Study site and study plant. Two experiments were conducted between August 2020 and March 2021, and between June 2022 and October 2022 at the International Centre of Insect Physiology and Ecology (icipe), Nairobi, Kenya (latitude 1°13’S and longitude 36°53’E and a mean elevation of 1,587 m above sea level).
SC Duma 43 was the maize (Zea mays L.) variety selected for the experiments and its seeds were sourced from Kenya Seed Company (Nairobi, Kenya). It constitutes mostly grown by smallholder farmers in many regions of Kenya. Seeds were sowed in 25.5 cm high by 30 cm diameter plastic pots (one seed per pot), each containing a blend growing substrate at a ratio of 2:1:1 of topsoil, compost, and sand soil. One week following germination, each plant was top-dressed with 2–3 grams of NPK fertiliser (Yara East Africa Limited, Nairobi, Kenya) comprising nitrogen, phosphorus, and potassium in a ratio of 17:17:17. The potted plants were maintained in an open field under sufficient natural light (12 L:12 D photoperiod), at a mean daily temperature range of 23–27 ℃. All good agronomic practices, including watering, topdressing and weeding, were applied. Artificial infestation of these crops was conducted following a procedure adapted from Harrison47 (1986) and was applied when the maize plants were 3–4 weeks old. The artificial infestation was subsequently followed by application treatments.
Fall armyworm colony. Fall armyworm eggs and larvae were obtained from a continuous colony reared at icipe. The rearing of fall armyworm colony at icipe is described by Tefera et al.48. Fall armyworm egg masses were deposited on wax/butter paper. The rearing was conducted under laboratory conditions of 25 ± 2°C, 72 ± 3% RH, and L12:D12 photoperiod, with the larvae feeding on maize leaves.
Preparation of treatments. Freshly harvested and sun-dried small pelagic fish, Rastrineobola argentea (locally known as “omena” in Luo, “dagaa” in Swahili, and “mukene” in Luganda), was used. Rastrineobola argentea fish was obtained from the Gikomba market of Nairobi, Kenya. Two kilograms of grounded fish were boiled in 5 L of water for 45 minutes. The soup was separated from the boiled fish by sieving the mixture through 4.5-mm mesh, and the resultant soup was left to cool to room temperature in a 5 L plastic bucket. Approximately 450 g of white sugar (Kabras Sugar Mills Ltd., Kakamega, Kenya) was stirred into the fish soup to homogenise the solution. The solution was decanted, and the liquid was poured into a 5 L bucket.
For the first experiment, a series of dilutions of 50% and 10% were prepared from the initial concentration (100%) with the addition of distilled water. Three treatments (100%, 50%, and 10% fish soup + sugar) were prepared for immediate subsequent spray, while distilled water was used as a control.
Four solutions (fish soup, sugar, fish soup + sugar, and fish residue) were prepared for use in the second experiment. The fish soup was prepared as described above, and the resultant solid residue of fish was placed separately in clean plastic cups. Three solutions, at 25% fish soup, sugar, and fish soup + sugar (90g/L), were separately prepared. As a positive control, the chemical insecticide, Habel™ 5 (Emamectin benzoate), was mixed with water at the recommended concentration of 0.25 g/L. The negative control consisted of distilled water for this experiment.
Experimental design and treatments application. The experimental layout followed a randomised complete block design. In the first experiment, four treatments (10% fish soup + sugar, 50% fish soup + sugar, 100% fish soup + sugar, and distilled water) were sprayed on the maize plants. Five potted plants were arranged in 4 rows in a plot, at distances of 1 m between plants and 1.5 m between rows (Fig. 10). The same design was replicated three times in three plots situated 5 m apart. Before applying treatments, all plants were artificially infested with egg masses of fall armyworm. The sections of wax/butter paper containing the eggs were removed and clipped onto the underside of a maize plant leaf, near the whorl, at the 3–4 leaf maize stage. Each maize plant received 30–50 eggs, and after 4 days, when most of the eggs had hatched, plants were sprayed with four treatments, as follows. All treatments were separately loaded in a 1.5 L hand sprayer, and applied onto maize plants in a block designated for each treatment by spraying until dripping. Data collection began immediately after the application of the treatments.
In the second experiment, five treatments were compared, which comprised a negative control (distilled water), fish soup, sugar solution, a blend of fish soup + sugar, and a positive control (the insecticide, 5-Emamectin benzoate (Habel™)). The experiment was laid out in four plots with 5 potted plants per row in a plot. The plots were replicated three times with random allocation of treatments. Artificial infestation of FAW larvae was carried out as described for the first experiment, when maize plants reached the 3–4 leaf stage, and repeated 5 weeks later (presented here as week 0 and week 5, respectively). Following the initial application, treatments were repeated weekly for seven weeks.
