Potential global distribution of Aleurocanthus woglumi considering climate change and irrigation

Among various methods used for model validation, most CLIMEX validation exercises focus on comparing the modelled potential distributions with the known occurrences in geo- graphical areas not used for parameter fitting in pests’ native ranges [61,62]. In our study, we also used observations about voltinism of A. woglumi in three known locations taken from the literature (south Florida in USA, Kuala Lumpur in Malaysia, and Panama Canal Zone) and compared them with the modelled establishment results for the pest in these locations [63]. Irrigation scenario Non-climatic habitat factors can have a substantial impact on cultivation and consequently on many pest species’ distributions. Irrigation is one such factor that affects soil moisture. Aleuro- canthus woglumi has location records in the Middle East and India where agriculture is only possible because of irrigation. In the CLIMEX model under current climate conditions (natu- ral rainfall scenario), using biologically plausible soil moisture parameters, many of these loca- tions would be modelled as unsuitable due to drought stress. To account for the effects of irrigation, an irrigation scenario of 4.5 mm day-1 was applied globally as a top-up, using a com- posite map developed from a global map of irrigated areas [64]. The global map of irrigated areas (GMIA) indicates the irrigated areas within each 5’ grid cell. For this study, the GMIA dataset was aggregated to 30’ grid cells. A threshold of 10 ha of irrigated area was applied; for

each 30’ cell, if the irrigated area was greater than or equal to 10 ha it was considered to be irri- gated, otherwise it was classified as natural rainfall. A composite risk map was developed using a method that is described in detail in Yonow et al. [59]. In short, for cells classified as irriga- tion, the greater EI value under the irrigation or natural rainfall scenario was used. For the remaining cells, the EI value of the natural rainfall scenario was used. Parameter sensitivity and model uncertainty Parametric sensitivity is the strength of the effect of each parameter on selected state variables. CLIMEX Version 4 provides tools to automate analyses. In the sensitivity analysis, the model is run for a scenario where each parameter is adjusted upward and then downward by the rele- vant perturbation value. In CLIMEX the perturbation ranges are set as follows: for temperature +/− 1˚C, soil moisture +/− 10%, rate parameters +/− 10%, degree day temperature thresholds +/− 1˚C and degree day sum (PDD) +/− 20˚C days [44]. A table of all the parameters is pro- duced, indicating the amount by which each has been adjusted and the effect this adjustment had on each of the state variables. Model uncertainty is an estimate of incertitude in the results of the model. The automated uncertainty analysis provided by CLIMEX Version 4 uses a Latin hypercube framework to effi- ciently sample parameter uncertainty. The analysis produces a map indicating the propor- tional model agreement (%) for climate suitability. For each of the sensitivity and uncertainty

analyses, the default model parameters were run for the entire world using the CM30 1995H V2 dataset. Results and discussion Current climate projections and model validation The potential distribution of A. woglumi under historical climatic conditions is presented in Fig 1. Globally, around 42 and 51 million km2 of land are modelled as climatically suitable for A. woglumi in the absence and presence of irrigation under current climate, respectively (Table 2). Under current climate conditions, A. woglumi is well distributed in tropical and sub- tropical regions throughout Asia and the Pacific, to central and southern Africa as well as parts PLOS ONE Potential global distribution of Aleurocanthus woglumi PLOS ONE | https://doi.org/10.1371/journal.pone.0261626 December 20, 2021 7 / 18