Maize Yield Response to Chemical Control of Spodoptera frugiperda at Different Plant Growth Stages in South Africa †

: Fall armyworm (FAW), Spodoptera frugiperda (Lepidoptera: Noctuidae), is a major pest of maize. Yield losses between 30 and 70% in the Americas and between 11 and 100% in Africa have been reported. Little information exists on the effect of pest damage during different plant growth stages on yield loss. Previous studies showed that insecticide applications at weekly intervals did not always provide a yield gain comparable to only a single or two well-timed applications. In this study, we completed four field trials under high natural pest pressure. Treatments consisted of different regimes of insecticide applications that provided protection against damage during different growth stages. In one trial, the mean incidence of infested plants was 65%, and the yield benefit gained from four insecticide applications was 32.6%. The other three trials had 16 treatments which were divided into two spray sequences to protect plants against FAW damage for different lengths of time, between early vegetative stages and tasseling. Yield losses were 41.9, 26.5 and 56.8% for the three respective trials if no insecticides were applied. Yield loss of plants protected during earlier growth stages was significantly lower than that of the treatments which provided protection during later growth stages. More than three spray applications generally completed did not provide further yield gains. Plants that were protected more during early growth stages yield higher than plants protected during later growths stages.


Introduction
Fall armyworm (FAW), Spodoptera frugiperda (Lepidoptera: Noctuidae), infests maize during all plant growth stages, with infestation during vegetative stages leading to serious leaf-feeding injury, while late infestations lead to ear damage [1][2][3][4][5]. This damage results in both quantitative and qualitative losses [6][7][8]. FAW damage to maize during the late growth stages, just prior to tassel emergence, may result in ear damage and comparatively higher yield losses [1,3,[9][10][11]. It was also reported that very early infestations, especially during the seedling stages, can lead to total defoliation and crop loss [3,12]. Yield losses suffered by maize are therefore mostly dependent on the plant tissue type where larval damage occurs [3], with infestations during the vegetative growth stages leading to extensive foliar damage.
Insecticide applications have been the main method of FAW control in South America [13,14] and in Africa since its arrival in 2017 [8,[15][16][17]. Insecticide applications against FAW in maize are often reported to not be as effective due to incorrect application methods, cryptic feeding behavior of larvae and application of insecticides when larvae are too large and not susceptible anymore [4,8,18]. Chemical control strategies are only effective when larvae are small and require timely or regular applications [4,19]. However, the efficacy of insecticides applied at high rates against FAW has been reported to be rapid and efficient, compared to other control measures, if applications are completed timeously and correctly [13,14,16]. A short generation period, which can be completed within 20 to 30 days, may result in reinfestation of the same crop by FAW, resulting in a need for repeated insecticide applications [13,14,16], which may lead to resistance evolution against insecticides [18,20].
It is important to estimate the effect of insecticide applications at different growth stages under African conditions since the protection of plants for specific periods may have significant yield benefits [3,21,22]. It is also important to identify the duration of the period of protection that would result in significant yield gains [3,21,22], based on the infestation levels observed in maize fields [23,24].
For chemical control to be economical, it must be applied according to economic threshold levels [24,25]. However, most literature from North and South America indicate no to poor correlations between the level of FAW infestation, foliar damage and actual yield loss. No data on damage-yield loss relationships under African farming conditions are available. Several authors from the Americas reported losses between 17% and 45% [26,27], with occasional losses as high as 100% [12].
Scientific publications provide contradicting results on whether chemical control of FAW is effective in reducing yield loss. Del Rosario et al. [28] and Del Rosario and Dilco [29] showed that chemical control was economically justifiable with significantly higher yields. Studies conducted in the southern parts of America and South America [12,30,31] showed that chemical control might be justified, but reports also indicate that control may only provide small yield gains [28,29].
Economic threshold levels for FAW control have not been determined [25], and only guidelines exist for use in the decision-making process [32]. Large variation exists in the injury resulting from a given level of FAW infestation and plant response to injury [6,25,27,33]. Improved knowledge on plant response to FAW damage can contribute to estimating ETLs for this pest and will contribute to decision support systems for FAW management.
The aim of this study was to determine the effect of insecticide applications at different plant growth stages between seedling emergence and flowering on yield losses caused by FAW in maize.