Assessment of foliar damage and plant recovery. The foliar damage caused by FAW larvae feeding on crop leaves exposed to various treatments was assessed. In the first experiment, leaf (foliar) damage was evaluated through using an arbitrary scale of 0–100%, adapted from Williams et al.49. Monitoring was done daily, and the foliar damage and plant recovery observations were made once weekly for two weeks. The reduction of foliar damage after two weeks was expressed as plant recovery. Recovery was calculated by the difference between the damage scores of weeks 1 and 2, using the following formulae:
\(Foliar recovery \left(\%\right)=b-\) a
where a is the foliar damage (%) score on week one, and b is the foliar damage (%) score on week two. When the calculation of recovery yielded negative values, a score of zero was recorded, which meant that damage was continuous and did not allow plant recovery.
In the second experiment, foliar damage, but not foliar recovery, was scored for seven weeks, using the protocol of Williams et al.49.
Assessment of abundance and diversity of visiting insects. In the first experiment, visiting insects were collected through using two trapping methods: pitfall traps and yellow sticky cards. Immediately after spraying potted maize plants, a pitfall trap (consisting of a 150 mL cylindrical cup, 74 mm in height with a 70 mm opening diameter) was placed on a pot, adjacent to the maize stem, with the cup opening at soil level to capture potential crawling insects visiting the treated maize plants. The pitfall traps were filled with water containing a few drops of unscented multipurpose liquid detergent (Teepol®) to trap crawling animals by drowning. Traps were collected after 24 hours for the morphological identification of the insects. The pitfall traps were placed after applying treatments, one week after treatment, and two weeks after treatment.
To capture potential flying insects visiting the plants, 10 cm × 25 cm yellow sticky cards (Horiver®, Koppert Biological Systems, Nairobi, Kenya) were suspended 30 cm above the crop canopy along each block of maize plants. These polystyrene sticky cards were covered with non-toxic glue on both sides. They were set after the application of treatments once weekly for two weeks. The sticky cards were removed after 24 hours, and the tanglefoot glue was dissolved by using kerosene to suspend the trapped insects (adapted from Muvea et al. 50). The trapped visiting insects were labelled and arranged per treatment and organised per taxonomic group, i.e. order and family. Insects were then collected, using a fine brush, and transferred into glass vials containing 70% ethanol for further processing, counting and identification using a dichotomous key up to family level.
Assessment of plant growth parameters. The plant growth parameters (plant height and chlorophyll content) were recorded in blocks sprayed with 50% fish soup + sugar solution and distilled water in the first experiment. Measurements of plant height (in cm) and chlorophyll content (SPAD value) were taken weekly for three weeks. Above-ground plant height was measured using a 1-metre tape measure. A SPAD 502 Plus Chlorophyll Meter (Honor Test Technology Co., Ltd., Shijiazhuang, China) was used to measure the amount of chlorophyll in the leaves. Weekly measurements of the chlorophyll amounts were done from the 3–4 leaves stage until V14 stage of the maize crop by using a chlorophyll meter (SPAD). The uppermost fully expanded leaf was selected for the first reading, and the second and third readings were of the second and third leaves under the uppermost expanded leaf in a plant. The reading for each leaf was taken at between 40 and 70% distance from the base, and readings from the three leaves were averaged.
Proximate analysis and mineral composition of fish soup. The proximate analyses and mineral compositions of the three samples (fish soup, fish soup + sugar, and fish residue) were submitted to Crop Nutrition Laboratory Services Limited (Cropnuts Ltd), Limuru, Kenya. The proximate compositions were estimated according to the methods of the Association of Official Analytical Chemists (AOAC) 51. The crude protein (N × 6.25) was determined using the Kjeldahl method (method 978.04) 52. Crude ash, crude fat, crude fibre, and dry matter were determined through using the following methods, ISO 5984, ISO 6492, ISO 13906, and ISO 6496, respectively. Mineral composition analysis was carried out using the wet chemistry Inductive Coupled Plasma – Mass Spectrometry technique (Wetchem, ICP-MS).
Volatile collections and analysis. Volatile compounds present in the fish soup samples and control groups were collected by capturing the headspace volatiles. Each treatment was cooled to room temperature and then transferred to separate 2-L Quickfit® glass chambers (Analytical Research Systems, Gainesville, FL, USA). To facilitate the collection process, activated charcoal-filtered and humidified air was circulated over the samples at a flow rate of 340 mL/min, using a push–pull Gast pump (Gast Manufacturing, Benton Harbor, MI, USA). The volatiles were subsequently absorbed onto Super-Q traps (30 mg, Analytical Research Systems, Gainesville, FL, USA) at a flow rate of 170 mL/min, employing a Vacuubrand CVC2 vacuum pump (Vacuubrand, Wertheim, Germany). All volatile collections were carried out for a duration of 24 hours.
