United States Environmental Protection Agency | US EPA



SummarySSDs were fit to predicted test results for fish exposed to Malathion. Five distributions were tested against four sets of predicted values. The first set was predicted from rainbow trout (Onchorhynchus mykiss) data, the second from bluegill (Lepomis macrochirus) data, the third from fathead minnow data (Pimephales promelas), and the fourth was predicted from a mixture of all three surrogates. Only data for species typically tested in freshwater were used. Summary statistics for SSDs are presented below. Note that the statistics presented in the table below, with the exception of noting the best distribution for each species, are model-averaged estimates from across all fitted distributions.Table 1. Summary statistics for SSDs fit to Malathion test resultsstatisticrainbow troutbluegillfatheadminnowmixtureempiricalBest Distribution (by AICc)logistictriangularlogistictriangularnormalCV of the HC050.150.260.220.260.51HC0543.126.01645.723.944.8HC1049.932.62090.032.789.7HC5093.479.65235.7137.91364.9HC90225.2213.013728.81315.629214.6HC95309.8288.118199.33112.480293.0Mortality Thresh.1 (slope = 4.5)3.82.3144.62.13.9Indirect Effects Threshold1 (slope = 4.5)22.413.5854.212.423.31Slope of dose-response curve = 4.5Predicted values from rainbow trout, bluegill, and the mixture of all three surrogates produced estimates of the HC05 that were lower than, but comparable to the empirical data (Table 1). However, predicted values from fathead minnow produced estimates of the HC05 that were almost two orders of magnitude greater than the empirical value. HC05 estimates from the predicted data appeared to be more precisely estimated (smaller coefficients of variation and smaller estimates of sampling variance in the HC05). Plotted SSDs for the mixed data set show severe lack of fit, and this is confirmed in many significant P-values for lack of fit in the individual model tables below. For each fitted distribution, I also calculated where the upper confidence limit of the HC05 fell on the fitted distribution and this typically ranged from the tenth to the 25th quantile. Though there are no hard guidelines, values above the 20th percentile perhaps should be viewed with some suspicion, as indicating substantial uncertainty in the HC05 relative to the variance of the distribution. Predicted value SSDs had smaller variances than the SSD generated from empirical data. When the surrogate species rainbow trout and bluegill are examined on the empirical dataset (Fig. 5) they are seen to be among the more sensitive species. In contrast, the other surrogate species (fathead minnow) is among the more tolerant species (Fig. 5). Thus, even with a much smaller variance, the estimated HC05 comes out close to that estimated using the empirical data for bluegill and rainbow trout, but not for fathead minnow. Crosstabulation of predicted values with empirical test results show that Web ICE predictions are significantly positively correlated with empirically measured toxicity values (correlation coefficients range from 0.72 – 0.80). However, for all three surrogates, the discrepancy between predicted and measured toxicity values increases with dissimilarity between the surrogate species used to derive the predicted values and the predicted species. For species with quite dissimilar tolerances, the discrepancies between measured and predicted toxicity approached 2 orders of magnitude. This effect may be related to phylogeny. Thus predicted values get worse with taxonomic distance, as the dissimilarity in tolerance grows between two taxa (but perhaps not for predicted values from fathead minnow). This point should be investigated further. I. DataData were received from Elizabeth Donovan on 2 October 2014 in the file:“Malathion