Migration dynamic of fall armyworm Spodoptera frugiperda (J. E. Smith) between China's southern Yunnan and the Indo‐China Peninsula

Abstract

BACKGROUND

Since the invasion of fall armyworm (FAW) into Asia, it has posed a threat to agricultural production across Asian countries. The southwestern border region of China serves as a key pathway for FAW migration from the Indo‐China Peninsula into China. Understanding the transboundary migration dynamics of this pest facilitates source control and early prevention and control measures.

RESULTS

Systematic monitoring for transboundary migration dynamics of FAW was conducted during 2019–2024 in Jiangcheng County of Yunnan province located at the border area between Southwestern China and the Indo‐China Peninsula. The results indicate typical northward migration of FAW during spring, summer, and autumn, while winter was characterized by local dispersal population. Northward migratory FAW populations primarily originated from Myanmar, Laos, Vietnam, and Thailand (immigrant FAW population sources accounted for 93.56% in spring, 89.35% in summer, and 82.93% in autumn, 98.89% in winter). In terms of distribution in the migration destination areas, the landing points were predominantly located in Yunnan Province, Guangxi Province, and Guizhou Province in China (97.18% in spring, 89.89% in summer, 89.22% in autumn, and 93.89% in winter).

CONCLUSION

FAW populations in the Indo‐China Peninsula undergo continuous cross‐border northward migration into China through south Yunnan during spring, summer, and autumn, with no backward migration observed in all seasons. These findings provide a scientific basis for establishing precise monitoring and regional integrated management for FAW. © 2026 Society of Chemical Industry.

Population dynamic of cross‐border migration of fall armyworm in the border areas between southern Yunnan, China, and the Indo‐China Peninsula during 2019–2024 confirm northward migration in spring, summer, and autumn and no back‐migration in all seasons: (A) daily catches dynamic, (B) monthly catches dynamic, (C) migration period delineation, and (D) pathway simulation for typical migration peak event in different seasons).

1. INTRODUCTION

The fall armyworm (FAW), Spodoptera frugiperda (J. E. Smith) (Lepidoptera: Noctuidae), is a major global agricultural pest originating from tropical and subtropical regions of the Americas. ¹ , ² It causes substantial economic losses to agricultural systems. ³ , ⁴ Characterized by its strong migratory capacity, broad host range, high reproductive potential, and resistance to insecticides (including Bt toxins), S. frugiperda poses a significant challenge to global management efforts. Frequent intercontinental transport (via cargo or aircraft) is considered as the primary reason for its invasion of Africa in 2016, thereby altering its global distribution and occurrence patterns. ⁵ , ⁶ The resulting crop losses threaten both the food security and economic livelihoods of smallholder farmers worldwide. In Africa alone, annual maize yield losses due to S. frugiperda are estimated at up to 17.7 million tons, equivalent to an economic loss of approximately US$ 4.661 billion. ⁷ The pest's invasion has also led to increased pesticide inputs and economic costs for small‐scale farming systems, ⁸ as well as a range of ecological and social issues, including interspecific competition between invasive and native species, imbalances in biodiversity, and threats to environmental sustainability and human health. ⁴ , ⁹ , ¹⁰ , ¹¹ The migratory behavior of FAW moths has accelerated their geographical expansion and host adaptation across the Americas, Africa, Asia, and Oceania. ⁴ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ In Africa, the migratory habit of FAW is a major factor driving its rapid spread across most of the continent. ⁴ , ⁵ S. frugiperda occurs year‐round in much of Africa, though in regions such as South Africa, Zimbabwe, Sudan, and Egypt migration is influenced by dry conditions and lower temperatures. ⁴ , ⁷ In recent years the pest has expanded and established in most suitable regions worldwide, except for Europe (https://gd.eppo.int/taxon/LAPHFR/distribution). Unlike its introduction into Africa, the remarkable migratory capacity of S. frugiperda played a key role in its rapid invasion from Africa into Asia and Oceania. ⁴ , ¹¹ , ¹⁷ , ¹⁸ , ¹⁹

FAW was first detected in the border region of Yunnan, China, in December 2018. By April 1 the following year, long‐distance migration across the South China Sea was recorded on Yongxing Island, with an estimated migration distance exceeding 600 km. ²⁰ In East Asia, influenced by cropping patterns, the tropical southwest monsoon, and the East Asian monsoon, the pest generally exhibits a relatively clear ‘northward migration and southward return‐migration’ pattern along a southwest‐northeast trajectory. ²¹ , ²² , ²³ This aligns with its expansion route across Southeast and East Asia, ²⁴ indicating its strong adaptability to regional climatic and agricultural conditions and its rapid completion of the ecological process of ‘invasion–establishment–spread’. ²¹