Prior to collection, volatile collection traps, Super-Q® traps (SQ International, Seoul, Korea) were pre-cleaned, using GC-grade dichloromethane, and dried with a stream of high purity nitrogen gas provided by a nitrogen generator (Peak Scientific Instruments Ltd., model 600 cc, Renfrewshire, Scotland). At the end of the 24-hour collection period, the Super-Q® traps containing adsorbed volatiles were eluted with 200 µL of dichloromethane into 2-mL clear glass vials. Each vial was equipped with a 250 µL conical point glass insert (Supelco, Bellefonte, PA, USA). The eluted samples were promptly subjected to analysis through using gas chromatography–mass spectrometry (GC-MS).
For the analyses, the extracted fish soup volatiles were injected, using a splitless technique (1 µL), into a 7890A gas chromatograph (GC), coupled with a 5975C mass selective detector (MSD, Agilent Technologies, Santa Clara, CA, USA). The GC system was equipped with a 5%-phenyl-methylpolysiloxane (HP5 MS) low-bleed capillary column (30 m × 0.25 mm i.d., 0.25 µm; J&W, Folsom, CA, USA). The oven was programmed with the following settings: helium flow rate at 1.25 mL/min, initial oven temperature held at 35°C for 5 minutes, followed by a rise at a rate of 10°C/min to 280°C, and then held at this temperature for 20.4 minutes. The MSD was operated with an ion source temperature of 230°C and a quadrupole temperature of 180°C. Electron impact (EI) mass spectra were obtained at 70 eV, and the fragment ions were analysed over a mass range of 40 to 550 m/z in full scan mode. A solvent delay of 3.3 min was implemented. Experiment-specific retention indices (RIs) were calculated relative to C8–C32 n-alkanes.
The relative integration of each detected peak was determined by using the ChemStation integrator and reported as the relative abundance. To eliminate contaminant peaks or peak areas originating from the adsorbent, column, or solvent, blank runs were performed on empty collection systems and analysed. Detected peaks were tentatively identified by comparing the mass spectral data with reference spectra published in library MS databases, considering retention times and retention indices and, where available, identification was made through co-injection with an authentic sample.
Data analyses. The data collected were summarised by descriptive statistics (counts and percentages). The extent (percentage) of foliar damage, percentage of plant recovery, the numbers of insect species collected on traps, plant heights and chlorophyll contents were recorded. The insects collected on the pitfall traps and sticky cards were counted based on treatments, and identified according to order, family, and genus level.
Subsequently, the means of foliar damage and percent plant recovery were subjected to analysis of variance (ANOVA).
To establish the percentage peak recovery and least damage of maize plants after treatments with 10%, 50% and 100% fish soup + sugar solutions, the data were fitted to a modified Gompertz model 53:
$$Y\left(D\right)={Y}_{Asym}exp \left(-exp \left(\left(\frac{{u}_{{g}^{e}}}{{Y}_{Asym}}\right)({\lambda }_{g}-C\right)+1 \right)$$
where Y(D) is the expected level (percentage) of recovery or damage of maize plants as a function of the concentration of fish soup + sugar solution, YAsym is the asymptotic recovery or damage level (percentage), λg is the inflection point of the curve (having concentration units), ug is the rate of recovery or damage, and C denotes the specific concentration of fish soup + sugar solution tested. To get weighted least–square estimates of these parameters, the data were fitted in the Gompertz model’s equation using the nlsLM function, and start values for the model to achieve convergence tolerance were based on hypothetical estimations. The corresponding least concentration (Copt) required for the peak recovery or lowest damage of maize plant (expressed as YAsym) was calculated from a mathematical equation where C was the subject of the formula.
Shannon-Weiner diversity was used to estimate the diversity of insects that visited the artificially FAW-infested maize after spraying with three fish soup + sugar solutions, alongside the control. The following parameters were assessed. The relative abundance of order was determined as:
$$Relative abundance=\frac{n}{N}$$
where n is the total number of specimens of a particular insect family, and N is the total number of all insect families in a particular order. Insect family richness was estimated for each treatment 54. To assess the insect diversity, the Shannon-Weiner index (H′) 55 was computed using the Shannon and Weaver 56 formula:
$${H}^{{\prime }}=-⌊\sum Pi*LN\left(Pi\right)⌋$$
where H′ is the Diversity Index, Pi is the proportion of each family in the sample, and LN (Pi) is the natural logarithm of Pi. The evenness of insect families compares the similarity of the population size of each family 57.
$$Evenness Index \left({J}^{{\prime }}\right)=\frac{{H}^{{\prime }}}{Hmax}$$
where Hmax is the natural log of the total number of families.
A word cloud analysis was conducted to show the abundance of different families of insects attracted by the fish soup + sugar treated maize plants.
A probit regression was used to predict the count of different families across the different concentrations of fish soup. The predicted count was plotted against the fish soup concentration.
Datasets on plant growth parameters (plant height and chlorophyll content) were subjected to a generalised linear model (GLM). Post-hoc analyses were performed for factors showing significant differences by using Tukey’s honestly significant difference (HSD) test, at p < 0.05. Word cloud analysis was conducted to illustrate the families of insects attracted to fish soup + sugar solution sprayed on maize plants. All statistical analyses were conducted with R Software version 4.2.3 58.