Potential distribution of fall armyworm in Africa and beyond, considering climate change and irrigation patterns

The fall armyworm, Spodoptera frugiperda (FAW), first invaded Africa in 2016 and has since become established in many areas across the continent where it poses a serious threat to food and nutrition security. We re-parameterized the existing CLIMEX model to assess the FAW global invasion threat, emphasizing the risk of transient and permanent population establishment in Africa under current and projected future climates, considering irrigation patterns. FAW can establish itself in almost all countries in eastern and central Africa and a large part of western Africa under the current climate. Climatic barriers, such as heat and dry stresses, may limit the spread of FAW to North and South Africa. Future projections suggest that FAW invasive range will retract from both northern and southern regions towards the equator. However, a large area in eastern and central Africa is projected to have an optimal climate for FAW persistence. These areas will serve as FAW ‘hotspots’ from where it may migrate to the north and south during favorable seasons and then pose an economic threat. Our projections can be used to identify countries at risk for permanent and transient FAW-population establishment and inform timely integrated pest management interventions under present and future climate in Africa.

www.nature.com/scientificreports/ FAW monitoring forecasting and Early Warning (CBFAMFEW) system 44, 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 platform 44, 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.
Climate data. The current suitability of FAW was modeled with the CliMond historical dataset interpolated at 10-arc minute (available at https:// www. climo nd. 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 format 46 . 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, Australia 47 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 century 46,48 . Similarly, these two GCMs predict different rainfall patterns 49 .
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 scenario 50 . The temperature increase at the Table 1. CLIMEX parameter values used for modeling the distribution and invasion risk of FAW (Spodoptera frugiperda). Changes made to the Du Plessis et al. 35 parameter values are given in bold.  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/ aquas tat/ en/ geosp atial-infor mation/ global-maps-irrig atedareas/).
Host crops data. Geographic distribution of maize and sorghum, two major host crops of FAW, were obtained from the EarthStat database (http:// www. earth stat. 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 FAW [53][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 FAW 35,38 . 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 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 eclose 55,59 . www.nature.com/scientificreports/ 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 emergence 55 .
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 studies 35,38 , 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-Hing 60 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 earlier 60 . Although a heavy downpour reduces adult emergence by trapping moths in their pupation tunnel 61 , FAW larvae can tolerate substantial waterlogging conditions 35 . 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, respectively 35 . 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 countries 35 . 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 irrigation 35,42,57 . 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 www.nature.com/scientificreports/ 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 published 43 . No such analysis is available for other host crops.

Results
FAW global potential geographical distribution under the current climate. The potential projected global geographical distribution of FAW, ignoring the distribution of crop hosts and non-climatic barriers, corresponded well with the known present distribution of this species. The result shows that most of the world's tropical and subtropical climates are climatically suitable for year-round FAW establishment (with EI > 30) (Fig. 2a). This area increases substantially when looking at seasonal FAW population growth (with GI > 0): our www.nature.com/scientificreports/ model predicts that areas with humid-continental and Mediterranean climates have the potential to support FAW population growth during sometime of the year (Fig. 2b). The model projected a notable difference in the potential distribution of FAW between irrigation scenario-II and the rainfed scenario, especially in North Africa, the Middle East and Australia (Fig. 2c,d). Many parts of North Africa and the Middle East become climatically suitable for FAW population establishment, when irrigation is available in these areas.
Native range. Potential FAW distribution in the Americas, its native range, is consistent with its known geographical range. The model confirmed that FAW over-wintering locations in the Americas are climatically suitable (with EI > 30) for the persistence of the FAW populations (Fig. 2a,c) under both rainfed condition and irrigation scenario-II. Furthermore, our model projected the expansion of the FAW geographical range to the eastern parts of North America with humid-continental and humid-subtropical climates during favorable seasons (with GI > 0, EI = 0) (Fig. 2b,d). In South America, the entire continent, except the Andes region and Argentina, has an optimal climate for FAW establishments.
