Key Il Shin

Simulation output analysis using the threshold bootstrap

Abstract The threshold bootstrap extends the bootstrap method of inference to autocorrelated data series, such as the outputs of discrete event simulations. The method works by resampling random chunks that are some multiple of a cycle . A cycle consists of alternating high and low runs that are created when the time series wanders back and forth across a threshold . Monte Carlo simulations show that the threshold bootstrap performs well in estimating the standard error of the sample mean and constructing confidence intervals with appropriate coverage and compact half-widths. We establish the asymptotic unbiasedness and consistency of threshold bootstrap estimates in the case of the sample mean. Comparison with the method of batch means and the moving blocks bootstrap shows that the threshold bootstrap is an attractive alternative for simulation output analysis.

Evaluation of resistance pattern to fenpyroximate and pyridaben in Tetranychus urticae collected from greenhouses and apple orchards using lethal concentration-slope relationship

This study aimed to monitor the present and future developments of the resistance of Tetranychus urticae Koch to fenpyroximate and pyridaben, using the relationship of the LC50 and slope of the concentration-mortality line in a probit model, for the provision of reliable resistance management tactics. Tetranychus urticae populations were collected from 16 commercial greenhouses, where various crops were cultivated, as well as from 10 apple orchards throughout Korea. The resistance to fenpyroximate and pyridaben of each population was estimated by calculating the median lethal concentration (LC50), resistance ratio (RR) and slope of the concentration-mortality regression. Most of the greenhouse populations exhibited moderate levels of resistance, whereas the apple orchard populations showed only low levels, indicating that T. urticae populations in greenhouses were more strongly selected than those in apple orchards. Four population groups were established based on either the habitats (greenhouse and apple orchard) or acaricides (fenpyroximate and pyridaben). To test the hypothesis, “the slope is greatest at low and high levels of resistance,” the slopes were regressed as a function of the LC50, and fitted to a polynomial regression. The polynomial regression model explained this relationship well for the four population groups (p < 0.05), indicating that the development of resistance toward fenpyroximate or pyridaben was consistent with the gradient. A laboratory selection study agreed with the results from both acaricide field populations. These results suggest that the gradient was a good indicator of the susceptibility of T. urticae to genetic variations, which was related to the LC50. The application of these findings is also discussed in relation to the resistance management of T. urticae.

Response surface model for predicting chronic toxicity of cadmium to Paronychiurus kimi (Collembola), with a special emphasis on the importance of soil characteristics in the reproduction test

A central composite design (CCD) was employed to investigate the effects of organic matter (OM) content and soil pH on the reproduction, and chronic toxicity (28-d EC(50-reproduction)) of cadmium for Paronychiurus kimi after 28days exposure in a standard artificial soil. Two statistical models were developed, one describing reproduction in control artificial soils as a function of OM content and pH, and the other describing cadmium toxicity to the same soil parameters. In the reproduction model, pH was the most important factor, followed by two quadratic factors of OM(2) and pH(2). The parameter pH alone could explain 75.5% of the response variation. The reproduction model will allow us to predict a mean reproduction in the non-treated control soils that contain various combinations of OM content and different pH values. In the chronic toxicity model, only the linear factor of the OM content and pH significantly (p<0.05) affect cadmium toxicity, which explains the 78.9% and 14.9% of total response variance, respectively. Therefore, the final polynomial regression describing the chronic toxicity of cadmium to P. kimi is as follows: predicted 28-d EC(50) of cadmium (mgkg(-1))=-21.231+2.794 x OM+4.874 x pH. The present study show that soil characteristics, which can alter the toxicity of cadmium, can also act as stressors themselves in regards to the reproduction of P. kimi. Based on the physico-chemical characteristics of the test media, the response surface model developed in this study can be used to provide initial toxicity information for cadmium within a region of interest in terms of OM content and pH, and may lead to more scientific based risk assessment for metals.

Comparison of spatial interpolation techniques for predicting climate factors in Korea

A variety of statistical interpolation techniques have been used with climate factors. This study compared three statistical interpolation techniques such as PRISM, Kriging, and IDW with temperature and precipitation. Mean monthly cumulative values on temperature and precipitation from 1977 to 2006, which were provided by Korea Meteorological Administration, were used for this study. The aim of this study is to find better appropriate process which can consider to the topographical characteristics of Korea and to produce high‐resolution climate‐maps using statistical models. As a result, Kriging showed gradual and smooth pattern, while IDW showed rather irregularly distributed with discontinuous borders. And PRISM generated the most detailed pattern. This paper will contribute to coping with climate‐change such as urban heat island, flood, disaster, and to make alterative suggestion for the meteorological factors on forest and vegetation models in the future.

Evaluation of Data Transformations and Validation of a Spatial Model for Spatial Dependency of Trialeurodes vaporariorum Populations in a Cherry Tomato Greenhouse

Abstract An evaluation of data transformations was made for constructing the reliable spatial models of the greenhouse whitefly {Trialeurodes vaporariorum Westwood) populations in a commercial cherry tomato greenhouse. A Box-Cox power transformation that is useful family of transformations was applied to original data sets. The ability of the transformations to correct for the heterogeneity of variance was tested with Shapiro-Wilk W statistics. After finding the appropriate transformations, empirical variograms were calculated and fitted to spherical model. In this study, the data transformations can stable variogram modeling by means of converting non-normal data to normal. The model was validated with new data set by comparing the deviation between observed and predicted values, using a leave-one-out method. Among the data transformations tested, loge (x+0.5) and (x+0.5) transformations were found to be appropriate at correcting for the heterogeneity of variance. According to the leave-one-out cross validation, the (x+0.5) transformation was better than the loge(x+0.5) transformation. However, both transformations produced a systemic deviation: the predicted mean was always smaller than the observed mean. No transformations were found to be appropriate, when a proportion of empty sample units (no individuals were observed) was higher than 0.2. Moreover, in this study, the abnormal high density of sample units made inappropriate spherical variogram modeling.

