Damage Assessment of Melanagromyza sojae (Diptera: Agromyzidae) on Soybean in Brazil

Soybean stem fly, Melanagromyza sojae Zehntner (Diptera: Agromyzidae), is an important soybean ( Glycine max ) pest in Eastern Asia that has recently colonized South America. The region colonized by M. sojae includes Brazil and several other major soybean growing countries. Management strategies for this pest remain largely undeveloped due to lack of information regarding its potential to injury soybeans. The objective of this study was to quantify soybean yield reduction caused by M. sojae injury. One experiment was carried out during two summer crop seasons (2020 and 2021) at Santa Maria, RS state, Brazil. Soybean was planted during late-season to ensure that high pressure of M. sojae adults were present in the fields. The number of seeds, 1,000-seed weight, seed yield and number of pods were quantified for the lower, middle and upper canopy, and plant height was compared to the amount of stem injured to determine percentage of injured stem. Each 1% of injured stem in the lower, middle and upper canopy segments significantly reduced the number of seeds per plant, 1,000-seed weight, and yield. Across all canopy segments, yield reduction reached 0.9 g per plant for every 1% of injured stem. Treatments where insecticide applications started during the vegetative phase presented the lowest damage by M. sojae. These data suggest that M. sojae is an economically important herbivore of soybeans under Brazilian growing conditions and highlight the need to develop efficient and sustainable management strategies for this pest.


Introduction
Soybean (Glycine max (L.) Merril) is the main cash crop grown in South America.Approximately 57 million hectares in South America are currently cultivated with soybean (FAOSTAT, 2020), 38.5 million of which were located within Brazil during the 2020/2021 crop season (CONAB, 2021).Extensive cultivation and variation in planting dates make this crop highly vulnerable to pests, including several invasive species (Pozebon et al., 2020).Soybean stem fly, Melanagromyza sojae Zehntner (Diptera: Agromyzidae), is native to eastern Asia and was recently detected in South America in 2015 (Arnemann et al., 2016), but it is suspected to have been present in southern Brazil since the 1980s (Gassen & Schneider, 1985).Since its detection in Brazil, M. sojae has spread across the South American soybean belt, reaching Paraguay (Guedes et al., 2017), Bolivia (Vitorio et al., 2019) and Argentina (Trossero et al., 2020).In Southern Brazil, outbreaks of M. sojae have been repeatedly observed on late-season soybeans, which are planted after maize (Zea mays L.) harvest beginning in late December (Folmann et al., 2017;Pozebon et al., 2020).within the same plant.Melanagromyza sojae larvae damage soybean plants by boring into the main stem and impairing xylem tissues (Talekar, 1989).Adult females feed and lay their eggs on upper (i.e., younger) leaves, piercing holes where one or two eggs are placed.After two to four days, the eggs hatch and the larvae bore into the leaf's main vein, reaching the main stem of the plant via the petiole.Each larva consumes around 1.4 mm of leaf tissue per hour (Lee, 1962), taking two days on average to reach the main stem after emergence and drastically reducing the possibility for control.
Melanagromyza sojae can attack soybean plants throughout the whole crop cycle, but population densities vary according to environmental factors such as rainfall, temperature, and availability of plant hosts.During periods of the year when mild temperatures predominate, M. sojae populations decrease drastically.The highest incidence is observed in the hottest and driest months, as intense rainfall restrains feeding and oviposition by adult females (Talekar & Chen, 1985;Yadav et al., 2015).In Asian countries, stem flies typically overwinter as pupae within dead soybean stems (Pozebon et al., 2020).In Brazil, winter survival has been facilitated by the presence of volunteer soybean plants in fields (Czepak et al., 2018) and alternative plant hosts, such as Trifolium resupinatum L. (Ferreira et al., 2020).
As a novel pest in South America, there is a paucity of data regarding the potential of M. sojae to reduce soybean yield, and integrated management programs targeting this pest remain largely undeveloped.Yield losses due to M. sojae have been estimated at 30% in Indonesia (Du & Hong, 1982) and 42% in India (Jadhav et al., 2013), varying according to region, crop fertilizer nutrition, soybean cultivar, planting date and management strategies adopted (Savajji, 2006).However, as South America has a completely different soybean growing environment than the endemic region of M. sojae, this information is still lacking for our continent.Thus, the objective of this work was to quantify M. sojae injury and its effect on yield under South American soybean growing conditions.