Non-native range. Almost all areas in Africa, the non-native range, with known presence records were projected to be climatically suitable (with EI > 30) for FAW population persistence under rainfed conditions (Fig. 2a). A few occurrence records in North Africa did not fall into the region projected as suitable by the model. However, irrigation is used to sustain agriculture in those areas 63 (Supplementary Fig. S2). Most of these locations become suitable when the irrigation scenario II is included in the model (Fig. 2c). About 27% of Africa's total area is projected to have an optimal climate for FAW population establishment, under irrigation scenario II (Fig. 2c). In eastern and central Africa, all countries except Somalia, Eritrea and Chad were projected to have suitable to optimal climatic for FAW population establishment and are under high risk. In western Africa, Guinea, Sierra Leone, Liberia, Côte d'Ivoire, Ghana, Togo, Benin, Nigeria, Sao Tome and Principe were also projected to potentially support year-round FAW populations. North African countries with tropical and subtropical desert climates were projected to be most unsuitable, with only irrigated areas and areas along the Mediterranean coast projected to have marginally suitable climates for FAW persistence. Heat and dry stresses would limit population establishment and growth in North African countries under the current climate (Fig. 3). Similarly, the Indian Ocean coast in South Africa is the only area in southern Africa that has an optimal environment for FAW persistence. Other parts in this region were not projected to support FAW population establishment due to dry and cold stress (Fig. 3b,d). However, availability of irrigation would render these dry areas suitable for FAW invasion and population growth (Supplementary Fig. S3). www.nature.com/scientificreports/ Although Europe has an unsuitable climate for FAW year-round establishment (Fig. 2a,c), a few areas with humid-continental and Mediterranean climates are potentially suitable to support migrant FAW populations (with GI > 10) for a few generations per year (Fig. 2b,d). The risk of FAW invasion in Europe has become increasingly possible considering that FAW presence has recently been reported for Egypt and Mauritania. Furthermore, since extensive areas of host crop plants are readily available in southern Mediterranean Europe ( Supplementary  Fig. S4), these areas have potential to support transient populations under the current climate. Small regions of southern Italy, Spain, and Portugal have suitable to marginal climates for FAW year-round establishment, and these areas could be expected to be invaded first, as FAW has already reached the Nile Valley in Sudan, Egypt, and Mauritania. Yet, low winter temperatures in the majority of Europe would limit population growth and persistent establishment under the current climate (Fig. 3b).
FAW invasion has been confirmed in many countries in Asia in early 2018 [20][21][22][23] . There is a considerable chance that FAW will extend beyond its current range in Asia. Based on historical climate data, South East Asian and South Asian countries have optimal climate conditions for FAW population establishment (Fig. 2a,c). In addition, areas in southeast China with a humid subtropical climate are also optimal for FAW population establishment. When weather is favorable, FAW could migrate seasonally to maize-growing areas with unsuitable or marginal climates for seasonal population establishment (Fig. 2b,d).
A large part of Australia is unsuitable for FAW persistence because it has a tropical and subtropical desert climate. However, areas with tropical wet and dry climate are projected to range from suitable to optimal category for FAW persistence (Fig. 2a,c), and regions with humid-subtropical climate are suitable and have the potential to support migrant FAW populations (Fig. 2b,d).   Table 2, Fig. 4, and Supplementary Fig. S7). For 2030, both CSIRO-Mk3.0 and MIROC-H GCMs projected a reduction of approximately 2% in areas with optimal climatic suitability, with a shift from optimal to suitable or marginal climate. All currently suitable countries in West Africa were projected to gradually retract from the optimal range for FAW by 2030, and many countries lose it by 2080 due to heat and dry stress (Figs. 4 and 5). Areas of optimal suitability are also projected to decrease in countries, such as South Sudan and Mozambique. For the remaining countries with current climatic suitability, the projected changes in climate for the next 10 years would not affect their FAW suitability. Some areas in Angola, Zambia, Zimbabwe, and Tanzania currently classified as suitable category are projected to improve to optimal in the warmer future climates.  Under the climate change scenario, the current potential range of FAW is expected to decrease mainly due to heat and dry stress in Africa (Fig. 5). Hot and dry summers will create stressful conditions that prevent FAW from completing its life cycle and persisting year-round in areas that are now suitable for year-round persistence. However, there is still a chance that FAW escapes the heat and dry stress during hot summers by migrating to Central and Eastern Africa and reinvading the marginal or unsuitable areas during the warm winter season each year.
Those African countries that lie along the equator will likely support the long-term persistence of FAW populations and experience a stable risk throughout the year (Fig. 4). These areas include parts of Ethiopia, Kenya, Uganda, Rwanda, Burundi, Tanzania, Zimbabwe, Zambia, Angola, Democratic Republic of Congo, Republic of Congo, Gabon, Equatorial Guinea, and Madagascar. These areas could also serve as FAW hot spots from where FAW migrate each year to re-infest a succession of cropping areas in other regions with marginal and suitable growth index (GI > 0) but unsuitable eco-climatic index (i.e. EI = 0) (Fig. 4).