Based on overwintering area surveys, ²⁵ meteorological trajectory analysis, ²² , ²³ , ²⁶ , ²⁷ and monitoring of population dynamics using attractants/light traps, ²⁸ the seasonal migration routes of FAW in China have been clarified, which shows some similarities with that in North America. ⁴ , ¹⁵ The northern overwintering limit of FAW in China is located between 28 and 31° N, ²⁵ while in North America it is around 28° N. ² In terms of year‐round breeding areas, southern and southwestern regions of China, such as Yunnan, Guangxi, Guangdong, and Hainan, face persistent pressure from FAW reproduction and damage throughout the year, similar to the situation in Texas and southern Florida in the United States. ² , ²⁵ , ²⁹ However, China's year‐round breeding areas are different from those in North America, as southern China and border regions between China and Indo‐China also face persistent pressure from immigrant populations originating from outside China. ²⁵ , ²⁷ The migration routes of FAW in China can be divided into eastern and western pathways based on upper‐air wind patterns. ²⁶ For the eastern pathway, northward migration begins from March–April to areas south of the Yangtze River Basin, reaches the Yangtze River Basin in May–June, and can further threaten North China by July‐August. ³⁰ For the western pathway, populations from year‐round breeding areas migrate into the Sichuan Basin in April and subsequently move northward generation by generation through the Sichuan Basin towards northern and eastern regions from May to June. ³¹ Populations from year‐round breeding areas migrate over multiple generations from spring to summer into the Yangtze River Basin, the Huang‐Huai‐Hai region, and even northeast China, exhibiting a nationwide southwest–northeast migration pattern. In autumn, they begin to return southwestward. This phenomenon has been empirically supported by monitoring with high‐altitude insect searchlights, isotope analysis, and trajectory analysis. In the southwestern border regions of China, such as Ruili and Lancang in Yunnan, FAW exhibits distinct seasonal northward migration in spring/summer and southward return in autumn. ²⁹ , ³² In the Ruili area on the China–Myanmar border, populations from Myanmar, India, and Bangladesh continuously migrate eastward into central and eastern Yunnan, China, primarily from April to June, facilitated by the South Asian monsoon, with return migration occurring in October. ²⁹ Monitoring in Lancang County, Pu'er City, also confirmed persistent northward migration of FAW from Myanmar into China during spring and summer, and return migration in autumn. ³² Similarly, long‐distance cross‐sea migration of FAW populations from southern Laos, southern Vietnam, and southern Thailand into Guangdong, Fujian, and other areas in southern China were observed over the South China Sea from April to August (spring and autumn), with return migration in September. ²⁰ Furthermore, FAW populations in eastern China are also significant sources for other East Asian countries like Japan and Korea. ³³ Analysis using the GEARN‐insect model by Otuka suggested that FAW found in Japan likely originated mainly from Fujian, Zhejiang, Jiangxi, and Taiwan. ³⁴ This indicates that for the entire East Asian region, the ultimate source of FAW populations is the Indo‐China Peninsula.

Maize, a preferred host crop for FAW, is widely cultivated in various countries of the Indo‐China Peninsula, ranking as the second to sixth most planted crop, and has shown a stable increasing trend in recent years in countries like Myanmar and Cambodia. ²⁴ In the Indo‐China Peninsula, FAW can reproduce and cause damage year‐round, occurring from January to December, with peak occurrence in Thailand, Vietnam, and other areas typically from May to October. ²⁴ FAW infests maize fields to varying degrees in most provinces across the peninsula, with Thailand being particularly severely affected, leading to decreased maize yields, for example an 8% reduction in 2019. ²⁴ In Myanmar, Ayeyarwady and Shan reported the highest FAW infestation areas, and FAW has become the most common pest in maize fields. ³⁵ China is a major country in terms of maize planting area, with 44.74 million hectares sown in 2024 (https://data.stats.gov.cn/easyquery.htm?cn=C01). China's main maize‐producing regions span a wide latitudinal range, encompassing FAW's year‐round breeding areas, migratory transition zones, and key control areas, and provides favorable resource conditions for FAW's migration and damage from low to high latitudes. Field surveys by Yang et al. indicated that FAW can reproduce and cause damage year‐round in China's south subtropical and tropical regions, with the northern boundary of the year‐round breeding area approximately near the January 10 °C isotherm. ³⁶ This region falls within China's winter‐sown maize area, which also supports FAW persistence. Under the influence of monsoon circulation, the risk of large‐scale migration and damage by FAW is higher. Therefore, the border region between southwestern China and the Indo‐China Peninsula serves as a bridgehead for FAW migration in East Asia, and the continuous influx of FAW from Myanmar, Laos, Thailand, and Vietnam exacerbates its occurrence and damage in East Asia.

Facing the challenge of global FAW management, the Chinese government has proposed regional integrated management strategies based on migration patterns and cropping systems, implementing measures focused on source control and comprehensive prevention. Precise monitoring and effective control of migratory FAW populations at their source and along migration pathways are key to source control. This study employs vertical searching light trap monitoring to conduct long‐term surveillance of cross‐border migration activities between southern Yunnan, China, and the Indo‐China Peninsula from 2019 to 2024, aiming to further clarify the dynamics of cross‐border migratory FAW populations in this region.

2. MATERIALS AND METHODS

2.1. Study site

The study site was selected at the Jiangcheng Experimental Station of the Institute of Plant Protection, Chinese Academy of Agricultural Sciences (hereinafter referred to as Jiangcheng Station; 22°41′13.13″ N, 101°38′40.63″ E, altitude approx. 800 m) (Fig. 1), located in Baozang Town, Jiangcheng Hani and Yi Autonomous County, Pu'er City, Yunnan Province (hereinafter referred to as Jiangcheng). Jiangcheng County is situated at the junction of China, Laos, and Vietnam, also known as ‘one city connecting three countries’. Climatically, it belongs to the south subtropical humid climate zone and is a typical river valley agricultural area.

Figure 1.

Schematic map of study sites and design of vertical searching light trap and scenario of actual effect for population dynamics monitoring of fall armyworm at Jiangcheng Station (JC Station).