Experimental Sites
The same experiment was carried out during two summer crop seasons (2020 and 2021), at the Federal University of Santa Maria (29º42′48″S, 53º43′59″W), in Santa Maria, RS, Brazil.Soybean varieties TMG 7063 IPRO (planted on 27 January 2020) and 6968RSF (planted on 25 January 2021) were used in crop seasons 2020 and 2021, respectively.Soybean seeds were treated with 30 g a.i. of carbendazim + 70 g a.i. of thiram per 100 kg of seeds.Weeds were controlled prior to planting with an application of 1,005 g a.i./ha of 2,4-D + 1,040 g a.i./ha of glyphosate, and at soybean growth stage V3 (Fehr & Caviness, 1977) with an application of 1,040 g a.i./ha of glyphosate.Foliar sprays of strobilurin and triazole fungicides were carried out at growth stages V3, V7, R1, R4 and R5.2 for disease control.The soybean variety used in the crop season 2020 contained an insecticidal Bt protein, Cry1Ac, to control defoliating caterpillars, but the soybean variety used in the crop season 2021 lacked such a trait.Sap-sucking pests (stink bugs and whiteflies) were monitored at the field borders and managed with an application of 60 g a.i./ha of acetamiprid + 30 g a.i./ha of pyriproxyfen and 970 g a.i./ha of acephate, before they reached the experimental plots.

Experimental Design and Treatments
The experimental design was completely randomized with one factor (spray timing) and seven treatments, plus one additional treatment in the crop season 2020 (Tables 1 and 2).Ten replicates were used for evaluation of percentage of injured stem during weekly evaluations, and four replicates were used for evaluation of soybean yield and its components at crop senescence.Each plot was 6 × 20 m (12 soybean rows spaced 0.5 m) and contained approximately 2,400 soybean plants.We used weekly foliar applications of an insecticide to develop treatments of varying M. sojae densities.The number of applications ranged from 0 (untreated control) to 7 (additional treatment in the crop season 2020), which represented an attempt to keep the plants free of M. sojae (i.e., pest-free control).These applications occurred at distinct moments of the soybean growth cycle, starting at the early vegetative growth stages and ending at pod formation (i.e., growth stage R5.2).The insecticide used was 26.5 g a.i./ha of lambda-cyhalothrin + 35.2 g a.i./ha of thiamethoxam.Sprays were carried out using a CO 2 -pressurized backpack sprayer, with a spray volume of 150 L/ha and six spray nozzles (model XR 110 020) spaced 0.5 m from each other.

Evaluations
Evaluations were carried out weekly by randomly sampling 10 soybean plants from each plot, prior to the insecticide application, at growth stages V3, V6, V8, R2, R3, R4, R5.2 and R5.5 for crop season 2020, and V5, V7, R1, R2, R3, R4 and R5.3 for crop season 2021.Plant height was measured from the soil line to the last node of the main stem, and the presence of M. sojae tunnels was assessed and tunnels measured by opening the main stem longitudinally, from bottom to top.The percentage of injured stem was determined as a ratio between plant height and tunnel length in the main stem.For quantification of soybean yield and its components, four replicates were used, each being the average of four harvested plants (i.e., 16 soybean plants per treatment).Seed yield (g/plant) and yield components (number of pods, number of seeds and 1,000-seed weight) were quantified for each segment of the plant canopy (lower, middle and upper) and for the entire plant.Yield was estimated based on number of seeds and 1,000-seed weight.

Statistical Analysis
We used ANOVA and Scott-Knott test to determine if the means of the variables measured varied from each other (P ≤ 0.05).Linear regression analysis was used to determine the relationship between the percentage of injured stem and components of soybean yield (seeds/plant, 1,000-seed weight, pods/plant and yield/plant).The percentage of injured stem was analysed using the evaluations from growth stages R1, R2, R3, R4 and R5.5 for crop season 2020 and growth stages R4 and R5.3 for crop season 2021, which presented statistical differences among treatments.Figures for the linear regression analysis are presented for crop season 2020, which showed the best correlation among variables.Statistical analyses were carried out using Microsoft Excel and SISVAR (Ferreira, 2014).