Materials and Methods
Four field trials were conducted in the Lowveld region of South Africa. Three of these trials (Trials 1, 2 and 4) were conducted on adjacent fields at an experimental farm in Malelane (−25.564559, 31.657301), and trial 3 at an experimental farm in Nelspruit (−25.468072, 31.059030), South Africa. Sub-tropical climatic conditions allowed for natural FAW infestation throughout the season at both sites. The trials were conducted at these two different sites to provide for different ecoregions and to account for possible differences in results due to the effect that higher altitude and lower temperatures could have on pest ecology and plant response to damage. Malelane is situated in a hot and humid Lowveld region of South Africa, 320 m a.s.l., while the altitude at Nelspruit is 720 m a.s.l. The daily mean temperature at Malelane is 24-26 °C, and the mean daily minimum and maximum temperatures range between 12 and 18 °C, and between 28 and 35 °C, respectively. The daily mean temperature at Nelspruit is 20-22 °C, and the mean daily minimum and maximum temperatures range between 12 and 16 °C, and 28 and 30 °C, respectively [34]. The climatic conditions at the Malelane site during January and February are similar, and no effect on plant growth or insect ecology is expected. Trials 1 and 2 were planted during February 2019, with a 7-day interval between the two plantings. The experimental design was a randomized block with 16 treatments, replicated four times for each trial. Each treatment plot consisted of two adjacent rows that were 5 m in length. The inter-row spacing was 1.0 m. Each plot contained approximately 40 plants. Trial 3 was planted on 20 February 2019. The 16 treatments were replicated five times, and each replicate consisted of a single row, 10 m in length. Each plot contained approximately 40 plants, and the inter-row spacing between plots was 1.2 m. Trial 4 was planted in January 2019. Plots consisted of two adjacent 5 m rows at a 1.0 m inter-row spacing. There were two guard rows of maize at the sides of each trial.
A herbicide-tolerant maize hybrid, DKC78-35R, was used in all the trials. Plants of this hybrid take between 64 and 83 days to tassel and between 120 and 148 days to become physiologically mature [35].

Insecticide Spray Schedule
The treatments used in trials 1, 2 and 3 are indicated in Table 1. The first insecticide application was completed 1 week after seedling emergence (WAE), while the final application was completed 8 WAE, when all plants were starting to tassel. Due to the higher altitude and slightly lower temperatures at the site where trial 3 was conducted, the plant developmental period was prolonged by approximately 7 days compared to those of trials 1 and 2. This implied that the final insecticide application was completed when plants were in the V14 growth stage, with only a few tassels starting to emerge ( Table 2).
Treatments were divided into two spray sequences (Sequences I and II) ( Table 1). The purpose was to protect plants against FAW damage for different periods of time, between the early vegetative stages and tassel emergence. The upper check treatment received an insecticide application at weekly intervals from 1 WAE to early tassel emergence. The control treatment did not receive any insecticide application. Plants that received insecticide applications during spray Sequence I therefore received protection mainly during later development stages, while those of Sequence II treatments were protected largely during early growth stages.
Methomyl 900 SP was applied by means of a knap-sac sprayer. A hollow-cone nozzle was used, and the application rate was 250 g/ha at a spray volume of approximately 200 L spray solution per hectare. Applications were directed into the whorls of plants, and care was taken to ensure full plant coverage. Supplementary irrigation was provided as needed at the Malelane site. Table 1. Insecticide application schedule for control of fall armyworm on maize plants, from the seedling stage to tasselling, 8 weeks after seedling emergence in trials 1, 2 and 3. This spray schedule provided different periods of protection over different maize growing stages. The crosses indicate the time that insecticide applications were completed. WAE = weeks after seeding emergence. The onset of the tassel stage in trials 1 and 2 was at 8 WAE, and 9 WAE in trial 3.