2.2. Vertical searching light trap monitoring

This study used a 1000 W metal halide bulb (Model JLZ1000BT; Shanghai Yaming Lighting Co., Ltd., Shanghai, China) as the light source. This bulb emits white light with a brightness of 105 000 lm and a color temperature of 4000 K, coupled with a Yaming GT193 luminaire with a reflective profile, effectively projecting the light to an altitude of about 500 m above the ground to attract phototactic insects migrating through the air (Fig. 1). The insect collection device consisted of a funnel and a 60‐mesh nylon net cage for collecting insect samples, followed by sorting and identification. To minimize interference with ground insect communities, the high‐altitude searchlight was installed on a farmer's house roof near farmland, approximately 5 m above the ground. The searchlight was set to turn on daily at 19:00 (Coordinated Universal Time (UTC) +08:00) and off at 07:00 the next day (UTC + 08:00). The monitoring site in Jiangcheng County, Yunnan Province, is within the FAW's year‐round breeding area. Barring power outages or system failures, uninterrupted monitoring was conducted from 18 January 2019 to 31 December 2021. Collected insect samples were placed in a −20 °C freezer each morning for freezing, followed by morphological sorting, identification, and preservation. The numbers of male and female FAW moths were recorded daily, forming a systematic daily catches dataset.

2.3. Data processing and statistical analysis

This study employed various statistical methods to systematically analyze the high‐altitude searchlight monitoring data. To delineate the annual migration cycle of FAW, Fisher's optimal partitioning method was applied to the time series of daily average catch data (using a 10‐days unit) for chronological segmentation, determining the optimal partition points based on the inflection point of the minimum loss function. Simultaneously, Morlet wavelet analysis is specifically designed to handle non‐stationary signals by providing time‐frequency localization, and it allows for the identification of periodic components and how these components evolve over time. ³⁷ , ³⁸ Here, Morlet wavelet analysis was used to analyze the time series of weekly catch data of FAW, revealing the multi‐scale periodic characteristics of population occurrence during the monitoring period. For sex ratio analysis, one‐way ANOVA was used to compare the proportion of females among different months, and paired t‐test were used to analyze significant differences in the male‐to‐female ratio within each month. A generalized linear mixed model (GLMM) was used to analyze the effects of different years and months on the monthly total catch of FAW. Fisher's optimal partitioning, one‐way ANOVA, and paired t‐test were performed using SAS 9.4 (SAS Institute, Cary, NC, USA). Morlet wavelet analysis was implemented in MATLAB R2020a (MathWorks, Natick, MA, USA). GLMM analysis was performed using R language (with packages lme4, glmmTMB, MASS, etc.).

2.4. Definition of peak migration day

To systematically identify migratory events for subsequent trajectory analysis, we adopted and adapted the criteria established by Lin ³⁹ and Cang et al. ⁴⁰ regarding the population dynamics of oriental armyworm (Mythimna separate). A peak migration day was defined based on the following hierarchical conditions: Initial identification (Condition A): A day is identified as the start of a peak when the daily catch (Nt) is ≥ 10 individuals and represents at least a two‐fold increase over the previous day (Nt/Nt‐1 ≥ 2). Duration of peak (Condition B): Once a peak is initiated, consecutive days with catches ≥ 10 individuals are maintained within the same peak period. The period concludes when the daily catch drops below the 10‐individual threshold. Successive peaks (Condition C): To distinguish between independent migratory waves, a new peak is only recorded if Condition A is met again after an interval of at least 2‐days following the end of a prior peak. Rationale for threshold selection: The selection of these criteria, particularly the numerical threshold of 10 individuals, is strategically aligned with the ecological context of our study. Firstly, ecological adaptation to year‐round breeding areas. The study site is in a year‐round breeding zone for S. frugiperda, characterized by continuous moth activity and significant overlap of generations. Unlike the ‘all‐or‐nothing’ surges typical of seasonal invasion zones, migration here often manifests as frequent, low‐amplitude pulses. By prioritizing the relative growth rate (two‐fold increase), we can effectively distinguish migratory ‘pulses’ from the fluctuating local background population. Secondly, sensitivity in low‐density years. During the 2022–2024 period, the overall population density remained relatively low. Utilizing a high absolute threshold would lead to the omission of critical migratory signals. This sensitive approach ensures a comprehensive and representative dataset, which is essential for the accuracy and seasonal coverage of the subsequent high‐resolution trajectory simulations.

2.5. Trajectory analysis method

Based on the identified peak migration days from 2019 to 2023, this study used HYSPLIT trajectory analysis software based on Weather Research and Forecasting (WRF) and ECMWF Reanalysis v5 (ERA5) data to simulate the migration paths on these migration peak days. ⁴¹ , ⁴² WRF 4.7.1 software was used for meteorological data processing under the operating system Ubuntu OS 20.04. Python 3.8 was used to automate the batch execution of the arw2arl.exe program, converting the output results of the WRF model into ARL files, using these files as the meteorological background data for the HYSPLIT trajectory analysis model.

2.6. Parameter settings of the WRF model

The numerical simulation was performed using the WRF‐Advanced Research WRF (ARW) model. The simulation employed a two‐way nested grid configuration with horizontal resolutions of 27 km (d01) and 9 km (d02), respectively. The Mercator projection was selected to minimize geometric distortion within the tropical/subtropical study area. The initial and boundary conditions were driven by the National Centers for Environmental Prediction Final Analysis data, which provides comprehensive atmospheric state variables. The integration time step was set to 120 s to ensure numerical stability. Model outputs for the inner domain (d02) were archived at a 1‐h temporal resolution to provide high‐frequency wind field data for subsequent trajectory analysis.

2.6.1. The physical parameterization schemes selection for the WRF model

Microphysics: The Thompson scheme, a doubly moment‐sensitive scheme for improved cloud‐aerosol interactions.

Radiation: The rapid radiative transfer model for general circulation models (GCMs) for both longwave and shortwave radiation processes.

Planetary boundary layer: The Yonsei University scheme, a non‐local first‐order closure scheme.

Surface layer: The Monin–Obukhov similarity theory.

Land surface: The Noah land surface model, which manages soil moisture and temperature.

Cumulus parameterization: The Kain–Fritsch scheme, applied to account for sub‐grid scale convective processes.

