In agricultural ecosystems, the relationship between insect pests and hosts is important, as insect pests can invade hosts, increasing insect pest density that threatens the hosts’ health. Insect pests and hosts are negatively correlated and affect the environment around them. i.e., host health, environment, and insect pest density are causally related, and the environment affects insect pest density. Deep learning is method of machine learning based on neural network theory. This approach enables handling uncertain environmental factors that simultaneously impact the density of F. occidentalis. Environmental factors affecting the density fluctuation of F. occidentalis selected atmosphere factors, soil factors, and host factors. This study aims to F. occidentalis monitoring using deep learning models inputting environmental factors.

As climate changes and global trade volume increases, the spread of invasive alien species accelerates. Early prevention before occurrence is crucial for invasive pest control. Therefore, this study modeled the current and future potential distribution of the tomato leafminer(Tuta absoluta) (Meyrick) (Lepidoptera: Gelechiidae), the most significant pest affecting tomatoes, in Korea. This pest primarily feeds on Solanaceae crops and can cause extensive damage, resulting in 50-100% loss of crops in greenhouses or fields. While previously unreported in Korea, it invaded China in 2017, indicating a potential threat to Korea. The potential distribution of the tomato leafminer in Korea under current and three future climate scenarios (SSP1-26, SSP3-70, SSP5-85) was predicted using the MaxEnt model. Additionally, elevation and land cover were incorporated as abiotic factors considering the ecological characteristics of the pest.

Density survey should be carried out for applying integrated pest management strategies, but it is labor-intensive, time- and cost-consuming. Therefore, binomial sampling plans are developed for estimating and classifying the population density of whiteflies late larvae based on the relationship between the mean density per sample unit (7 leaflets) and the proportion of leaflets infested with less than T whiteflies ( ). In this study, models were examined using tally thresholds ranging from 1 to 5 late larvae per 7 leaflets. Regardless of tally thresholds, increasing the sample size had little effect on the precision of the binomial sampling plan. Based on the precision of the model, T=3 was the best tally threshold for estimating the densities of late larvae. Models developed using T=3 validated by Resampling Validation for Sampling Plan program. Above all, the binomial model with T=3 performed well in estimating the mean density of whiteflies in greenhouse tomato.

담배가루이(Bemisia tabaci)는 광식성 해충으로 토마토황화잎말림바이러스(Tomato Yellow Leaf Curl Virus; TYLCV), 카사바갈색줄무늬병 (Cassava Brown Streak Disease; CBSD)를 매개하는 해충이다. 담배가루이 방제를 위해 화학적 방제가 주로 시행되지만 저항성으로 인한 한계로 인 해 종합적해충방제를 위한 고정정확도를 이용한 표본조사법(Fixed precision sampling plan)을 개발하였다. 표본추출은 토마토 식물이 50 cm 높이의 레일 위에 위치한 배지를 이용해 재배되고 있어 배지로부터 130 cm 이상(지상에서 180 cm 이상)을 상단, 70 cm~100 cm (지상에서 120 cm~150 cm)를 중단, 50 cm 이하(지상에서 100 cm 이하)를 하단으로 나누어 각 위치별 토마토 7엽의 잎 뒷면에서 관찰된 담배가루이 노숙 유충 마리 수를 조사 하였다. 담배가루이 노숙유충은 이동성이 거의 없어 알에서 우화한 뒤 고착화하여 용과 성충 단계를 거치기 때문에 중단, 하단에 밀도가 높았다. 공간분 포분석은 Taylor’s power law (TPL)를 이용하여 도출된 TPL의 회귀계수를 통해 분석하였고, TPL 계수의 차이는 공분산분석(ANCOVA)하여 차이 가 없어 자료를 통합(pooling)하여 계산된 새로운 TPL 계수를 이용하여 표본추출정지선과 방제의사결정법을 개발하였다. 개발된 표본추출법의 적합 성을 판단하기 위해 분석에 사용하지 않은 독립된 자료를 이용하여 Resampling Validation for Sampling Plan (RVSP) 프로그램으로 평가하였다.

A causality exists between insect density and plant health, where plant health is affected by both the plant’s potential and environmental factors. In other words, causality is possible between insect density and environmental factors, allowing for the analysis of insect density based on these environmental factors. Machine learning enables studying insect density alongside environmental factors, providing insights into the causality between insects, the environment, and plant health. Machine learning is a methodology that involves the design of models by learning patterns from input data. This study aims to predict F. occidentalis density by sampling environmental factors and applying them to machine learning models.

