Research model and software
CLIMEX model
FAW growth and development are primarily related to climate conditions, especially temperature patterns17. The current study used CLIMEX (version 4)42, a semi-mechanistic niche modeling platform, to project FAW distribution in relation to climate. The model parameters that describe the species’ response to climate were overlaid onto FAW occurrence data and climate data to project the species’ potential global distribution. Briefly, the annual growth index (GI) was used to describe the potential for FAW population growth during favorable climatic conditions, while stress indices (SI: cold, wet, hot, and dry) and interaction stresses (SX: hot-dry, hot-wet, cold-dry, and cold-wet) (Table 1) were applied to describe the probability that FAW populations could survive unfavorable conditions. The Ecoclimatic index (EI) was derived from a combination of GI, SI, and SX indices to provide an overall annual index of climatic suitability on a scale of 0–10042. An EI value of 0 indicates that the location is not suitable for the long-term survival of the species, whereas an EI value of 100 indicates maximum climatic suitability comparable to conditions in incubators. EI values of more than 30 indicate the optimal climate for a species. In this study, the climatic suitability was classified into four arbitrary categories; unsuitable for EI = 0, marginal for 0 < EI ≤ 10, suitable for 10 < EI ≤ 30, and optimal for 30 < EI ≤ 10042.
ArcGIS software
The ArcGIS software 10.8 (US Environment Systems Research Institute—ESRI, Redlands, CA, USA, https://desktop.arcgis.com/en/arcmap/) was used to visualize the result obtained from the CLIMEX analysis and calculate the areas under various EI categories for the species.
Data collection
Fall armyworm occurrence data
FAW occurrence observations (n = 304) within its native range were obtained from the Global Biodiversity Information Facility (GBIF, www.gbif.org), PestWatch (www.pestwatch.psu.edu), Butterflies and Moths of North America (BAMONA, www.butterfliesandmoths.org), and literature resources (Supplementary Table S1). Real-time occurrence records (n = 1186) in six East African countries (Burundi, Ethiopia, Kenya, Rwanda, Tanzania, and Uganda) were collected from the Community Based FAW monitoring forecasting and Early Warning (CBFAMFEW) system44,45. The CBFAMFEW system relies on pheromone traps, field scouting and mobile applications for field data collection. Briefly, FAW pheromone traps were established in five districts in each country, and, in each district, 10 villages were sampled under the coordination of two community focal persons. Thus collected FAW occurrence records were validated by national FAW focal persons and published in FAMEWS (Fall Armyworm Monitoring and Early Warning System) global platform44,45.
The additional records of FAW occurrence were obtained from PlantVillage FAMEWS survey. PlantVillage is a public good platform that integrates AI, satellites, cloud computing, and local networks to help smallholder farmers adapt to climate changes and increased pest pressure.
Overall, a total of 13,460 FAW global distribution records with either pheromone trapping (with confirmed FAW moth) or field scouting (with confirmed FAW larvae) data were collected. To make data visualization and manipulation easier, FAW occurrence records were spatially filtered to retain a single record in each 10-arc minute (~ 18 km) grid. This resulted in 2968 records (Africa—2591, Asia—150, North America—171 and South America—56) used for further analysis. The distribution records used in this study are shown in Fig. 1.
FAW presence confirmed locations in the world. Triangles represent FAW occurrence records from its native range—black triangles show areas that support seasonal population growth and red triangles show areas that support year-round population establishment. Blue circles show FAW occurrence records from its invasive range. ArcMap 10.8 (https://desktop.arcgis.com/en/arcmap/).
Climate data
The current suitability of FAW was modeled with the CliMond historical dataset interpolated at 10-arc minute (available at https://www.climond.org)46. This dataset consisted of long-term averages centered at the year 1975 for maximum and minimum temperatures, precipitation and relative humidity at 09:00 and 15:00 h. Future climatic suitability for FAW was projected using the 10-arc minutes gridded spatial resolution climate data for 2030, 2050, and 2080 retrieved from the CliMond (Version 2) in CLIMEX format46. The future climate projections used in this study were based on two global climate models (GCM), CSIRO-Mk3.0 GCM developed by CSIRO Atmospheric Research, Australia47 and MIROC-H GCM developed by Center for Climate Research, Japan. These were run with the A1B SRES (Special Report on Emission Scenarios) emission scenarios. The SRES A1B was chosen with the assumption that, in the future, the use of fossil intensive and non-fossil energy sources will be balanced. For an A1B emission scenario, the CSIRO-Mk3.0 and MIROCH-H GCMs predict a rise in temperature of 2.11 °C and 4.31 °C, respectively, by the end of twenty-first century46,48. Similarly, these two GCMs predict different rainfall patterns49.
The Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5) released a new family of emission scenario called Representative Concentration Pathway (RCP), to replace the SRES family. This consists of four-climate change scenarios- RCP2.6, RCP4.5, RCP6 and RCP8.5. The closest similar RCP scenario to SRES A1B is RCP 6.0, which represents an intermediate emission scenario50. The temperature increase at the end of the twenty-first century for RCP6.0 scenario is projected to be 2.2 °C with a range of 1.4–3.1 °C, while for SRES A1B scenario, it is projected to be 2.8 °C with a range of 1.7–4.451. Furthermore, the CO2 concentration by the end of the century for RCP6.0 is expected to reach 670 ppm, just below A1B (703 ppm)50.
Irrigation data
The information on global irrigation areas was derived from the Food and Agriculture Organization of the United Nations (http://www.fao.org/aquastat/en/geospatial-information/global-maps-irrigated-areas/).
Host crops data
Geographic distribution of maize and sorghum, two major host crops of FAW, were obtained from the EarthStat database (http://www.earthstat.org/) created by Monfreda et al.52.
Model fitting
Using the “Compare Locations” modules in CLIMEX, the values of FAW parameters (Table 1) were determined (1) from previous studies’ findings on requirements for growth and development of FAW53,54,55,56, and (2) by fitting the projected distributions to the occurrence records of FAW in its native range and the recently invaded range in East Africa. We initially adopted parameter values from a previously conducted CLIMEX study on FAW35,38. Then, parameter values were determined through an iterative process to fit the simulated CLIMEX results to the known distribution of FAW in the world in 2020. FAW occurrence records in North America include reports of transient or migrant populations (Fig. 1, Supplementary Table S1). Therefore, during the model fitting process, we confirmed that the areas with migrant populations have a positive annual growth index (GI > 0) and an unsuitable eco-climatic index (i.e. EI = 0)57,58. The fitted values of CLIMEX parameters were then validated by comparing EI distribution of FAW with independent sets of FAW occurrence data collected in eight African countries (Burkina Faso, Ghana, Madagascar, Malawi, Mozambique, Liberia, Sudan, and Zambia), Asia, and Australia. The process of parameter fitting was repeated until the reliability and consistency of the model projections were established. The parameter values used in this study are listed in Table 1, and details are provided in the following sections.
Growth indices
Temperature index (TI)
The lower (DV1) and upper (DV2) optimal temperatures for FAW population growth were left unchanged at 25 °C and 30 °C, respectively35. These values are supported by multiple publications53,54,55,56. FAW reared at 25 °C constant temperature are less likely to emerge deformed55, and the adult moths have the highest adult longevity and fecundity53. A recent study by Du Plessis et al.17 identified 30 °C as the upper optimal temperature for FAW growth and development. The development rate of FAW increases linearly with increasing temperature from 18 to 30 °C, but declines when temperature increases above 30 °C17.
The limiting low temperature (DV0) was also kept the same at 12 °C35,38. FAW lacks any diapause mechanisms and overwinters only in warm and humid areas; hence, FAW cannot tolerate freezing temperature14. Wood et al.59 reported that a temperature above 10 °C is required for pupal eclosion. The pupae held at 10 °C live for 50–62 days but do not eclose55,59.
The limiting high temperature (DV3) was set at 36 °C because more than 50% of FAW reared at 35 °C or above exhibit physical deformity and die within 24 h of emergence55.
Moisture index (MI)
Not much is known about the relationship between soil moisture and the FAW lifecycle, so we adopted the species parameters from previous studies35,38, with some assumptions. The lower soil moisture threshold (SM0) was kept the same at 0.15 to allow FAW invasion in semi-arid areas in Africa. Silvain & Ti-A-Hing60 reported higher FAW populations (both adult moths and larvae) during rainy seasons than in the dry seasons. At any time, the larval population is affected by the amount of rainfall that was experienced three weeks earlier60. Although a heavy downpour reduces adult emergence by trapping moths in their pupation tunnel61, FAW larvae can tolerate substantial waterlogging conditions35. Therefore, the upper soil moisture threshold (SM3) was set to 2. Reducing the value of SM3 from 2.5 to 2 had no effect on defining the potential range of FAW. The lower (SM1) and upper (SM2) limits for optimal growth were left unchanged at 0.8 and 1.5, respectively35. The upper optimal soil moisture (SM2) value allows the persistence of FAW in tropical areas that experience high rainfall, such as Central America. Here, the SM value 0 indicates no soil moisture; SM 0.5 indicates soil moisture is 50% of soil water holding capacity, and SM > 1 indicates a run-off situation.
Stress indices (SI)
CLIMEX mainly uses four stress indices (SI: heat, cold, wet, and dry) to determine the species’ geographical distribution. Species population growth occurs between the temperature parameters DV0 and DV3, and moisture parameters SM0 and SM3 (Table 1). Values outside of this range result in negative population growth.
Cold stress (CS)
The cold stress temperature threshold (TTCS) and cold stress accumulation rate (THCS) was decreased to 8 °C and − 0.005, respectively, to fit the FAW distribution in the Rio Grande valley Texas (overwintering site in North America), Mediterranean coast in North Africa and the Yunnan province (first FAW-invaded province) in China.