2.6.2. HYSPLIT trajectory analysis model

In this study, three‐dimensional trajectory simulations were performed using the HYSPLIT model (v 5.4.2), driven by high‐resolution meteorological fields generated by the WRF model. In HYSPLIT model, altitude layers were set at 300, 500, 700, 900, and 1100 m above ground level, which were assumed release heights for FAW moths. Trajectory analysis time, here both forward and backward trajectory, durations were set to 12 h (UTC 11:00–23:00 and UTC 23:00–11:00, i.e. one flight night not exceeding 12 h) to infer the immigration area and emigration source of FAW populations passing through Jiangcheng Station.

We introduced the vector sum of its flight speed of insects and drift compensation angle into the HYSPLIT model to output the trajectory and landing point. Due to the lack of direct evidence of self‐flight speed of S. frugiperda, we employed trajectory tracing to analyze S. frugiperda as a typical nocturnal moth, aiming to determine its possible source areas and landing points. We also assumed that the high‐altitude S. frugiperda moths migrated in a downwind mode with an active self‐powered flight speed of 4.5 m s⁻¹. This setting 4.5 m s⁻¹ (4–5 m s⁻¹) represents the optimized biological airspeed for migratory noctuids under free‐flying conditions, as empirically observed in radar studies and other trajectory analysis settings. ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ Drift compensation behaviors promote optimal migration trajectories for compass‐mediated migration insects. ⁴⁸ , ⁴⁹ The inclusion of a 30° directional deflection to the right of the wind vector is a biologically grounded parameter intended to simulate the common orientation and drift‐compensation behavior characteristic of migratory noctuids. This assumption setting has been specifically validated as a highly effective constant for simulating the seasonal invasion of S. frugiperda across the complex terrains of East Asia, aligning model outputs with historical field‐trapping data. ²² , ²³ , ²⁷ , ³⁰ This configuration has been successfully applied in simulating the seasonal migration pathways of S. frugiperda across East Asia.

All endpoint coordinates were processed in batch using R language (with packages ggspatial, maps, mapdata, sf, etc.) to plot trajectories and reverse‐geocoded to obtain geographical locations. The distribution of endpoints for forward and backward trajectories in different years and seasons was statistically analyzed.

3. RESULTS AND ANALYSIS

3.1. Cross‐border migratory population dynamics

After the invasion of FAW into southern Yunnan in December 2018, damage to winter maize by FAW was first confirmed on January 11, 2019, in Shuicheng Village, Baozang Town, Jiangcheng (22°40'59.15” N, 101°38'25.22” E). From 2019 to 2024, a total of 34 772 FAW adults were captured, with an annual average of approximately 5800 individuals. The earliest detection date of FAW moths was 24 February 2019 (Fig. 2(A)). The annual capture quantity of FAW was as follows: 2019, 11 615; 2020, 9607; 2021, 5511; 2022, 2451; 2023, 3323; 2024, 2265 (Fig. 2(B)). FAW moths were captured in each month, showing clear seasonal patterns in migratory activity. Visually, from the daily catch data from 2019 to 2024, FAW showed multiple sharp increases and decreases in catch numbers during spring, summer, and autumn. Furthermore, based on the average daily catch over the 6 years, distinct population peaks indicative of typical peak FAW migration activity occurred in mid‐March, mid‐April, mid‐May, mid‐June, early July, late September, and October (Figs 2(A) and 3(A)).

Figure 2.

The population dynamics of migratory fall armyworm (FAW) crossing the border in southern Yunnan from 2019 to 2024: (A) daily captures and (B) annual captures. The dotted blue lines in (A) are seasonal segments.

Figure 3.

Monthly captures of fall armyworm (FAW) (A) and Fisher's optimal segmentation for migration period of FAW (B) in the vertical searching light trap from 2019 to 2024 at JC Station: (a) local dispersal period; (b) first north toward migration; (c) second north toward migration period; (d) transitional period between second and third north toward migrations; (e) third north toward migration.

3.2. Division of cross‐border migration periods

To clarify the migration periods of FAW in this region, Fisher's optimal partitioning method was used for ordered sample clustering analysis on the time series data of weekly average catch data. The results indicated that the migratory activity of FAW in this area can be divided into five periods: a, winter dispersal period (28 October to 4 March of the following year); b, spring northward migration period (4 March to 20 May); c, summer northward migration period (20 May to 15 July); d, summer–autumn transition period (15 July to 23 September); e, autumn northward migration period (23 September to 28 October) (Fig. 3).

3.3. Sex composition of migratory populations

Regarding the dynamics of the sex composition of the migratory FAW populations from 2019 to 2024, there was a highly significant difference in the proportion of females among different months (F _11,67 = 4.62, P = 0.00013), and female ratios in different months were all below 50% (Fig. 4). The highest proportion of females occurred in February (45.40 ± 7.22%) and the lowest occurred in November (23.35 ± 5.32%) (Fig. 4). Paired t‐test on monthly captures data indicated that, except for February and March, the female catch was significantly lower than the male catch in all other months. In other words, the sex composition of the migratory FAW populations captured at the Jiangcheng monitoring station was generally male‐biased (Fig. 4 and Table 1).

Figure 4.

Sex composition of the fall armyworm (FAW) populations trapped by vertical searching light traps from 2019 to 2024 at JC Station. The stacked columns represent the average captures of FAW for each month. The dotted line shows the 50% position and asterisks indicate significance by paired t‐test. *P < 0.05, **P < 0.001, ***P < 0.0001; ns, no significance.

Table 1.