Crop Age (Weeks)
x x  x  x  4  II  14  x  x  x  3  II  15  x  x  2  II  16 x 1 II Table 2. Crop age (weeks after seedling emergence), at which insecticide applications were completed for control of Spodoptera frugiperda in maize under natural infestation (trials 1, 2 and 3). Trial 4 consisted of a block of maize which was divided into two sub-blocks, each with 24 treatment plots. Each treatment plot consisted of two adjacent 5 m rows, each with approximately 20 plants. The two sub-blocks were separated by a 1.5 m barren area where maize plants were removed. Plant stand per treatment plot, as well as the number of plants exhibiting whorl damage, were recorded 4 WAE (V6/7). The one sub-block (24 treatment plots) was sprayed with methomyl SP immediately after the incidence of infested plants was determined. This was followed by three additional sprays at weekly intervals until plant tassels were visible in the whorls. The unsprayed sub-block remained unprotected till harvest to allow for continuous larval feeding damage and reinfestation throughout the trial. Two rows on each side of the block served as side rows, and the eight central rows were used for this experiment. Insecticide applications were completed as described above for trials 1, 2 and 3. Plants were left until harvest to determine yield. Ears were harvested on a per plot basis and hand threshed. Yield loss was expressed as a percentage relative to the yield of the upper check treatment as follows:
Yield of trial 4 was determined for each of the sprayed and unsprayed treatment plots, after which it was calculated and expressed as yield per plot and hectare. Moisture analysis was conducted on a sample of the maize kernels of each plot. Grain mass was calculated at a moisture content of 11%.

Data Analysis
The relationship between incidence of plants exhibiting feeding damage and plot yield was analyzed by means of regression analyses, and scatterplots were used to illustrate relationships. The yield per plot and percentage yield loss of each treatment were compared by means of one-way ANOVAs conducted separately for each trial, followed by Tukey's HSD test at p = 0.05. Data were tested for normality (Shapiro-Wilk test) and homogeneity of variance (Levene's test), followed by ANOVAs.

Results
Although the infestation level was not quantified for trials 1 and 2, the incidence of damaged plants in the untreated control plots, based on what was quantified in the adjacent maize field described in trial 4, was high (>40%).

Yield Data
The tendencies observed in trials 1, 2 and 3 were similar. There was a large variation in plot yield between treatments. The highest yields were obtained from the fully protected treatments, which received eight insecticide applications ( Table 3). The lowest yields were obtained from plots that received none or only a single insecticide application, at either 1 WAE (treatment 16) or 8 WAE (treatment 9), or two and three insecticide applications applied either at very early or very late plant growth stages (treatment numbers 7, 8, 14, 15).
There was a general tendency that yields of Sequence I treatment plots that received 5, 6 or 7 insecticide applications were statistically similar to that of the fully protected control. With the exception of treatment 16, which received an insecticide application only at the early tasselling stage, all treatment yields of Sequence II were statistically similar to that of the fully protected control. The mean yield of treatments receiving Sequence I sprays was lower than those of plants that were protected largely during the early growth stages (Sequence II sprays) ( Table 3).

Yield Loss Data
The yield losses suffered by the unsprayed control treatments, in relation to the fully protected control treatment, which received 8 insecticide applications, were 41.9%, 29.0% and 51.9% for trials 1, 2 and 3, respectively (Table 3).
Yield loss (%) differed significantly between treatments in each of the trials ( Table 3). The lowest yield losses in all trials were observed in treatments in which plants were protected from 2 to 8 WAE, although these losses did not differ significantly from those of many other treatments. Marked differences were, however, observed between the mean yield loss of treatments subjected to Sequence I and Sequence II treatments. The mean yield loss percentage of Sequence I treatments, which protected plants largely during late plant growth stages, were 28.3, 21.7 and 28.0% for trials 1, 2 and 3, respectively, while yield loss of Sequence II treatments were 16.8, 18.6 and 10.4%, respectively. Figures 1 and 2 describe the effect of the different numbers of insecticide applications and periods of protection (Table 1; Sequence I and II sprays) on plot yield. The green line represents the mean plot yield of the treatments that received weekly insecticide applications from 1 to 8 WAE, while the red line represents the mean plot yield of the unsprayed plots. For example, a single spay application, which was completed either at 1 WAE or 8 WAE, resulted in low yields. The latter are indicated on the figures as "early protect: 1 WAE" and "late protect: 8 WAE", respectively.
A single application on very young plants (1 WAE; treatment 16), as well as a single application when tassels started to develop (8 WAE; treatment 9), did not provide any yield benefit compared to the unsprayed control treatment. Similarly, protection of plants from 5-8 WAE only did not provide any yield benefit. Two repeated applications (1 and 2 WAE) resulted in a yield similar to that obtained with two very late applications (7 and 8 WAE). Three insecticide applications that provided protection during early vegetative stages (1-3 WAE; Sequence II) yielded the highest (Figure 1).
In all three trials, additional insecticide applications at 4, 5 and 6 WAE (treatment 10), following three applications during the early whorl period (1, 2 and 3 WAE) (treatment 14) did not provide any significant contribution in terms of limiting yield loss compared to two or three early insecticide treatments (Figure 2).     In trial 4, the incidence of infested plants, 4 WAE, ranged between 33 and 93% per plot, and the mean percentage infested plants for the whole block was 65%. A scatterplot indicating the wide range in percentage infested plants between treatment plots at the time of the first insecticide application (4 WAE) is provided in Figure 3. All the plots that were protected against further FAW damage (4-7 WAE) yielded higher than unsprayed treatment plots.  Regression lines indicating the yield of sprayed (red) and unsprayed (blue) plots (Figure 4) showed that large yield gains resulted from the protection of plants from 4 WAE onwards. There was no relationship (r = 0.13; F = 0.325; p = 0.574) between the incidence of damaged plants and yield in unprotected plots. There was, however, a slight but significant relationship between the incidence of infested plants at 4 WAE and yield in sprayed plots (r = 0.51; F = 7.613; p = 0.011). These regressions do, however, need to be interpreted with care, since there were no uninfested plots that could have provided data on attainable yield.