For effective control of Frankliniella occidentalis, one of polyphagous pests with resistance to insecticides, necessitates the implementation of an integrated pest management strategy. Therefore, estimation of pest density is essential and this is achieved through the application of spatial statistical analysis methods. Because traditional methods often overlook the correlation between sampling locations and data, geostatistical analysis using variogram and kriging is introduced. Variogram provides information on the independent distance between data points. Kriging is a spatial interpolation technique for estimating the values at unsampled locations. For assessing model fitness, cross-validation is used by comparing predicted values with actual observations. This study focuses on the application of geostatistical techniques to estimate F. occidentalis density in hot pepper greenhouse, thereby contributing to making decision.

Climate change and biological invasions are the greatest threats to biodiversity, agriculture, health and the global economy. Tomato leafminer(Tuta absoluta) (Meyrick) (Lepidoptera: Gelechiidae) is one of the most important threats to agriculture worldwide. This pest is characterized by rapid reproduction, strong dispersal ability, and highly overlapping of generations. Plants are damaged by direct feeding on leaves, stems, buds, calyces, young ripe fruits and by the invasion of secondary pathogens which enter through the wounds made by the pest. Since it invaded Spain in 2006, it has spread to Europe, the Mediterranean region, and, in 2010, to some countries in Central Asia and Southeast Asia. In East Asia, Tomato leafminer was first detected in China in Yili, Xinjiang Uygur Autonomous Region, in 2017. There is a possibility that this pest will invade South Korea as well. This study provides this by the use of MaxEnt algorithm for modelling the potential geographical distribution of Tomato Leafminer in South Korea Using presence-only data.

Since the importance of integrated pest management to minimize environmental damage and maximize pest control effectiveness has emerged, efforts to put it into practice have continued. To implement IPM, it is necessary to estimate the economic injury level to determine the control method by identifying pests and weeds that damage the quantity and quality of crops in the field, investigating the occurrence level, and calculating the ratio of cost and effectiveness. Also, damage to host plants caused by increased density of insect pests appears to change plant’s health that key factor for managing crops. Therefore, understanding the relationship between the density of pests and the damage to the host plants is necessary. This study aims to analyze the causal relationship between the density of insect pests and damage to the host plants for estimating the economic injury level of insect pests on the host plants and investigating the possibility of pest control decision-making using plant health status.

Among migratory insect pests, Mythimna seperata and Cnaphalocrocis medinalis are invasive pests introduced into South Korea through westerlies from southern China. M. seperata and C. medinalis are insect pests that use rice as a host. They injure rice leaves and inhibit rice growth. To understand the distribution of M. seperata and C. medinalis, it is important to understand environmental factors such as temperature and humidity of their habitat. This study predicted current and future habitat suitability models for understanding the distribution of M. seperata and C. medinalis. Occurrence data, SSPs (Shared Socio-economic Pathways) scenario, and RCP (Representative Concentration Pathway) were applied to MaxEnt (Maximum Entropy), a machine learning model among SDM (Species Distribution Model). As a result, M. seperata and C. medinalis are aggregated on the west and south coasts where they have a host after migration from China. As a result of MaxEnt analysis, the contribution was high in the order of Land-cover data and DEM (Digital Elevation Model). In bioclimatic variables, BIO_4 (Temperature seasonality) was high in M. seperata and BIO_2 (Mean Diurnal Range) was found in C. medinalis. The habitat suitability model predicted that M. seperata and C. medinalis could inhabit most rice paddies.

Frankliniella occidentalis is an invasive pest insect, which affects over 500 different species of host plants and transmits viruses (tomato spotted wilt virus; TSWV). Despite their efficiency in controling insect pests, pesticides are limited by residence, cost and environmental burden. Therefore, a fixed-precision level sampling plan was developed. The sampling method for F. occidentalis adults in pepper greenhouses consists of spatial distribution analysis, sampling stop line, and control decision making. For sampling, the plant was divided into the upper part (180 cm above ground), middle part (120-160 cm above ground), and lower part (70-110 cm above ground). Through ANCOVA, the P values of intercept and slope were estimated to be 0.94 and 0.87, respectively, which meant there were no significant differences between values of all the levels of the pepper plant. In spatial distribution analysis, the coefficients were derived from Taylor’s power law (TPL) at pooling data of each level in the plant, based on the 3-flowers sampling unit. F. occidentalis adults showed aggregated distribution in greenhouse peppers. TPL coefficients were used to develop a fixed-precision sampling stop line. For control decision making, the pre-referred action thresholds were set at 3 and 18. With two action thresholds, Nmax values were calculated at 97 and 1149, respectively. Using the Resampling Validation for Sampling Program (RVSP) and the results gained from the greenhouses, the simulated validation of our sampling method showed a reasonable level of precision.