Joint toxic action of binary metal mixtures of copper, manganese and nickel to Paronychiurus kimi (Collembola)

The joint toxic effects of binary metal mixtures of copper (Cu), manganese (Mn) and nickel (Ni) on reproduction of Paronhchiurus kimi (Lee) was evaluated using a toxic unit (TU) approach by judging additivity across a range of effect levels (10-90%). For all metal mixtures, the joint toxic effects of metal mixtures on reproduction of P. kimi decreased in a TU-dependent manner. The joint toxic effects of metal mixtures also changed from less than additive to more than additive at an effect level lower than or equal to 50%, while a more than additive toxic effects were apparent at higher effect levels. These results indicate that the joint toxicity of metal mixtures is substantially different from that of individual metals based on additivity. Moreover, the close relationship of toxicity to effect level suggests that it is necessary to encompass a whole range of effect levels rather than a specific effect level when judging mixture toxicity. In conclusion, the less than additive toxicity at low effect levels suggests that the additivity assumption is sufficiently conservative to warrant predicting joint toxicity of metal mixtures, which may give an additional margin of safety when setting soil quality standards for ecological risk assessment.

Detecting and cleaning outliers for robust estimation of variogram models in insect count data

Outlier detection and cleaning procedures were evaluated to estimate mathematical restricted variogram models with discrete insect population count data. Because variogram modeling is significantly affected by outliers, methods to detect and clean outliers from data sets are critical for proper variogram modeling. In this study, we examined spatial data in the form of discrete measurements of insect counts on a rectangular grid. Two well-known insect pest population data were analyzed; one data set was the western flower thrips, Frankliniella occidentalis (Pergande) on greenhouse cucumbers and the other was the greenhouse whitefly, Trialeurodes vaporariorum (Westwood) on greenhouse cherry tomatoes. A spatial additive outlier model was constructed to detect outliers in both the isolated and patchy spatial distributions of outliers, and the outliers were cleaned with the neighboring median cleaner. To analyze the effect of outliers, we compared the relative nugget effects of data cleaned of outliers and data still containing outliers after transformation. In addition, the correlation coefficients between the actual and predicted values were compared using the leave-one-out cross-validation method with data cleaned of outliers and non-cleaned data after unbiased back transformation. The outlier detection and cleaning procedure improved geostatistical analysis, particularly by reducing the nugget effect, which greatly impacts the prediction variance of kriging. Consequently, the outlier detection and cleaning procedures used here improved the results of geostatistical analysis with highly skewed and extremely fluctuating data, such as insect counts.

Combined effects of organic matter and pH on acute toxicity of cadmium to Paronychiurus kimi (Collembola): development of response surface model.

A second-order central composite design was employed to investigate the effect of organic matter (OM) content and soil acidity (pH) on the cadmium toxicity to Paronychiurus kimi (Lee), a collembolan species native to Korea. Two independent variables, OM and pH, were adjusted from 0 to 10% (in total dry wt) and 4.5 to 7.0, respectively. Cadmium concentrations tested were 10, 20, 40, 80, and 160 mg/kg dry soil. The toxic effects that cadmium exerted on P. kimi (28-day median lethal concentration (LC50s)) varied significantly with different combinations of OM content and pH. The second-order response surface model was appropriate for describing combined effects of OM and pH on the cadmium toxicity to P. kimi. Linear effects (OM and pH) were predominant contributors accounting for ≈70% of the total variance of cadmium toxicity to P. kimi. No significant quadratic (pH × pH) and interaction (OM × pH) effects were observed; therefore, the final second-order response surface model could be reduced as follows: 28-day LC50 of cadmium (mg/kg) = –94.87 + 26.69 × OM + 16.64 × pH – 1.82 × OM2. Validation of the developed response surface model using 4 soil media with different OM content and pH resulted in a significant correlation coefficient (R2 = 0.97) between predicted and observed 28-day LC50. Based on physico-chemical properties of the test media, developed models can be used to explain the variation of cadmium toxicities at various environmental conditions and may improve current environmental risk assessment procedures of metals.

Evaluation of binomial sequential classification sampling plan for leafmine of Liriomyza trifolii (Diptera: Agromyzidae) in greenhouse tomatoes

Abstract Sequential binomial sampling plans for classifying the leafmine density of Liriomyza trifolii (Burgess) were developed and evaluated in tomato greenhouses in southern Korea during 2003 – 2004. Two action thresholds (m AT), three and seven leafmines per tomato leaf, were set by examining the relationship between the leafmine density and the decrease in the tomato yield. An empirical P T -m model, which is expressed as ln(−ln(1−P T )) = γ + δ ln(m), was used to examine the relationship between the proportion of the infested tomato leaves (P T ) with at least T (tally threshold) leafmines and the mean leafmine density (m). The empirical model showed an improvement in the fit up to T = 5, which then stabilised at T > 5. Walds sequential probability ratio test was used to formulate the sequential sampling stop lines relative to m AT = 3 and 7 with T = 1, 2, 3, 4, and 5. The sampling plans were evaluated using the operating characteristic (OC) and the average sample number (ASN) functions. The robustness of the sampling plans improved with increasing T values but the improvements were negligible at T ≥ 3. Simulation analysis with the OC and ASN functions indicated that a binomial model with T = 3 was optimal and less than 70 samples were required to classify the number of leafmines relative to either m AT values. Resampling simulation of eight independent data sets with T = 3 showed that the performance of the sampling plans was superior to that expected from the established OC and ASN functions. The correct classification rate was at least 99% and the matching sample size was <57.