SSD_WebICE_10.02.14.xlsx”From this file I created three working files:1. MalathionFWFishIcePredRainbowTrout.xlsx2. MalathionFWFishIcePredBluegill.xlsx3. MalathionFWFishIcePredFathead.xlsx4. MalathionFWFishIcePredAllThree.xlsxThese working files each contain three fields (Table 2).Table 2. Fields in working filesField in working filesField in received fileColumn in received file1. GenusScientificC2. SpeciesScientificC3. ToxicityMeasured Toxicity (ug/L)DTable 3. Distribution of predicted values available for MalathionSurrogate SpeciesPredicted Valuesrainbow trout46bluegill36fathead minnow45All three33For each analysis I also added the surrogate species empirical test value(s) to the dataset to allow comparison of the predicted distribution to the value of the surrogate species. For further comparison, I also reanalyzed the previous empirical dataset on freshwater fish provided by Elizabeth Donovan (“Malathion_SSD Criteria_UPDATE.xlsx”, 9 July 2014) after converting that dataset from mg/L to ?g/L. Previous analyses of that data were provided in the draft report “Species Sensitivity Distributions for Aquatic Vertebrates Malathion.docx” (16 July 2014).I considered five potential distributions for the Malathion data (log-normal, log-logistic, log-triangular, log-gumbel, and Burr). To fit each of the first four distributions, I first common log (log10) transformed the toxicity values. I also explored the best-fit distribution using Akaike’s information criterion. Finally I calculated the direct and indirect effect thresholds and report five quantiles from the fitted SSDs (HC05, HC10, HC50, HC90, HC95).II. Comparison of distributions using AICcI began by using AICc to compare the five distributions for both datasets. For these comparisons all SSDs were fit using maximum likelihood. Based on AICc the datasets disagreed on which distribution best fit the predicted values. For values predicted from rainbow trout, the gumbel distribution fit best, but was followed closely by the burr distribution (Table 4). For the bluegill and fathead minnow predicted value datasets, the triangular distribution, followed by the normal distribution were selected. For the predicted values from a mixture of all three surrogates, the gumbel and burr distribution were selected as best by AICc, but both showed considerable evidence of lack-of-fit. Finally, for the empirical data, the normal distribution provided the best fit, followed closely by the gumbel and triangular distributions (Table 7). Table 4. Comparison of distributions for freshwater fish toxicity values predicted from rainbow trout.distributionHC05AICc?AICcWeightgumbel45.0310.570.000.60burr45.0313.102.520.17triangular38.7314.153.570.10normal36.7315.094.520.06logistic34.2315.164.590.06Table 5. Comparison of distributions for freshwater fish toxicity values predicted from bluegill.distributionHC05AICc?AICcWeighttriangular27.2191.080.000.50normal24.0192.831.750.21gumbel27.5193.352.270.16logistic21.2194.383.300.10burr27.5196.345.260.04Table 6. Comparison of distributions for freshwater fish toxicity values predicted from fathead minnow.distributionHC05AICc?AICcWeighttriangular1694.9314.020.000.54normal1601.8315.591.570.25logistic1495.1316.982.960.12gumbel1681.1318.154.130.07burr1664.1320.976.950.02Table 7. Comparison of distributions for freshwater fish toxicity values predicted from all three surrogates.distributionHC05AICc?AICcWeightgumbel23.9466.850.000.77burr23.9469.312.460.23logistic5.6481.3614.510.00triangular6.8481.4214.570.00normal7.2482.5215.670.00Table 8. Comparison of distributions for empirical freshwater fish toxicity data for MalathiondistributionHC05AICc?AICcWeightnormal38.9790.710.000.44gumbel56.9792.121.410.22triangular48.9792.631.920.17logistic30.9793.462.750.11burr56.9794.443.730.07The cumulative