Potential overlap between FAW and its major host plant maize.
Under the current climate conditions, 10.1 million km 2 (i.e., 33%) of the land area in Africa has suitable to optimal climate for both maize and FAW (Table 3). Under the projected future climates, the area with co-occurrence potential would decrease over time (Table 3, Supplementary Fig. S8). However, by the end of 2080 under CSIRO-Mk3.0 and MIROC-H GCMs, 13% to 16% of the area in Africa, respectively, would still support both FAW and its host plant maize. These areas of co-occurrence would have high potential to serve as a breeding spot or hot spot, like the southern part of Florida and Texas in the USA 28 .
Parameter sensitivity and performance analysis. The modeled potential distribution of FAW using the recent CLIMEX model is highly sensitive to change in the cold stress temperature threshold (TTCS) and limiting low moisture (SM0), with 1 and 0.8% impact, respectively (Supplementary Table S3). Other parameters have less than 0.5% sensitivity to the projected potential range, suggesting that TTCS and SM0 were the fundamental parameters for fitting the model simulation into the actual distribution.
The model uncertainty analysis indicates that there is a greater degree of geographical uncertainty about the ability of FAW to persist in drier areas (e.g. Sub-saharan Africa, Australia and the Middle East) than colder areas (e.g. USA and Mexico, Southern China) (Fig. 6).
The high proportion (0.98) of the occurrence records in the validation area falls within the area of projected suitable to optimal categories for FAW establishment (Supplementary Table S4), thereby reflecting the reliability, accuracy and robustness of the model. The CLIMEX model under current climate had an EI mean of 38.8 and EI median of 37.1, suggesting that a considerable proportion of the extracted EI values were within the optimal suitability threshold of 30-100 ( Supplementary Fig. S9). The model under the current conditions had the highest EI mean of 38.8 and EI median of 37.1, while the CSIRO 2080 model had the lowest EI mean of 30.9 and EI median of 31.8. This indicates that the accuracy of models reduces with future projections. This implies that there is a possibility of a niche range shift of FAW based on the climate suitability in future. www.nature.com/scientificreports/

Discussion
The global geographical distribution of FAW projected with the CLIMEX model closely matched the current distribution of FAW. The model set out to predict areas where FAW populations could establish themselves permanently; that is, their presence is year-round with multiple generations. Based on the CLIMEX projections under present climate conditions, the world's tropical and subtropical climates are suitable for the year-round establishment of FAW, whereas temperate climates are at risk of seasonal invasions. In Africa, a large part of east, west and central Africa (i.e., 57.2% of the total area) is currently suitable for FAW population establishment.
In the future, areas of climatic suitability for FAW establishment are expected to gradually decrease over time mainly due to heat and dry stress. However, FAW may still survive and become established in significant areas in Africa (i.e., 47.2% and 53% of the total area under the CSIRO-Mk3.0 and MIROC-H GCMs, respectively) by the end of 2080. If certain conditions are met, these established persistent populations, could, in turn, serve as a source of seasonal invasions and migrate into less favorable climatic regions. Therefore, it is likely that FAW will establish permanent populations and cause substantial damage and economic losses to certain crops production every year in Africa unless FAW populations can be effectively managed.
Our CLIMEX model projects that, under the current climate, there is an increased risk of global FAW invasion and establishment. Outside of its native range in North and South America, the current climatic conditions in many parts of Africa, South and South-East Asia, southeastern parts of China, the north coast of Australia, and a few pockets in Europe are favorable for FAW invasion and establishment. Until 2016, FAW was confined to areas with a subtropical wet and dry climate in the Americas. After invading Africa in 2016, FAW has spread rapidly throughout vast regions of Africa that have a climate similar to its native range. By 2020, FAW has further spread to the Middle East, South Asia, South East Asia, and Australia 18,24 . While FAW has not yet invaded Europe, our model identifies several pockets in Europe that are climatically suitable for FAW invasion, suggesting that FAW has not reached its full potential range and its range is still expanding.
