Global Potential Distribution of Bactrocera carambolae and the Risks for Fruit Production in Brazil

Abstract

The carambola fruit fly, Bactrocera carambolae, is a tephritid native to Asia that has invaded South America through small-scale trade of fruits from Indonesia. The economic losses associated with biological invasions of other fruit flies around the world and the polyphagous behaviour of B. carambolae have prompted much concern among government agencies and farmers with the potential spread of this pest. Here, ecological niche models were employed to identify suitable environments available to B. carambolae in a global scale and assess the extent of the fruit acreage that may be at risk of attack in Brazil. Overall, 30 MaxEnt models built with different combinations of environmental predictors and settings were evaluated for predicting the potential distribution of the carambola fruit fly. The best model was selected based on threshold-independent and threshold-dependent metrics. Climatically suitable areas were identified in tropical and subtropical regions of Central and South America, Sub-Saharan Africa, west and east coast of India and northern Australia. The suitability map of B. carambola was intersected against maps of fruit acreage in Brazil. The acreage under potential risk of attack varied widely among fruit species, which is expected because the production areas are concentrated in different regions of the country. The production of cashew is the one that is at higher risk, with almost 90% of its acreage within the suitable range of B. carambolae, followed by papaya (78%), tangerine (51%), guava (38%), lemon (30%), orange (29%), mango (24%) and avocado (20%). This study provides an important contribution to the knowledge of the ecology of B. carambolae, and the information generated here can be used by government agencies as a decision-making tool to prevent the carambola fruit fly spread across the world.

Potential global distribution of Diabrotica species and the risks for agricultural production

“…Recent cases of biological invasion of insect pests suggest that this is a real threat. 29,67 Combined with data on propagule pressure and the main invasion pathways of the Diabrotica species, our findings can contribute to the development of pest risk analysis by policy makers and plant protection organizations in order to determine Pest Manag Sci 2018; 74: 2100-2109 priority areas for the adoption of phytosanitary measures aiming to prevent the introduction and spread of these species. This is an important issue considering that the history of invasions shows that prevention is more cost-effective than eradicating or controlling the invading species.…”

“…Apart from potential economic losses associated with direct insect damage on plants, there is the risk of environmental contamination as a result of increased use of highly toxic pesticides to promote pest control in a region where the pest may not be controlled by indigenous natural enemies. Recent cases of biological invasion of insect pests suggest that this is a real threat …”

“…After selection, the final model was run with the following changes in the default settings of MaxEnt: (i) create response curves option was enabled to evaluate the species response to individual predictors, (ii) a jackknife procedure was performed to measure variable importance, (iii) 75% of the occurrence data were used for training and the remaining data for the test, and (iv) replicate run type was used as a subsample with 50 replicates. Binary maps displaying marginal and optimal environments were generated using the minimum training presence threshold (MPT) and the 10% training presence threshold (TP10) applied to the resulted maps showing logistic scores, respectively . While MTP uses the suitability associated with the least suitable training presence record as the threshold, TP10 uses the suitability associated with the presence record that occurs at the 10th percentile of presence records.…”

“…These variables are derived from weather stations across the world, 27 and are commonly used in ecological niche modeling studies because they capture annual ranges and limiting temperature and precipitation conditions that are known to constraint the occurrence of insects. 28,29 Additionally, data on soil clay, silt and sand content was obtained from SoilGrids (www.soilgrids.org) considering the influence of soil physical properties on the survival of the immature Diabrotica species. [30][31][32][33] The inclusion of highly correlated variables is known to generate complex and over-fitting models, 26,34 and therefore we carried out a Pearson's correlation test to ensure the lack of multicollinearity among the variables.…”

“…Binary maps displaying marginal and optimal environments were generated using the minimum training presence threshold (MPT) and the 10% training presence threshold (TP10) applied to the resulted maps showing logistic scores, respectively. 29,47 While MTP uses the suitability associated with the least suitable training presence record as the threshold, TP10 uses the suitability associated with the presence record that occurs at the 10th percentile of presence records.…”