distribution functions for each of the above 4 datasets are plotted below. Figure 1. Model-averaged SSD for WebICE predicted values for freshwater fish LC50s for Malathion, based on rainbow trout regressions.Figure 2. Model-averaged SSD for WebICE predicted values for freshwater fish LC50s for Malathion, based on bluegill regressions.Figure 3. Model-averaged SSD for WebICE predicted values for freshwater fish LC50s for Malathion, based on fathead minnow regressions.Figure 4. Model-averaged SSD for WebICE predicted values for freshwater fish LC50s for Malathion, based on all three surrogates. Figure 5. Model-averaged SSD for Malathion LC50s for freshwater fish. Red points indicate single toxicity values. Black points indicate multiple toxicity values. Blue line indicates full range of toxicity values for a given taxon. IV. Goodness of fit and the importance of fitting methodTo test goodness-of-fit all five distributions were fit to each of the five datasets and bootstrap goodness-of-fit tests with 5,000 bootstrap replicates were run. Three different fitting methods were used (maximum likelihood, moment estimators, and graphical methods), though not all methods are available for all distributions. Tables 9 – 13 give results of these fitting exercises. Significant lack of fit was observed for many distributions fit to the mixture data from all three surrogates. The burr distribution showed significant lack of fit for the empirical data.Table 9. Range of HC05 values (μg/L) for Malathion SSDs estimated from Web ICE predicted values using rainbow troutdistributionmethodHC05SECVLCLUCLLCQUCQPgumbelML44.954.940.1137.5656.890.010.161.00gumbelMO44.565.980.1334.0357.490.010.171.00gumbelGR40.706.220.1525.9750.710.000.131.00normalML36.706.790.1826.4152.840.010.151.00normalMO36.036.720.1925.6252.010.010.141.00normalGR32.676.370.1919.6044.110.010.111.00logisticML34.216.950.2023.0650.220.020.140.98logisticMO36.527.520.2124.0153.820.020.140.98logisticGR31.516.960.2215.5542.930.010.100.93triangularML38.716.120.1632.0955.940.020.161.00triangularMO35.366.270.1825.8050.290.010.151.00triangularGR33.366.050.1822.4245.860.000.131.00burrML44.954.960.1137.1156.610.010.160.42Table 10. Range of HC05 values (μg/L) for Malathion SSDs estimated from Web ICE predicted values using bluegilldistributionmethodHC05SECVLCLUCLLCQUCQPgumbelML27.465.520.2020.2941.650.010.200.99gumbelMO29.956.290.2119.7044.530.000.211.00gumbelGR25.666.290.2512.0637.050.000.170.93normalML23.977.170.3014.8343.190.010.201.00normalMO23.087.140.3113.5341.330.010.191.00normalGR19.396.370.338.0432.450.000.150.97logisticML21.227.640.3611.4240.820.010.180.86logisticMO23.467.910.3412.2442.920.010.180.90logisticGR18.196.510.365.5130.440.010.130.71triangularML27.176.650.2420.9046.430.010.211.00triangularMO22.556.700.3013.7939.440.000.191.00triangularGR20.176.220.3110.0533.800.000.161.00burrML27.455.830.2118.9441.280.010.200.41Table 11. Range of HC05 values (μg/L) for Malathion SSDs estimated from Web ICE predicted values using fathead minnowdistributionmethodHC05SECVLCLUCLLCQUCQPgumbelML1681.13384.200.231204.862702.850.010.211.00gumbelMO1993.72430.860.221314.503006.580.000.221.00gumbelGR1699.49428.580.25778.362449.150.000.171.00normalML1601.82513.660.32963.242970.800.010.211.00normalMO1539.30481.430.31886.352746.430.010.191.00normalGR1285.06422.820.33518.722180.060.000.151.00logisticML1495.09542.920.36793.642868.720.010.181.00logisticMO1564.73535.680.34791.212918.150.010.191.00logisticGR1202.95438.840.36344.012018.110.000.131.00triangularML1694.92439.580.261304.833019.070.010.231.00triangularMO1504.45454.930.30912.672667.030.000.191.00triangularGR1338.79426.940.32666.122291.670.000.171.00burrML1664.12429.660.261135.972810.550.010.220.18Table 12. Range of HC05 values (μg/L) for Malathion SSDs estimated from Web