The model suggests that extreme temperatures, cold winters and hot summers, as well as limited soil moisture, are the most critical factors constraining the survival of FAW. Currently, cold stress limits FAW's potential to permanently establish in northern North America, southern parts of South America, the northern part of Asia, and Europe. Presently, the permanent establishment of FAW in these regions is not possible because FAW does not diapause 14 . However, if global temperature increases as projected, some areas currently classified as climatically unsuitable may become suitable for FAW invasion and establishment in the future. Expansion towards the north has already been reported for several invasive insect species under climate change 64,65 .
Similarly, dry stress limits FAW's establishment in North African countries, the Middle East, and a large part of Australia. In the future, these areas are projected to experience a reduction in total annual rainfall and would not be expected to support FAW populations. However, irrigated crop fields in these areas could support seasonal FAW populations during warm winters.
In Africa, central, eastern, and western African countries with sub-tropical wet and dry climates offer optimal conditions for FAW population establishment and are under high risk. These areas could serve as overwintering areas and sources of seasonal invasions to other parts of Africa, in the same way, that southern Florida, Texas, www.nature.com/scientificreports/ and Mexico serve in North America 28 . The projected FAW overwintering areas in the present study are consistent with those in previous studies 26,35 . Further, the projected climatic suitability for FAW also closely matches the projected distribution of maize, its preferred host under current and future climates (see Ramirez-Cabral et al. 43 for global distribution maps of Zea mays), which further enhances the probability of FAW to invade and even establish persistent populations in the areas projected by our model. In Africa, the northern part of FAW's current distribution is currently limited by extreme heat and dry stress. The permanent establishment of FAW in this region is unlikely, as FAW cannot tolerate extreme heat [53][54][55] . However, microclimates near maize growing areas in the Nile River basins and irrigated zones could provide channels of suitable habitat 66 and support the establishment of FAW permanent populations. In addition, a small portion of the Mediterranean coast in North Africa is suitable for FAW persistence but has not yet been invaded. The arrival of FAW in those areas by its own means is unlikely because FAW has to migrate across the Sahara and face these harsh desert conditions. Yet, since FAW has already invaded the humid irrigated zones of Aswan Governorate in South Egypt 19 , we should not rule out the possibility of seasonal invasion of the north through natural migration. Migrant FAW moths could use maize fields along the Nile River basins as 'stepping-stones' before finding a suitable location along the North African coast. The flight biology of FAW in North America suggests that it can complete such a long-distance migration 28 . The establishment of significant populations along the North African coast and the Nile River basin will increase the threat to North Africa and Europe through seasonal migration 67 . The risk posed by migrant FAW populations north of the Sahara should not be ignored, since a similar scenario occurs in North America where the climate also does not support FAW establishment, yet crops are still devastated by seasonal invasions of FAW 28 .
Furthermore, several pest species have crossed unfavorable geographic barriers by 'piggybacking' on human transit routes. Therefore, intra-continental travel and trade increase the risk of FAW introduction to the North African coast and Europe 26,27 . Rwomushana et al. 4 reported several interceptions of FAW in consignments from Africa. Therefore, to prevent travel-assisted introduction of the FAW in these places, close monitoring of the FAW invasion and activity in North Africa as well as adopting strict phytosanitary measures are and will continue to be necessary. In addition, pest species, including FAW, can move long distances aided by meteorological phenomena such as storm fronts 68 . While the present study did not consider it, an analysis of synoptic and mesoscale wind and storm patterns may be useful, especially since these patterns are projected to change over time 69,70 .
The climate around the globe is changing, and the distribution of areas climatically favorable to FAW persistence will change accordingly. Our results suggest that under future climate scenarios, the projected distribution of FAW in Africa will contract in both northern and southern regions towards the equator. These findings are in line with the results of a previous study, which also projected a decrease in climatically suitable areas in the Americas under the climate change scenario 38 . These range contraction projections are similar to those made for its preferred host, Z. mays 43 (also see Supplementary Fig. S4). The projected decrease in suitable climate in the northern and southern range in Africa is due to a significant increase in heat and dry stress caused by increased temperatures. During hot and dry summers, FAW pupae do not generally hatch, and if they do, their ability to fly and search for mating partners is compromised 55,56 . Similarly, exposure to high temperatures for extended periods also significantly reduces fecundity and increases mortality in FAW 53 , leading to population decline in formerly suitable areas. However, these areas could be re-infested every year from migrant populations, thereby resulting in potential damage.