Discussion
Overton et al. [25] found large differences in the mean yield losses caused by FAW damage, and cautioned that when crop losses are reported through farmer surveys, they may be overestimated. A similar observation was made by Baudron et al. [32], who reported that the FAW infestation levels on smallholder farmers' fields in Zimbabwe ranged between 32 and 48%, and estimated yield loss to be 11.6%, which is lower than that reported in many other studies conducted in the Americas. The impact of FAW on maize yield in Africa has been reported as very large. Day et al. [3] estimated yield losses between 22 and 67% in Ghana and Zambia, and Kumela et al. [37] estimated 32% loss in Ethiopia and 47% in Kenya. The losses reported in this study were, however, in the same range as those reported in farmer surveys conducted in Africa.
Several studies conducted on the relationships between FAW infestation and crop response indicated that maize plants might tolerate damage. The level of tolerance is, however, dependent on the particular environment and pest ecology. Maize plants have been reported to recover from injury and to be tolerant to damage in agroecological zones, where the climate does not allow for rapid pest development and continuous reinfestation of the crop [2,3,38]. Therefore, in regions with favorable environmental conditions which allow for continuous FAW infestations, it is recommended that the first application of insecticides should be made no earlier than 2-3 weeks after seedling emergence [6,39].
The response of maize plants to FAW damage is influenced by several biotic and abiotic factors. These include plant growth stage and plant health, the incidence of infested plants, the severity of foliar damage symptoms, duration of larval feeding and whether or not reinfestation of the crops takes place [1][2][3]. Environmental and climatic conditions within a given geographical region are also important factors that determine the necessity of insecticide application for the control of FAW larvae [3,40,41].
Results from trials 1-3 showed, as expected, that yield was higher in plots that received increased numbers of insecticide applications. However, higher numbers of insecticide applications were not always correlated with a higher yield, and high yield losses still occurred, similar to what was reported by Dal Pogetto et al. [42]. The time of insecticide application based on the growth stage of maize plants and the infestation level of FAW is more important than the number of insecticide applications [42]. Susceptibility of maize plants varies greatly between the different growth stages [3], and it has been reported that plants are most susceptible to damage during seedling emergence and the early whorl stages, and then again during the period of pre-tassel formation [1,11]. The yield of plants that were protected against FAW damage for extended periods during late plant growth stages up to tassel emergence was generally lower than that of plants that were largely protected during the early growth stages. These results indicated that the protection of maize plants against FAW damage during the early vegetative growth stages provided the largest yield gain. Additional insecticide applications after 4 WAE (for example, treatment 14) did not provide any significant yield gain if plants were also protected during the very early growth stages.
Early detection of FAW infestation is therefore important to facilitate effective control. Small FAW larvae (first-third instar) are relatively easy to control and inflict low levels of injury [4,32]. Therefore, once these small larvae are detected within a maize field, and their population numbers or degree of injury exceeds threshold levels, chemical control needs to be implemented [2,32].
Insecticide application at very early growth stages, such as the first week after seedling emergence, may not be necessary or economically justifiable [38] since pest biology also influences the relationship between infestation and plant response to injury. During the seedling stage, only egg batches or small-instar larvae (<7 days old) at very low densities may be present. Such low population densities, which may lie below a damage threshold level that causes injury [43], may have no negative impact on the overall quantity or quality of the crops and their yield.
In terms of practical crop protection, insecticide applications at such early stages may be postponed to allow for an application 5 or 6 days later since there may still be eggs present on plants, and larvae that hatched from the first egg batches laid on plants will mostly still be in the first three instars, which can be easily controlled. Favorable microclimates enable overlapping generations, which complicate chemical pest control [37]. Therefore, in regions with favorable environmental conditions which allow for continuous FAW infestations, it is recommended that the first application of insecticides should be made no earlier than 2-3 weeks after seedling emergence [6,39]. However, the duration and intervals of application are highly dependent on the climatic factors within each specific region [7,24].