ICE predicted values using all three surrogatesdistributionmethodHC05SECVLCLUCLLCQUCQPgumbelML23.926.360.2715.9240.410.010.140.35gumbelMO13.826.380.466.0030.700.010.160.02gumbelGR10.504.580.442.7920.440.000.120.00normalML7.215.390.752.6423.040.020.140.13normalMO6.834.820.712.2720.360.010.130.11normalGR5.093.310.651.0613.410.010.110.05logisticML5.613.980.711.8816.750.020.130.01logisticMO7.145.800.811.8424.420.020.130.03logisticGR4.573.060.670.5611.710.010.100.01triangularML6.804.560.673.8621.020.020.150.44triangularMO6.424.360.682.4518.950.010.140.38triangularGR5.413.410.631.6114.510.000.130.25burrML23.896.490.2715.4140.890.010.150.00Table 13. Range of HC05 values for Malathion SSDs based on measured freshwater fish LC50sdistributionmethodHC05SECVLCLUCLLCQUCQPgumbelML56.9522.130.3931.96116.600.020.130.51gumbelMO77.7232.320.4235.99159.200.010.140.83gumbelGR61.1924.030.3919.77109.860.000.110.52normalML38.8925.600.6615.15112.620.020.130.60normalMO37.2624.480.6614.13105.560.020.120.57normalGR29.0515.690.547.3768.040.010.100.41logisticML30.9323.600.769.87100.160.020.120.19logisticMO39.0428.510.7312.11119.540.020.120.30logisticGR26.5416.200.613.9164.430.010.090.14triangularML48.9430.960.6327.39146.470.020.131.00triangularMO34.9120.610.5914.1892.930.010.131.00triangularGR30.3816.100.5310.2470.860.010.111.00burrML56.9221.850.3830.74114.940.020.130.00V. Calculation of other quantilesTables 14 - 18 provide estimates of the HC05 as well as other quantiles of the fitted SSDs.Table 14. Estimated quantiles of the fitted SSDs for Malathion Web ICE predicted LC50s based on rainbow trout.distmethodHC05HC10HC50HC90HC95gumbelML45.051.090.9224.9318.0gumbelMO44.650.791.0228.4324.5gumbelGR40.747.191.9262.2391.4normalML36.745.9100.9221.9277.4normalMO36.045.2100.9225.1282.6normalGR32.741.9100.9242.9311.6logisticML34.244.294.1200.1258.6logisticMO36.547.3100.9215.4278.8logisticGR31.542.3100.9240.5323.1triangularML38.747.1107.4245.2298.1triangularMO35.443.2100.9235.5287.9triangularGR33.441.2100.9246.9305.2burrML45.051.090.9224.8317.8Table 15. Estimated quantiles of the fitted SSDs for Malathion Web ICE predicted LC50s based on bluegill.distmethodHC05HC10HC50HC90HC95gumbelML27.532.772.0249.2400.4gumbelMO30.035.172.0222.7342.8gumbelGR25.731.073.3283.1474.4normalML24.031.481.7212.5278.6normalMO23.130.581.7218.8289.3normalGR19.426.681.7250.6344.3logisticML21.229.880.6218.1306.0logisticMO23.532.281.7207.3284.5logisticGR18.226.681.7250.7367.0triangularML27.233.581.5198.1244.5triangularMO22.628.981.7231.3296.0triangularGR20.226.481.7253.2330.9burrML27.532.772.0249.1400.2Table 16. Estimated quantiles of the fitted SSDs for Malathion Web ICE predicted LC50s based on fathead minnow.distmethodHC05HC10HC50HC90HC95gumbelML1681.12030.74807.718589.531166.6gumbelMO1993.72331.34758.914581.622367.7gumbelGR1699.52052.04848.418687.931293.9normalML1601.82094.55393.813890.518162.6normalMO1539.32030.55393.814328.118900.4normalGR1285.11764.15393.816491.922639.6logisticML1495.12088.35579.114904.920819.0logisticMO1564.72142.15393.813581.918593.1logisticGR1203.01760.45393.816526.224184.9triangularML1694.92097.05149.412645.115644.9triangularMO1504.41921.35393.815142.219338.2triangularGR1338.81748.45393.816639.821731.0burrML1664.12048.54919.017836.728964.3Table 17. Estimated quantiles of the fitted SSDs for Malathion Web ICE predicted LC50s based on all three surrogates.distmethodHC05HC10HC50HC90HC95gumbelML23.932.8137.91314.13110.1gumbelMO13.821.2147.73119.410004.4gumbelGR10.517.0151.94737.517634.1normalML7.215.1207.82851.95992.3normalMO6.814.5207.82974.06323.4normalGR5.111.6207.83737.28477.8logisticML5.612.5133.41419.83173.3logisticMO7.116.8207.82570.86047.3logisticGR4.612.0207.83586.89449.2triangularML6.813.0203.33170.46078.3triangularMO6.412.5207.83457.16730.4triangularGR5.410.9207.83968.57982.8burrML23.932.8137.91314.03109.3Table 