Our model projected a high risk of the permanent establishment of FAW in Africa. The year-round availability of host plants and warm and moist winters is optimal for FAW populations' long-term persistence in Africa. Although hot summer temperatures are likely to exclude the FAW from the warm habitat in humidtropics in Africa, the overwintering populations in sub-tropical wet and dry climate areas could re-infest these areas every year during favorable seasons. Records of FAW occurrence in hotter and drier areas than those in its native regions suggest that FAW is extremely heat-tolerant, more so than we thought in the past. Since there are no detailed studies on FAW biology and ecology in Africa, we used temperature-dependent life history data from the studies conducted in the Americas during the 1960s-1990s to predict infestation areas with suitable climatic conditions. However, FAW may have adapted and become more heat tolerant than previously reported. Therefore, any change in the insect's ability to survive in extreme climate conditions (in this case, tolerance to dry and heat stress) would expand the projected range of distribution 71,72 . Hence, further studies are necessary to assess FAW's biological and ecological adaptation to the African continent using natural populations sourced from different climatic niches 73 .
The large area overlap between the projected geographical distributions of FAW and its major host maize also indicates the substantial ecological and economical risk posed by this species. Under current climate, 33% of the total land area in Africa has suitable to optimal climate for FAW persistence and maize cultivation. Since this species feeds on over 350 plant species, the area overlap between FAW and it's host crops could be greater than 33%. This FAW-maize overlap area is expected to decrease over the next 50-60 years, mainly due to loss of climatic suitability for maize cultivation. FAW could still persist in areas outside the FAW-maize overlap areas if other host crops are present. Our analysis only shows the potential area overlap between FAW and host crop maize, assuming that maize is grown in all the areas climatically suitable for maize cultivation. Therefore, actual areas of maize and other host crops cultivation should be considered in the future studies.
The geographic distribution of FAW is determined by biotic interactions, abiotic factors, and its active or passive movement 74 . This study focuses on abiotic factors for temperature-dependent life history events to map climatically suitable areas for FAW because abiotic factors such as climatic variables are considered prime factors affecting species distribution at the continental and global scales 75 . Therefore, we examined the influence of climate change only on FAW's range of invasion and establishment. At these large scales, the results from the bioclimatic models alone are sufficient to suggest which regions are at high risk for pest invasion and establishment under future climates 75  www.nature.com/scientificreports/ Biotic factors should be incorporated in the bioclimatic model results to provide a more refined understanding of the species distribution under changing climates 74 . FAW requires a suitable place to live and complete its lifecycle over several generations. FAW also requires an adequate supply of host plants in synchrony with its lifecycle, without which it cannot persist in that environment. Furthermore, FAW's interactions with other species, such as natural enemies and competitors in a particular region, affect species distribution. Unfortunately, at this point combining all these factors in the bioclimatic model would increase the complexity and is not currently possible 74,75 .
The projected distribution of the FAW in Africa has indicated that FAW will severely impact agriculture over the next several decades. Insect pest management will be more challenging in a changing climate since climatic factors affect the timing of pest infestation, host preference, the efficacy of chemical and biological measures of control, and their utilization within integrated pest management strategies 76 . In this case, modeling habitat suitability for the pest under current and future climatic conditions provides robust tools and recommendations across multiple stakeholder levels and geographical scales. The current study will guide farmers, extension agents, researchers, policymakers, and public and private sectors to develop risk assessment protocols and climate-smart pest management strategies to prevent or reduce the economic loss due to FAW.

Conclusion
The FAW poses a considerable threat to farmers worldwide. Projection of the pest distribution and its potential to establish in a targeted geographical location is crucial in enhancing preparedness, particularly selecting appropriate pest management control. Since FAW performance and survival is limited by temperature and humidity, we constructed a model that predicts areas suitable for FAW occurrence under current and future climate conditions. Here we considered two fundamentally different scenarios: areas where FAW persist year-round with multiple generations per year, and areas where FAW invade seasonally and each invasion starts from persistent populations. Under current climatic conditions, FAW has not yet reached all areas where it could potentially establish year-round populations, and therefore also not all areas that it can occasionally or seasonally invade. Under projected global temperature increases, the optimal areas for FAW persistence will shrink and, with them, the areas of seasonal invasions. Our model does not consider the multitudes of biotic interactions. However, it appears that at least the distribution of maize, FAW's preferred host plant, follows a similar pattern, which means that under future climatic conditions successful cultivation of maize will require successful management of FAW populations.