Heat stress (HS)
The heat stress temperature threshold (TTHS) for HS was kept the same at 39ºC to allow pest development in western African countries35. Heat stress accumulation rate (THHS) was set to 0.0025 week−1 to allow the pest development in Nile River basins in Egypt and irrigated areas in Yemen and Pakistan.
Dry stress (DS)
Dry stress indices were the same as those of the existing model of Du Plessis et al. 35. Soil moisture dry stress threshold (SMDS) was set to the same value as SM0, i.e., 0.15, and dry stress rate (HDS) was set to 0.005.
Wet stress (WS)
Soil moisture threshold for wet stress (SMWS) was set at the same value as SM3 in our model. The wet stress accumulation rate (HWS) was increased to 0.01 week−1 to exclude extremely wet areas from being suitable. This change does not limit the FAW persistence in areas with known FAW distribution but reduces the modeled risk in extremely wet areas.
Effective degree-days (PDD)
In CLIMEX, the PDD parameter indicates the degree-day above the minimum base temperature (DV0) necessary for species to complete one generation. Hogg et al.62 estimated PDD for FAW at 346.2 degree-days with base temperature of 13.8 °C while Du Plessis et al.17 calculated PDD value of 390 degree-days with base temperature 12.57 °C. In the current study, the base temperature was set at 12 °C. To get the same number of generations per year, PDD value was increased to 400-degree days.
Irrigation
FAW was reported in dry areas in North Africa, Pakistan and Yemen. These areas did not fall within climatically suitable areas projected under rainfed conditions. These dry area records might reflect FAW populations able to persist only when irrigation is applied to sustain the crop. To simulate the effect of irrigation in FAW distribution, two irrigation scenarios were taken into account. First, the CLIMEX model for FAW was run using 2.5 mm day−1 as top-up irrigation throughout the year (Irrigation scenario I) to capture the risk posed by FAW in areas where cropping should be sustained by irrigation35,42,57. This top-up irrigation was added only when the weekly rainfall was less than 25 mm. Second, a composite FAW-risk map (Irrigation scenario II) was developed by combining the rainfed and irrigation scenario I results; EI from irrigation scenario I was mapped in areas under irrigation reported by Siebert et al.63 and EI for rainfed scenario was mapped elsewhere.
Model performance
FAW occurrence records were overlaid on the projected layer surfaces to evaluate the model performance. EI values of the pixel where each FAW occurrence records lie were extracted from the projected raster layers. A histogram and a normal distribution curve were fitted on the projected EI values of each dataset under the current and projected future climates. Descriptive statistics were also generated from the extracted EI values to recognize the projected models that captured the presence records better. This analysis was used to measure and confirm the ability of the developed model to predict the FAW habitat areas successfully.
Parameter sensitivity and model uncertainty analysis
CLIMEX model is a robust tool to assess the risk of pest invasion and establishment, but it includes several sources of uncertainty that need to be communicated to risk assessors and decision makers. The CLIMEX Version 4 has the parameter sensitivity and model uncertainty analysis tool available to evaluate the model. The sensitivity analysis identifies the degree to which each species parameter affects the projected areas of climatic suitability, whereas uncertainty analysis reflects the ability of the model to accurately predict the climatic suitability for a species. For the parameter sensitivity and uncertainty analysis, the default model parameters were run using the historic climate (CM10 1975H V1.2) under the rainfed scenario. Both parameter sensitivity and model uncertainty analysis were run for the entire world.
Potential overlap between FAW and its major host maize
The projected suitability areas for FAW was overlaid on the projected distribution of its major host maize, to assess the potential co-occurrence of FAW and maize under the current and future climates. We performed the CLIMEX suitability analysis for maize using the maize-CLIMEX parameters from Ramirez-Cabral et al.43 (Supplementary Table S2). The maize-CLIMEX model developed by Ramirez-Cabral et al.43 did not include irrigation. In the current study, the maize-CLIMEX model was updated to include irrigation. Similar to FAW projections, future climatic suitability for maize distribution was projected using the 10-arc minute gridded spatial resolution climate data for 2030, 2050, and 2080, assuming A1B emission scenario. EI maps for maize (Supplementary Fig. S1) were created considering Irrigation scenario-II (i.e., EI values from the maize-CLIMEX model with irrigation was used in the irrigated areas and EI values from the maize-CLIMEX model without irrigation was applied elsewhere). Areas with EI value above 10 (i.e., suitable and optimal categories) were used to calculate the potential area overlap between FAW and maize distributions. Areas with EI value less than 10 (i.e., unsuitable and marginal categories) do not support or marginally support the species distribution. Therefore, those were excluded from the analysis to increase the comparability between the pest and its host maize.
We considered only maize for the pest-host overlap analysis because CLIMEX suitability analysis for maize is already published43. No such analysis is available for other host crops.
Source: Ecology - nature.com
