 [cited
2022 Feb 24]. Available from: https://cran.r-project.org/web/packages/glmmTMB/vignettes/covstruct.
html.
24. Hilbe JM. Negative binomial regression. 2nd ed. California. Cambridge University Press. 2011.
25. González E, Seidl M, Ferrante M, Knapp M. Distribution of ecosystem services within oilseed rape fi
elds: Effects of field defects on pest and weed seed predation rates. Agric Ecosyst Environ. 2020;
295):1–9.
26. Brooks ME, Kristensen K, Benthem KJ Van, Magnusson A, Berg CW, Nielsen A, et al. GlmmTMB bal-
ances speed and flexibility among packages for zero-inflated generalized linear mixed modeling.The R
journal 2017; 9(2) 378–400. Available form https://journal.r-project.org/archive/2017/RJ-2017-066/
index.html.
27. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical
Computing. Version 1.0.23.1. https://www.r-project.org/. Vienna, Austria; 2021. Available from: https://
www.r-project.org/.
28. Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag. New York: Springer-Verlag;
2016. Available from: http://had.co.nz/ggplot2/book.
29. Magnusson A, Skaug HJ, Nielsen A, Berg C, Kristensen K, Maechler M, et al. Generalised Linear Mixed
Models using Template Model Builder. 2022. p. 1–45. Available from https://github.com/glmmTMB/
glmmTMB.
30. Bolker B, R Development Core Team. Tools for general maximum likelihood estimation [Internet].
CRAN, R package version 1.0.23.1. 2021. Available from: https://cran.r-project.org/package=bbmle.
31. Hartig F. Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. CRAN, Ver-
sion 0.4.5. 2022. p. 0–62.
32. Hothorn T, Bretz F, Westfall P, Heiberger RM, Schuetzenmeister A, Scheibe S. Simultaneous inference
in general parametric models. CRAN, version 1.4–18. CRAN; 2022. p. 1–35.
33. Armengaud P, Sulpice R, Miller AJ, Stitt M, Amtmann A, Gibon Y. Multilevel analysis of primary metabo-
lism provides new insights into the role of potassium nutrition for glycolysis and nitrogen assimilation in
Arabidopsis roots. Plant Physiol. 2009; 150(2):772–85. Available from: http://www.ncbi.nlm.nih.gov/
pubmed/19346439%5Cnhttp://www.ncbi.nlm.nih.gov/pubmed/19346439. https://doi.org/10.1104/pp.
108.133629 PMID: 19346439
34. Reynolds OL, Keeping MG, Meyer JH. Silicon-augmented resistance of plants to herbivorous insects: A
review. Ann Appl Biol. 2009; 155(2):171–86.
35. Kim SG, Kim KW, Park EW, Choi D. Silicon-induced cell wall fortification of rice leaves: a possible cellu-
lar mechanism of enhanced host resistance to blast. Genet Resist. 2002; 92(10):1095–103. Available
from: https://bsppjournals.onlinelibrary.wiley.com/doi/10.1111/ppa.12468
36. Coskun D, Deshmukh R, Sonah H, Menzies JG, Reynolds O, Ma JF, et al. The controversies of silicon’s
role in plant biology. New Phytol. 2019; 221(1):67–85. https://doi.org/10.1111/nph.15343 PMID:
30007071
37. Wang Y, Wang Y. Phytophthora sojae effectors orchestrate warfare with host immunity. Curr Opin
Microbiol. 2018; 46:7–13. Available from: https://doi.org/10.1016/j.mib.2018.01.008 PMID: 29454192
38. Wang J, Peiffer M, Hoover K, Rosa C, Zeng R, Felton GW. Helicoverpa zea gut-associated bacteria
indirectly induce defenses in tomato by triggering a salivary elicitor(s). New Phytol. 2017; 214(3):1294–
306. https://doi.org/10.1111/nph.14429 PMID: 28170113
39. Wu J, Baldwin IT. New insights into plant responses to the attack from insect herbivores. Annu Rev
Genet. 2010; 44:1–24. https://doi.org/10.1146/annurev-genet-102209-163500 PMID: 20649414
40. Bakaze E, Dzomeku BM, Wünsche JN. Banana defence responses to Cosmopolites sordidus feeding
and methyl jasmonate application. Ann Appl Biol. 2020; 178(1):98–108.
41. Zhong-xian L, Xiao-ping Y, Kong-luen H, Cui H. Effect of nitrogen fertilizer on herbivores and its stimula-
tion to major insect pests in rice. Rice Sci. 2007; 14(1):56–66.
42. Inbar M, Doostdar H, Mayer RT. Suitability of stressed and vigorous plants to various insect herbivores.
Oikos. 2001; 94(2):228–35.
PLOS ONE | https://doi.org/10.1371/journal.pone.0282493 March 10, 2023 12 / 12