Paired t‐test for monthly captures of female and male fall armyworm trapped by vertical searching light traps from 2019 to 2024 at Jiangcheng Station

Month	Average captures		Standard error		Ratio (%)		t value	P value
	Female	Male	Female	Male	Female	Male
Jan	25.20	46.00	5.86	9.10	35.96	64.04	3.597	0.0228*
Feb	44.67	47.50	17.28	15.00	45.40	54.60	1.139	0.3061ns
Mar	164.00	257.83	70.05	83.70	41.15	58.85	1.832	0.1265ns
Apr	123.17	199.00	36.01	51.30	38.00	62.00	5.309	0.0032**
May	253.67	388.33	99.27	112.95	34.25	65.75	4.316	0.0076**
Jun	473.67	1005.00	179.45	322.19	29.90	70.10	22.185	<0.0001***
Jul	312.67	672.17	121.06	261.19	30.35	69.65	10.158	0.0002***
Aug	126.00	204.00	44.22	55.54	36.16	63.84	10.993	<0.0001***
Sep	208.17	429.67	71.11	180.42	34.42	65.58	10.748	<0.0001***
Oct	211.67	402.33	121.28	110.59	26.13	73.87	9.07	0.0003***
Nov	49.33	98.50	29.68	31.24	23.35	76.65	11.236	<0.0001***
Dec	18.83	52.67	6.66	22.00	27.76	72.24	5.645	0.0024**

3.4. Seasonality and periodicity of cross‐border migration population dynamics

A generalized linear mixed model (GLMM) with a negative binomial distribution was used to analyze the monthly average catch for different years and months from 2019 to 2024. The model included month as a fixed effect and year as a random effect to assess both seasonal dynamics and interannual variation. The results showed significant differences in the monthly average catch of FAW among different months (Fig. 5). Month, as a fixed effect, explained 62.87% of the model variation (R ² = 0.6287, Akaike Information Criterion (AIC) = 485.7) (Fig. 5(E) and Table 2), indicating that month was the dominant factor influencing the monthly average catch. The variance for year as a random effect was only 0.2633 (ICC = 0.0141), indicating small interannual variation, meaning the seasonal effect dominated the monthly average catch. Catches from March to October were significantly higher than in January (P < 0.001), peaking in June (β = 2.9774, P < 0.001) (Table 2), indicating continuous population increase of FAW in the year‐round breeding area during spring and summer, reaching the peak of migratory population size in June.

Figure 5.

Results of generalized linear mixed model analysis for monthly captures of fall armyworm (FAW) at Jiangcheng Station from 2019 to 2024: (A) monthly captures dynamic of FAW; (B) monthly captures predictions with 95% confidence intervals; (C) coefficient estimated for fixed effect of month compared to January; (D) random effect of year. Intraclass Correlation Coefficient (ICC).

Table 2.

Analysis results of the negative binomial generalized linear mixed model (GLMM) for monthly captures of fall armyworm at Jiangcheng Station from 2019 to 2024

Effect	Parameter	Estimate β	Standard error	Z value	P value	Significance
Fixed effects	Intercept (Jan)	0.719	0.41	1.753	0.08	ns
	Feb	0.466	0.465	1.001	0.317	ns
	Mar	1.888	0.422	4.473	<0.001	***
	Apr	1.659	0.426	3.893	<0.001	***
	May	2.197	0.417	5.273	<0.001	***
	Jun	2.977	0.41	7.266	<0.001	***
	Jul	2.439	0.413	5.903	<0.001	***
	Aug	1.53	0.427	3.586	<0.001	***
	Sep	2.106	0.417	5.048	<0.001	***
	Oct	2.089	0.418	5.003	<0.001	***
	Nov	0.709	0.45	1.576	0.115	ns
	Dec	0.056	0.492	0.115	0.909	ns
Random effects	Variance		ICC
	0.263		0.0141
Model performance	AIC	BIC	Marginal R 2	Condition R 2	Dispersion parameter for nbinom2
	485.7	517.6	0.629	0.816	4.28

Model: mean monthly captures ~ month + (1 | year). ns, no significance.

Based on the Morlet wavelet analysis results, the raw (Fig. 6(A)) and normalized time series (Fig. 6(B)) data reveal a fluctuating population density with a prominent ‘front‐loaded’ distribution. High‐intensity capture events are concentrated within the first 150 weeks, followed by a transition to a lower‐amplitude, relatively stable phase in the latter half of the monitoring period. The Morlet wavelet transform identifies several regions of high power (warm colors in Fig. 6(C)). Significant oscillations (enclosed by black dashed lines, P < 0.05) are intermittently observed at scales of 2–8 weeks, 16 weeks, and ~64 weeks. The energy distribution is non‐stationary, indicating that the periodic strength of the population dynamics evolves over time (Fig. 6(C)). The global power spectrum identifies three dominant periodicities (Fig. 6(D)): T ₁ = 50.98 weeks, the primary peak, corresponding to a strong annual cycle, which underscores the influence of seasonal environmental drivers or host phenology; T₂ = 18.03 weeks, a secondary peak reflecting sub‐annual or seasonal population surges, likely linked to the regional voltinism (generation cycles) of the pest; T₃ = 4.31 weeks, a short‐term oscillation consistent with the development duration of a single generation.

Figure 6.

Morlet wavelet analysis for daily captures of fall armyworm in each week from 1 January 2019 to 31 December 2024 at Jiangcheng Station. (A) Time series data of weekly captures dynamic of FAW moths. (B) Z‐score standardization for time series data of weekly captures dynamic of FAW moths. (C) Morlet wavelet power spectrum (black dashed line represents the 95% confidence level against red noise). (D) Global wavelet power spectrum of Morlet wavelet analysis (red, blue, and green dashed lines represent the corresponding significance of 95%, 90%, and 80%, respectively, and the peak periods of the full line exceeds the dashed line represent significant period scale at different significant level).