Crop Season 2020 (Transgenic Soybean Variety)
3.1.1Amount of M. sojae Injury The number of pods, number of seeds, 1,000-seed weight and seed yield of the soybean plants were significantly affected (P ≤ 0.05) by M. sojae injury in the crop season 2020.The percentage of injured stem did not differ significantly among treatments for the evaluations from V3 to V8 and R5.2 (Table 3).The other evaluation timings (R1 to R4 and R5.5) presented statistical differences for this variable, resulting in reduction in yield components.The amount of injury caused by M. sojae varied according to the growth stage at which soybean plants were sprayed (Table 3).The regression models presented high coefficients of determination (R 2 > 0.60 for number of seeds, R 2 > 0.76 for 1,000-seed weight and R 2 > 0.74 for seed yield), indicating a likely relationship between M. sojae injury and reduction in soybean yield components.Note. 1 Coefficient of variation; 2 Standard error; 3 Means followed by the same letter do not differ among themselves by the Scott Knott test (p ≤ 0.05); 4 Non-significant.

Yield Component Analysis
There was no significant difference among treatments for number of pods in the lower (P = 0.9235) and middle (P = 0.1186) segments.Treatment had a significant effect on number of pods in the upper segment (P = 0.0229) in the ANOVA but not in the Scott-Knott test.However, all remaining yield components presented significant differences according to the Scott-Knott test (P ≤ 0.05), except for number of seeds in the lower segment (Table 4).The highest number of seeds in the middle segment (51.0 seeds/plant) was observed in treatment 1, whereas for the upper segments the highest values were obtained in treatments 1, 2 and 3, with 43.4,40.0 and 40.5 seeds/plant, respectively.Similarly, treatments 1, 2 and 3 presented the highest 1,000-seed weight values in all three segments, differing statistically from the remaining treatments.The values observed for 1,000-seed weight in treatments 1, 2 and 3 were: 198.2, 194.5 and 192.1 g, respectively, in the lower segment; 199.7, 203.0 and 197.5 g, respectively, in the middle segment; and 187.7, 188.4 and 174.1 g, respectively, in the upper segment (Table 4).Note. 1 Coefficient of variation; 2 Standard error; 3 Means followed by the same letter do not differ among themselves by the Scott Knott test (p ≤ 0.05); 4 Non-significant.

Yield Component Analysis
There was significant variation across the insecticide treatments for the following yield components: number of pods in the lower segment (P = 0.0023); number of seeds in the lower segment (P = 0.0073); 1,000-seed weight in the lower (P < 0.0001), middle (P < 0.0001) and upper (P = 0.0084) segments; and seed yield in the lower segment (P = 0.0021).The highest values for number of pods in the lower segment were observed in treatments 1, 3, 4 and 5, with 10.4, 11.1, 9.9 and 12.3 pods/plant, respectively, not differing significantly among themselves.
The highest values for number of seeds were observed in the lower segment of treatments 1, 3, 4 and 5, with 19.3, 20.0, 18.7 and 21.7 seeds/plant, respectively.As for 1,000-seed weight in the lower segment, the highest values were observed in treatments 1, 4 and 5, with 201.2, 206.1 and 204.0 g, respectively, not differing among themselves.In the middle segment, the highest values for 1,000-seed weight were observed in treatments 1 and 4, with 208.2 and 206.1 g, respectively, not differing from each other.In the upper segment, however, treatments 1, 2, 5 and 7 presented the highest values, with 179.3, 181.4,181.4 and 183.4 g, respectively, not differing among themselves (Table 6).Finally, the highest values for seed yield in the lower segment were observed in treatments 1, 3, 4, 5 and 7, with 4.0, 3.6, 3.8, 4.3 and 3.2 g/plant, respectively.Despite the differences observed among treatments, the resulting regression models between the aforementioned variables and the percentage of injured stem were not significant (except for 1,000-seed weight in the upper segment; P = 0.031), and have thus been omitted.

