The FAW is a tropical species mostly adapted to warmer parts of the New World [9]. In the current study, we modeled its distribution under tropical conditions in Eastern DR Congo. The existence of 3 bioclimatic zones for FAW was determined in South Kivu. One (zone 3) was found to correspond to the highest probability of FAW occurrence. Climate change has been reported to have different effects on insects, impacting directly their life cycles or indirectly their hosts and/or predators [3, 39]. However, the FAW may benefit from the climate change due to its polyphagous feeding behaviour, its phenotypic and genotypic plasticity [49]. Also, the adult migratory ability is one more adaptative trait to allow moving across regions to several miles (300 miles/generation in some years) [54, 58]. In an area such as South Kivu with an approximate surface area of 69,130 km2, the FAW migration would take place very quickly. Outbreaks of FAW are closely related to climate conditions and with good winter and spring conditions [49]. Cokola [10] noted that FAW incidence in South Kivu has been associated by temperature and rainfall. Moreover, study conducted by Liu et al. [32] founded that land-use was more important than climate factors, with larger potential distributions. In this study, among the 21 used bioclimatic variables, four of them influenced the potential distribution of FAW in the region. It is therefore seen that these four variables also play a major role in the spread of FAW in South Kivu. Wang et al. [56] modelled the distribution of FAW through MaxEnt with 19 bioclimatic variables related to temperature and humidity of which 10 influenced the FAW distribution. However, the FAW distribution may be influenced by other several non-climatic factors, such as host, natural enemy, management level and human activities [24], soil properties, land cover and agricultural management interventions (such as use of pesticides or fertilizers) [6]. This aspect need to be then incorporated into the model. Furthermore, it would also be important to model the FAW distribution by integrating local bioclimatic data into the model to minimize errors related to imported bioclimatic data. Soria-Auza et al. [53] reported that one of the least studied sources of uncertainty in species distribution modeling comes from the environmental data used to run the models, particularly the climate data, especially in the tropics, where comparatively few climatic stations are available. In the case of South Kivu province, however, it is difficult to obtain sufficient local bioclimatic data given the limited number of meteorological stations found in this region.
The accuracy of prediction of FAW MaxEnt model showed high values of AUC confirming a good model performance [33]. Comparing our results with other studies, including Wang et al. [56], an excellent AUC was found. For instance, AUC often increases with the size of the study area because it contributes to include background points that have environmental characteristics greatly distant from the species requirement, resulting in artificial increase of SDM validation [4]. The suitable areas of FAW in South Kivu province are divided into two corridors. The Eastern corridor covering the Eastern areas of Kalehe, Kabare, Walungu, Uvira and Fizi territories and the Western corridor covering the Western areas of Kalehe, Kabare, Walungu and Mwenga territories, southern Shabunda and north-western Fizi territories. Infestations are most prevalent in the first corridor. Differences in the FAW infestations within the said corridor, between the Ruzizi plain (low altitude) and Kabare (mid altitude) have been demonstrated [10]. According to the modeling realized by Early et al. [14], Sub-Saharan Africa, especially DR Congo, Gabon and Cameroon, appeared to have low suitability for FAW. Early et al. [14] explain that low suitability in these countries was more likely because of extensive forest cover. This is the case for example, here for Shabunda territory. However, this does not mean that pockets of the suitable habitats in the cited countries will not be severely affected, given the ability of the FAW to travel long distances [14].
Among the four environmental variables used as predictors in the FAW MaxEnt model, bio12 (annual rainfall) played a major role in the spread of FAW and contributed more to run the MaxEnt model. With the Jackknife test for variable importance, the environmental variable exhibited highest gain when used in isolation with bio12 (annual rainfall). Day et al. [12] found that rainfall in the wettest periods and the coldest annual temperatures were important variables in FAW migration. The effects of rainfall on the distribution of FAW have been documented. For example, Early et al. [14] reported that rainfall have a negative impact on FAW larvae. Furthermore, a suitability map provided by Du Plessis et al. [13] demonstrated that natural rainfall and irrigation scenario were important variables in FAW distribution. The coldest annual temperature and the rainfall during the wettest three months were consistently identified by Early et al. [14] as the environmental variables that most affected FAW distribution. In this work, most suitable habitat for FAW was found in places where annual rainfall was less than 1600 mm. According to Early et al. [14] and Nagoshi et al. [15], FAW was most commonly found in areas with very little forest cover, a minimum annual temperature of 18–26 °C and with 500–700 mm rainfall in the three wettest months. Furthermore, given that variable bio12 is strongly negatively correlated with bio7 (annual temperature range), it seems clear that FAW also favours locations with high annual temperature. Temperature was the main environmental factor affecting the growth and reproduction of the FAW [8, 25]. FAW was most likely to be found in areas characterized by very high annual temperature range, very long duration of the longest dry season, very high annual moisture index, high maximum temperature of the hottest month and very high mean temperature of the warmest quarter. The probability of FAW occurrence is high in environments where mean annual temperature (bio1) is comprised between 19 °C and 23 °C. Du Plessis et al. [22] found that the development rate of FAW increased linearly with increasing temperatures between 18 and 30 oC. Additionally, Wang et al. [56] found that when the Mean Temperature of the Warmest Quarter varies between 19.15–29.73 °C, the existence probability of the FAW is higher.