3.5. Trajectory analysis of cross‐border migration peak days

Based on population dynamics data and the peak day criteria, 661 peak migration days were identified between 2019 and 2023. Trajectory analysis for peak days in different seasons throughout the year showed that the source populations of FAW passing through the Jiangcheng monitoring station primarily originated from foreign countries in Indo‐China Peninsula (Fig. 7).

Figure 7.

Simulation of the cross‐border migration routes by trajectory analysis model based on HYSPLIT model Weather Research and Forecasting for fall armyworm (FAW) peak migration day in different seasons from 2019 to 2023. The track times for the backward trajectory (solid lines) and the forward trajectory (dashed lines) were both set to 12 h and 6 h. Altitude layers were set at 300 m (red line), 500 m (blue line), 700 m (green line), 900 m (purple line), and 1100 m (orange line) above ground level, which were assumed release heights for FAW moths.

Landing points of forward trajectory paths for fall armyworm trapped at Jiangcheng Station in different seasons were mainly concentrated in China (spring, 97.18%; summer, 89.89%; autumn, 89.22%; winter, 93.89%; Fig. 8(A) and Supporting Information S1). Specifically, landing points were predominantly concentrated in Guizhou (44.41%) and Guangxi (30.51%) during the spring. In summer, landing points were mainly terminated in Yunnan (53.14%) and Guizhou (21.90%). By autumn, landing points were concentrated in Yunnan (74.40%), which serves as a critical transit hub. During the winter, the landing points were primarily located in Guizhou (45.56%) and Yunnan (35.56%) ((Fig. 8(A)).

Figure 8.

Distribution of simulation of the cross‐border migration landing points of fall armyworm with trajectory analysis during peak migration days in different seasons from 2019 to 2023. (A) Forward trajectory landing points for 12 h, only the regions with a percentage greater than 1% are plotted. (B) Backward trajectory landing points for 12 h, only the regions with a percentage greater than 1.5% are plotted. The specific data can be found in Supporting Information S1 and S2).

Backward trajectory paths for fall armyworm captured at Jiangcheng Station in different seasons were mainly concentrated in Myanmar (spring, 70.17%; summer, 63.52%; autumn, 24.78%; winter, 48.89%) (Fig. 8(B) and Supporting Information S2). Specifically, during the spring, summer, autumn, and winter migration periods, populations mainly originated from Myanmar, Laos, Vietnam, and Thailand (foreign source percentages: spring 93.56%, summer 89.35%, autumn 82.93%, winter 98.89%).

The origin regions of FAW populations were distributed as follows: spring, mainly in Shan (49.83%) and Kachin (15.59%) in Myanmar, southern Yunnan, China (6.33%), and Chiang Rai in Thailand (4.0%) (Figs 7 and 8(B)); summer, mainly in Shan (57.47%) in Myanmar, southern Yunnan, China (8.51%), and Chiang Rai in Thailand (6.28%) (Figs 7 and 8(B)); autumn, mainly distributed in Shan (24.13%) in Myanmar, Sơn La (10.33%) in Vietnam, Houaphan (9.57%) in Laos, southern Yunnan, China (7.72%), and Chiang Rai in Thailand (2.93%) (Figs 7 and 8(B)); winter, mainly distributed in Shan in Myanmar (47.22%) and Houaphan in Laos (9.44%) (Figs 7 and 8(B)).

4. DISCUSSION

Southern Yunnan and the Indo‐China Peninsula are in tropical and subtropical climate zones, where suitable host crops for FAW such as maize, rice, wheat, sugarcane, and sorghum are cultivated continuously throughout the year, providing ample food resources and suitable habitats for the pest's population reproduction. Like the southern United States and Mexico, this region possesses the ecological conditions for year‐round reproduction and damage by FAW. ⁴ , ¹⁵ The management of migratory pests emphasizes precise monitoring and source control. ²¹ , ⁵⁰ This study, through long‐term systematic monitoring from 2019 to 2024, reveals the dynamics of FAW migration between southern Yunnan and the Indo‐China Peninsula. It clarifies the seasonal characteristics of population dynamics, divides the migration periods, and identifies periodic characteristics. Furthermore, this study simulates migration paths on peak‐event days through trajectory analysis, quantifying the source origins and landing distributions of the transiting FAW populations.