Number of Seeds and 1,000-Seed Weight
Soybean yield decreased as M. sojae injury increased, mainly as a result of reduced number of seeds and lower 1,000-seed weight.Despite receiving weekly insecticide sprays since growth stage V3 (crop season 2020) and V5 (crop season 2021), treatment 1 still resulted in plant injury by M. sojae and yield reduction (although lower than all other treatments), indicating that the potential for yield reduction can be even higher than observed in this study.
The results obtained in both crop seasons indicate that suppressing M. sojae contributed to an increase of seed yield, as observed for the middle and upper segments in the crop season 2020 and lower segment in the crop season 2021.The treatment where a spray occurred at the onset of M. sojae infestation presented the highest number of seeds in all canopy segments.The regression models for the middle and upper segments in the crop season 2020 indicated a reduction of 0.71 and 0.75 seeds/plant, respectively, for each 1% of injured stem, illustrating the potential of this pest to negatively affect one of the main yield components in soybean plants.
In the crop season 2020, 1,000-seed weight also decreased as injury by M. sojae increased.Means were grouped together according to a clear-cut pattern in the three canopy segments: treatments 1 to 3 did not differ significantly among themselves, and the lowest seed weight values were observed in the unsprayed control plot.
In the crop season 2021, treatment 1 presented the highest 1,000-seed weight values in the lower and middle segments, not differing significantly from treatment 4 in both segments and treatment 5 in the lower segment.All remaining treatments presented significant reductions in seed weight, which was expected as stem tunnelling by M. sojae larvae hinders xylem transport within soybean plants (Talekar, 1989), consequently affecting seed filling.

Seed Yield
Each 1% of soybean stem injured by M. sojae reduced seed yield by 0.9 g/plant in the crop season 2020.In comparison, studies carried out in East Asia during the 1980s estimated a yield reduction of 0.11 g/plant for each 1% of injured stem (Talekar & Chen, 1985).This value was estimated at 1.1 g/plant for the species Melanagromyza obtuse Malloch (Diptera: Agromyzidae) (Gangrade & Sing, 1976).Considering that modern soybean cultivars present lower leaf area index, lower height and less secondary stems than cultivars grown 40 years ago, each unit of injury was expected to produce a greater impact on yield components than previously reported.
Yield reductions as high as 63.7% were observed in treatment 8 (untreated) (Appendix B).Similarly, Jadhav et al. (2013) observed soybean yield losses ranging from 33 to 41% in India.This high potential for yield reduction is explained by the feeding behaviour of M. sojae larvae within the soybean stem, which damages vascular tissues and impairs xylem transport (Chiang & Norris, 1983).As a consequence, transport of water and nutrients from roots to stems and leaves is hindered (Talekar, 1989).Furthermore, soybean stems store most of the photoassimilates that are translocated to seeds during seed filling (Streeter & Jeffers, 1979).Thus, seed yield is directly affected by insect herbivory at this site within the plant.

Spray Timing and Management Strategies
Treatments where insecticide applications started during the vegetative phase (1, 2 and 3) presented the lowest damage by M. sojae in all evaluations of crop season 2020, not differing among themselves or treatments 8, 4 and 5 in the evaluations at R2, R4 and R5.2, respectively.Similarly, the lowest injury by M. sojae in the crop season 2021 was observed in treatments 1, 2 and 3, not differing among themselves according to the Scott-Knott test.Throughout the experiments, treatments with more sprays prior to evaluation presented lower percentage of stem injured by M. sojae.The reduction in percentage of injured stem in these treatments is probably linked to effective control of M. sojae adults, leading to fewer larvae boring into the main stem of the soybean plants.
After the larvae reach the main stem, the possibility for control is drastically reduced, as the insect becomes virtually unreachable by insecticide sprays.The high correlation found between M. sojae injury level (i.e., percentage of injured stem) and reduction in yield components (number of seeds, 1,000-seed weight and seed yield) points to a high potential for damage in soybean and highlights the need to develop efficient and effective management strategies for this pest.
Population outbreaks of M. sojae have so far remained restricted to late-season soybeans (i.e., planted from December onward) in southern Brazil.However, Pozebon et al. (2020) alerts that M. sojae infestations are likely to become frequent in main-season soybeans as well (planted from October onward), due to the pest's adaptability to Brazilian growing conditions.Besides reducing seed yield in main-season soybeans, the development of such a scenario would boost M. sojae population levels during late planting, potentially compromising late-season soybean cultivation altogether.Soybean growers in southern Brazil typically cut expenses during late-season cultivation, leading to lower use of insecticides.When Bt soybean cultivars are used, the crops are sprayed with insecticide only after growth stage R3, targeting phytophagous stinkbugs.Our study included this scenario and showed the potential for M. sojae to damage soybean plants and reduce seed yield.
Treatment 5, where two weekly insecticide sprays were carried out after growth stage R3, presented yield loss as high as 36.53%,not differing statistically from the unsprayed control plot.
Soybean pest management in Brazil relies heavily on the use of chemical insecticides and Bt cultivars.The transgenic cultivar MON 87701 × MON 89788, expressing insecticidal Bt protein Cry1Ac, was commercially released in Brazil in 2010 (Bernardi, 2012).Yu et al. (2014) studied the exposure of non-target arthropods to Cry1Ac in soybean fields and reported the presence of detectable Bt protein levels in M. sojae adults, with no apparent effect upon the insect's development.Bt proteins Cry4Aa, Cry4Ba, Cry11Aa, Cyt1Aa, Cry10Aa and Cyt2Ba knowingly affect dipteran insects (Ben-Dov, 2014), but no transgenic cultivar made available to date expresses these proteins.Soybean cultivars in India are classified according to their susceptibility to M. sojae attack, based on percentage of attacked plants (Savajji, 2006), but in Brazil this information is lacking (Curioletti, 2016).Unavailability of Bt cultivars with activity against dipteran pests, combined with the lack of information regarding susceptibility to M. sojae attack in current soybean cultivars, has constrained management strategies to insecticide spraying.The use of insecticide treated seeds, combined with foliar sprays shortly after plant emergence, protects soybean plants during the growth stages most vulnerable to M. sojae attack (Pozebon et al., 2021).Chlorantraniliprole and chlorpyrifos are recommended for seed treatment and foliar sprays, respectively (Curioletti et al., 2018).
The findings from this study highlight the importance of reducing the coexistence of pest and crop by spraying soybean plants during the vegetative phase, thus preventing yield losses and avoiding lost revenue.Further studies should assess M. sojae injury on main-season growing conditions and use the data to determine an economic injury level for this pest in soybean.
f pods per pla nts is shown e segment and by M. sojae.= 0.7678; P < segments, resp injured by Me R² = 0.6011; N plants.Data po rop season sojae and num 001) and upper art represent m Figure 2 lower (R² P