The trial conducted on the highly infested maize field of which the one sub-block was protected from 4 WAE onward (Trial 4) is typical of what farmers experience in tropical and sub-tropical areas where the continuous infestation of FAW may occur on a single crop. This study showed that the incidence of infested plants (%) during mid-vegetative growth stages, such as the scenario in this study, was not a good predictor of eventual yield loss. The yield gains obtained in the protected sub-plot, even when insecticide applications were completed only when the incidence of infestation was 50% or higher, indicated that chemical control might provide a benefit in terms of yield, but it will be essential to take the economics of control into account. The smallest yield gains in trial 4 were obtained with insecticide applications on plots that had higher than 70% infested plants ( Figure 4). A decrease in yield gain (the difference between the upper 95% confidence interval line (black) and the lower 95% confidence interval line (red) was observed with an increased incidence of infestation ( Figure 4). This indicated that insecticide applications in cases where infestation pressure approaches very high levels might not be economically warranted. These results are, however, only from a single cropping season, and careful interpretation of data together with further experimentation is required.
The mean yield benefit gained from the four insecticide applications in trial 4 was 32.6% (754 kg/ha). If the two treatment plots with the lowest incidences of infested plants (33 and 38%) are excluded from Figure 3, all other infestation levels were above 48%. This means that near-complete protection of a crop from the stage that an infestation level of 48% or higher was observed during the mid-vegetative stage resulted in a yield benefit of 0.754 t/ha after four insecticide applications. In a South African scenario, at a maize grain price of 270 USD per metric ton, the yield gain would equate to 203 USD per hectare. At an average cost of, for example, 25 USD/ha for a single insecticide application by means of a tractor-mounted sprayer, the input costs of applying the four sprays to offset the 32.6% yield gain would be approximately 100 USD/hectare. The yield gain ascribed to increased numbers of insecticide applications is, however, in many cases, not economically viable [42].
Large variation exists in the injury resulting from a given level of FAW infestation and in plant response to injury [6,25,27,33]. On-farm studies in Zimbabwe suggested that yield loss cannot be predicted from assessments of infestation and leaf damage alone [33]. McGrath et al. [32] suggested that, given the high degree of uncertainty surrounding relationships between infestation levels, plant damage and yield loss, much more conservative thresholds should be used, especially in the case of smallholder farmers.
Action threshold levels based on expert opinion have been recommended for FAW control [32]. These recommendations have been presented as different thresholds for different maize growth stages as follows: during the early whorl stage, if 20% (range of 10-30%) of the seedlings are infested, or during the late whorl stage, if 40% (range of 30-50%) of the plants are infested, an insecticide application is warranted. During the tassel and silk stages, if 20% (range of 10-30%) of plants are infested, an insecticide application may be justified [32].
Decisions to apply insecticides should be supported by action threshold levels, taking into account important aspects that may influence the response of maize plants to pest damage, for example, the agroecological zone with its specific environmental conditions, expected yield and time and level of pest attack, to prevent unwarranted insecticide applications [2,32]. The climatic conditions under which this study was conducted are similar to those in many other regions in Africa where FAW populations persist, and continuous reinfestation of maize crops occurs throughout the growing season. Results of this study could therefore also apply to similar ecoregions on the continent.

Conclusions
This study was conducted under conditions that allowed continuous reinfestation by FAW during the cropping season. Yield losses due to FAW damage ranged between 26.5 and 56.8%. More than three spray applications generally did not provide further yield gains. Plants that were protected more during early growth stages yield higher than plants protected during later growths stages. Further research on the relationship between FAW damage and yield loss is needed in different agroecological zones to guide the development of action threshold levels.