In the border area between southern Yunnan and Indo‐China Peninsula, large numbers of FAW adults can be trapped year‐round, with an annual average catch of 5795 individuals. There are three typical northward migration periods (spring, summer, autumn) and a winter dispersal period, with no obvious return migration observed. In another study conducted in Ruili on the China–Myanmar border, monitoring results indicated northward migration of populations from India, Bangladesh, and Myanmar entering China in spring/summer and southward return migration exiting China in autumn. ⁵¹ The same phenomenon was also confirmed in Lancang County from another study. ³² However, our study found that migration activities in Jiangcheng during spring, summer, and autumn were all northward entries, with no autumn return migration observed. This may be related to the influence of monsoons and topography. The study monitoring site is located near 100° E, where it is affected by the synergistic influence of both the East Asian monsoon and the South Asian monsoon systems, situated at their convergence zone. ⁵² , ⁵³ In autumn, upper‐air winds above this region may be more conducive to the continuous northward migration of Indo‐China Peninsula populations. Although Lancang (950 m) and Jiangcheng (800 m) are relatively close geographically, differences in altitude and topography might cause differences in the direction of autumn migration activity. Therefore, how FAW adapts to variations in monsoon strength, altitude, and topography at both population and individual levels requires further research. The capture of a substantial number of FAW adults during the winter dispersal period is consistent with monitoring results from Ruili and Lancang, and aligns with the characteristics of dispersal activity in year‐round breeding areas, ³² , ⁵¹ indicating that this region and the Indo‐China Peninsula still maintain a large FAW population base. This conclusion is relatively consistent with findings from monitoring via field surveys and pheromone traps in Xishuangbanna. ⁵⁴ Winter maize crops and non‐preferred hosts in this period may play an important role in the population persistence and maintenance of FAW. Furthermore, FAW first invaded Yunnan, China, on December 11, 2018. ²⁸ The population numbers from January to March 2019 were significantly lower than those in the same period in 2020–2021 (Fig. 2), and a single‐day catch peak of 427 individuals occurred on June 8, 2019, demonstrating the rapid population increase of this invasive pest. This population dynamic characteristic is like the monitoring results in Lancang. ³² This indicates the strong adaptability of FAW to the habitat environment of newly invaded areas. The substantial decline in FAW captures from 2019 to 2022 may be attributed to three potential driving factors. First, spatial shifts in migration corridors caused by inter‐annual wind variations may have led the ‘migration highway’ to bypass our single monitoring site, reducing sampling representativeness. Second, intensified management, including enhanced regional surveillance and precise chemical/biological control since 2019, has significantly lowered the migratory population base. Third, the population has likely transitioned from an invasive expansion phase into an ecological equilibrium, where environmental resistance—such as host–resource competition and the adaptation of indigenous natural enemies—suppresses major surges. Given the single‐site nature of this study, the current data are insufficient to isolate a sole cause, and the decline likely reflects the combined impact of these factors. The migration activity of FAW in the eastern part of the China–Indo‐China Peninsula border area also exhibits a certain generational cycle (4.31 weeks, ~30 days), which highly coincides with the generation duration based on the FAW life table from other studies. ⁵⁵ , ⁵⁶ Based on this, the estimated number of annual generations of FAW in this region can reach 12, which is close to the conclusion drawn by Chen Hui et al. regarding the division of occurrence areas in southern Yunnan. ⁵⁷ Population structure analysis further revealed a significant male bias in the captured migratory populations, which corroborates findings from indoor behavioral studies on differences in phototaxis between sexes. ⁵⁸ The underlying mechanism may be related to differences in light sensitivity caused by sex‐specific expression of visual proteins. ⁵⁹ , ⁶⁰ This finding not only provides a new perspective for explaining the mechanisms of migratory behavior but also has important implications for optimizing light trap monitoring techniques.

Regarding migration pathways, this region consistently faces continuous input of FAW populations from countries in the Indo‐China Peninsula such as Myanmar, Laos, Vietnam, and Thailand. The simulated proportions of foreign sources during the spring, summer, and autumn northward migration periods were 93.7%, 89.4%, and 83.1%, respectively. For the forward trajectories, settlements were mainly concentrated in eastern Yunnan, Guizhou, Guangxi, and a small number in Sichuan, China. Wu et al. inferred that FAW populations from Yunnan in spring and summer mainly migrated to Guizhou, Guangxi, and Sichuan. ²⁷ As this study primarily monitored populations immigrating from abroad, it can be determined that eastern and northern Myanmar, northern Laos, and northern Vietnam in the Indo‐China Peninsula are the main foreign sources of FAW migration into China. The trajectory analysis results based on population dynamics peak days are like the division of FAW migration routes into eastern and western pathways in China by Li. ²⁶ Our results show that simulated paths for different years and seasons indicate that the southern Yunnan border area is an important source for FAW migratory populations on both the eastern and western routes in China, overall showing a southeast–northwest migration trend (Fig. 7). It is noteworthy that the autumn migration paths in 2020 and 2022 were primarily southeast–northwest (Fig. 7), which might be related to anomalous climatic phenomena involving the East Asian and South Asian monsoons in those years. Like other insects, FAW primarily relies on upper‐air wind currents for long‐distance wind‐borne migration, with monsoons being the main external driver. Changes in the East Asian monsoon pattern affect the migration of rice planthoppers in eastern and southern China through alterations in rainfall and wind fields, leading to changes in their migration routes and population sizes. ⁶¹ This trajectory routes and landing areas for FAW corss‐border migration were results of simulation analysis based on typical migration events and atmospheric background dynamic conditions, and it should be noted that trajectory analysis hypothesis of 4.5 m s⁻¹ self‐flight speed and 30° drift compensation angle limited the accuracy of the predicted path. In fact, both of these parameters can be subject to dynamic changes. ⁴⁴ This has not been verified by light trapping data or mark‐recapture‐release data from the receiving area or the source area of the insects, which is one of the limitations of this study. However, the adaptive behavior of FAW to the characteristics of monsoons in China, Southeast Asia, and other Asian regions deserves attention.

The seasonal activities of the two major monsoon systems, the East Asian monsoon and the South Asian monsoon, constitute the meteorological background for the large‐scale regional migration of FAW in Asia. Suitable airflow background conditions drive FAW populations breeding in the Indo‐China Peninsula to cross borders northward into southern China and further promote the migration and dispersal of populations from year‐round breeding areas in southern China to the Yangtze River Basin, Yellow River Basin, and the main agricultural regions of northeast China. ²² , ²³ , ²⁷ Field surveys by Yang et al. indicated that FAW can reproduce and cause damage year‐round in China's south subtropical and tropical regions, with the northern boundary of the year‐round breeding area approximately near the January 10 °C isotherm, ³⁶ around 28° N. ²⁵ Yunnan is in the hilly maize region of southwestern China, where over 90% of its counties plant maize, and FAW occurs and causes severe damage year‐round. In 2022, the FAW infestation area in Yunnan was 2.11 million hectares, making it the primary disaster area for FAW in China. ⁶² It serves both as the first line of defense line against FAW invasion from abroad into China and as a bridgehead for FAW entering China's interior from the southwestern border. In the Indo‐China Peninsula, maize is also a major crop type in various countries, with a sown area of 2.84 million hectares. FAW can occur and cause damage year‐round in this region, making it an important population source for the entire East Asia region. ²⁴ Specifically, in 2019, the damage rate of FAW in southern Myanmar ranged from 62.501% to 97.50%, and the highest plant damage rate in the central‐northern region reached 100%, ⁶³ favoring the formation of a large population base. Suitable climatic conditions and food resources provide favorable conditions for the year‐round reproduction, population persistence, and migration activities of FAW in this region, creating a springboard for the generational migration of Indo‐China Peninsula FAW populations into China. The timing, scale, and destination areas of population emigration from the Indo‐China Peninsula determine the occurrence and severity of FAW infestation in inland China and even East Asia.