Table 1 .
Treatments, number of sprays and growth stages of the soybean plants at each spray.Crop season 2020, Santa Maria, RS, Brazil

Table 2 .
Treatments, number of sprays and growth stages of the soybean plants at each spray.Crop season 2021, Santa Maria, RS, Brazil

Table 3 .
Percentage of soybean stem length injured by Melanagromyza sojae in relation to insecticide sprays starting at different growth stages of the crop.Crop season 2020, Santa Maria, RS, Brazil P-value 0.6601 0.3152 0.6086 0.0448 0.0002 0.0015 < 0.0001 0.2491 < 0.0001

Table 4
oefficient of v s by the Scott of treatment w f seeds per p nt (R² = 0.6011 segment for of 2.4 (R² = 0. d weight, for th variation; 2 St Knott test (p ≤ was observed plant for all c 1; P < 0.0001) each 1% of s .8525;P < 0.00 he lower, midd

Table 5 .
Amount of M. sojae InjuryTreatments 1, 2 and 3 showed the lowest percentage of stem injured by M. sojae, not differing among themselves by Scott-Knott test (Table5).The remaining treatments presented higher percentage of injured stem and did not differ among themselves, ranging from 46.4 to 52.6% at growth stage R4, and from 56.3 to 65.5% at R5.3.Percentage of soybean stem length injured by Melanagromyza sojae in relation to insecticide sprays starting at different growth stages of the crop.Crop season 2021, Santa Maria, RS, Brazil

Table 6 .
Yield components following Melanagromyza sojae injury in relation to insecticide sprays starting at different growth stages of the crop, in lower, middle and upper canopy segments of the soybean plants.Crop season 2021, Santa Maria, RS, Brazil Note. 1 Coefficient of variation; 2 Standard error; 3 Means followed by the same letter do not differ among themselves by the Scott Knott test (p ≤ 0.05); 4 Non-significant.

Grain yield following Melanagromyza sojae injury in relation to insecticide sprays starting at different growth stages of the crop, in lower, middle and upper canopy segments and in whole soybean plants. Crop season 2021, Santa Maria, RS, Brazil
Note. 1 Coefficient of variation; 2 Standard error; 3 Means followed by the same letter do not differ among themselves by the Scott Knott test (p ≤ 0.05); 4 Non-significant.
Note. 1 Coefficient of variation; 2 Standard error; 3 Means followed by the same letter do not differ among themselves by the Scott Knott test (p ≤ 0.05); 4 Non-significant.