Our study confirmed that FAW seems to have a clearly one‐way northward migration pattern in this area, without any typical return migration phenomenon occurring. The migration pattern of fall armyworm in Asia is increasingly complex under the influence of global climate change and seasonal atmospheric circulation. While the Indo‐China Peninsula serves as a primary source, ²² , ²³ , ³² , ⁵¹ the absence of a significant, mass‐scale ‘backward migration’ (north‐to‐south) remains a critical gap in our understanding of the FAW's regional population dynamics. In East Asia, classic migratory pests such as the brown planthopper (Nilaparvata lugens) ⁶⁴ , ⁶⁵ and white‐backed planthopper (Sogatella furcifera) ⁶⁶ , ⁶⁷ , ⁶⁸ exhibit a distinct ‘round‐trip’ strategy: northward expansion driven by the southwest monsoon in spring and summer, followed by a southward return facilitated by the northeast monsoon in autumn. However, the dynamics of FAW migration are far from a simple unidirectional migration, and the assumed absence of a mass‐scale backward migration warrants deeper scrutiny in the context of broader entomological patterns. Furthermore, global climate change is fundamentally reshaping these migration templates. ⁶⁹ , ⁷⁰ Rising mean temperatures have expanded the potential overwintering boundaries northward, effectively shortening the required migration distance from the ‘springboard’ and allowing for earlier seasonal arrivals in temperate zones. ⁷¹ , ⁷² The changing patterns of the east monsoon also lead to significant changes in the abundance and patterns of pest migration in East Asia, such as N. lugens. ⁶¹ , ⁶² This shift, coupled with an increase in the frequency of extreme weather events, enhances the ‘pumping effect’ of the Indo‐China Peninsula population source of FAW. Consequently, the Indo‐China Peninsula is no longer just a static reservoir for FAW populations but a dynamic engine that, under a warming climate, may facilitate more frequent and erratic ‘trans‐border pulses’ of pest migration, challenging current trans‐regional early warning and management frameworks.

In summary, this study systematically monitored the cross‐border migratory population dynamics of FAW in the border area between China and Indo‐China Peninsula, clarified the source origins and settlement distributions of the transiting migratory populations, and provides strong support for establishing a regional cooperative prevention and control system based on monitoring and early warning for migratory pests. It holds significant practical value for promoting cross‐border joint management between China and Southeast Asian countries. ⁶² Governments in Southeast Asia, based on smallholder farming models, advocate for integrated pest management (IPM) oriented approaches to manage FAW damage. ⁷³ Based on the above conclusions, combined with management needs, we propose source control strategies for FAW. The tropical and subtropical regions of southern China and the Indo‐China Peninsula are year‐round occurrence areas for FAW and are the sources for its generational migration into the Yangtze River Basin, Yellow River Basin, and northeast regions. Therefore, monitoring of population dynamics, including occurrence period, density, and area (especially for the overwintering generation), should be strengthened in this region. An integrated domestic and international monitoring and early warning platform for FAW should be established to improve the accuracy of predicting FAW migration dynamics. Prevention and control technology integration focusing on adult trapping should be promoted. In major source areas, key migration corridors, and potential settlement areas, multiple physical and chemical attraction methods such as high‐altitude searchlights, black light traps, food attractants, and sex pheromone traps should be comprehensively utilized to trap emigrating and immigrating FAW adult populations. International cooperation should be strengthened, enhancing coordinated monitoring, early warning, and pest information sharing with relevant Southeast Asian countries to intercept transiting migratory FAW populations, reduce the occurrence of the next generation, block regional migration routes, and lower the regional disaster risk of FAW. A comprehensive IPM strategy for FAW should be established. Research and development of new pest control technologies should be enhanced. In year‐round breeding areas, technologies such as Bt maize planting and the insect sterile technique could be utilized to implement source control of migratory populations, reducing the size of emigrating populations from year‐round occurrence areas.

5. CONCLUSIONS

The cross‐border migration activities of FAW in the border area between southern Yunnan and the Indo‐China Peninsula are mainly concentrated in the spring, summer, and autumn northward migration periods, specifically 4 March to 20 May, 20 May to 15 July, and 23 September to 28 October. In terms of migratory population quantity dynamics, month was the dominant factor, with small interannual variation. The cross‐border migratory population dynamics exhibited a generational cycle (4.31 weeks). Myanmar, Laos, and Vietnam in the Indo‐China Peninsula are the main source areas for FAW populations migrating across the border into China during the migration periods.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

WKM designed the experiments. ZSY, LDZ, CS, LCY, and FYW conducted the experiment. ZSY, CXZ, and YXM analyzed the data. ZSY wrote the manuscript. All authors revised and approved the manuscript.

Supporting information

Supporting Information S1. Distribution of forward trajectory landing points for 12 h during migration peak day of FAW at Jiangcheng Station from 2019 to 2023.

Supporting Information S2. Distribution of backward trajectory landing points for 12 h during migration peak day of FAW at Jiangcheng Station from 2019 to 2023.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.