18. Estimated quantiles of the fitted SSDs for Malathion empirical LC50s for freshwater fishdistmethodHC05HC10HC50HC90HC95gumbelML56.993.9920.033028.5129749.7gumbelMO77.7121.2921.422212.374940.6gumbelGR61.2100.1942.831844.0122214.1normalML38.984.61315.320441.344491.3normalMO37.381.91315.321132.546431.4normalGR29.067.41315.325657.659561.3logisticML30.980.51343.322401.658329.7logisticMO39.095.31315.318152.444318.1logisticGR26.571.51315.324212.565197.8triangularML48.9102.02261.550146.8104507.9triangularMO34.970.01315.324726.749554.0triangularGR30.462.51315.327668.956947.4burrML56.993.9920.433007.9129594.1VI. Calculation of thresholdsThresholds were calculated using a probit curve with the HC05 as the mean and three different slopes (2, 4.5, and 9). Calculated thresholds are provided in Tables 19 - 23.Table 19. Thresholds (μg/L) for determination of action area based on Web ICE predicted values of Malathion LC50s from rainbow troutdistrib.methodMortality Threshold (10-6)Indirect Effects Threshold (10-1)slope = 4.5slope = 2slope = 9slope = 4.5slope = 2slope = 9gumbelML3.90.213.323.310.332.4gumbelMO3.90.213.223.110.232.1gumbelGR3.60.212.121.19.329.3normalML3.20.210.919.18.426.4normalMO3.20.210.718.78.226.0normalGR2.90.19.717.07.523.5logisticML3.00.110.117.87.824.6logisticMO3.20.210.819.08.426.3logisticGR2.80.19.316.47.222.7triangularML3.40.211.520.18.927.9triangularMO3.10.110.518.48.125.5triangularGR2.90.19.917.37.624.0burrML3.90.213.323.310.332.4Table 20. Thresholds (μg/L) for determination of action area based on Web ICE predicted values of Malathion LC50s from bluegilldistrib.methodMortality Threshold (10-6)Indirect Effects Threshold (10-1)slope = 4.5slope = 2slope = 9slope = 4.5slope = 2slope = 9gumbelML2.40.18.114.36.319.8gumbelMO2.60.18.915.56.821.6gumbelGR2.30.17.613.35.918.5normalML2.10.17.112.45.517.3normalMO2.00.16.812.05.316.6normalGR1.70.15.710.14.414.0logisticML1.90.16.311.04.915.3logisticMO2.10.17.012.25.416.9logisticGR1.60.15.49.44.213.1triangularML2.40.18.114.16.219.6triangularMO2.00.16.711.75.216.2triangularGR1.80.16.010.54.614.5burrML2.40.18.114.26.319.8Table 21. Thresholds (μg/L) for determination of action area based on Web ICE predicted values of Malathion LC50s from fathead minnowdistrib.methodMortality Threshold (10-6)Indirect Effects Threshold (10-1)slope = 4.5slope = 2slope = 9slope = 4.5slope = 2slope = 9gumbelML147.77.1498.2872.6384.41211.2gumbelMO175.18.4590.91034.8455.91436.4gumbelGR149.37.1503.7882.1388.61224.4normalML140.76.7474.7831.4366.31154.0normalMO135.26.5456.2799.0352.01109.0normalGR112.95.4380.9667.0293.9925.8logisticML131.36.3443.1776.0341.91077.1logisticMO137.46.6463.7812.2357.81127.3logisticGR105.75.1356.5624.4275.1866.7triangularML148.97.1502.3879.8387.61221.1triangularMO132.16.3445.9780.9344.01083.9triangularGR117.65.6396.8694.9306.2964.5burrML146.27.0493.2863.8380.51198.9Table 22. Thresholds (μg/L) for determination of action area based on Web ICE predicted values of Malathion LC50s from all three surrogatesdistrib.methodMortality Threshold (10-6)Indirect Effects Threshold (10-1)slope = 4.5slope = 2slope = 9slope = 4.5slope = 2slope = 9gumbelML2.10.17.112.45.517.2gumbelMO1.20.14.17.23.210.0gumbelGR0.90.03.15.42.47.6normalML0.60.02.13.71.65.2normalMO0.60.02.03.51.64.9normalGR0.40.01.52.61.23.7logisticML0.50.01.72.91.34.0logisticMO0.60.02.13.71.65.1logisticGR0.40.01.42.41.03.3triangularML0.60.02.03.51.64.9triangularMO0.60.01.93.31.54.6triangularGR0.50.01.62.81.23.9burrML2.10.17.112.45.517.2Table 23. Thresholds (μg/L) for determination of action area using empirical values for Malathion LC50s for freshwater fishdistrib.methodMortality Threshold (10-6)Indirect Effects Threshold (10-1)slope = 4.5slope = 2slope = 9slope = 4.5slope = 2slope = 9gumbelML5.00.216.929.613.041.0gumbelMO6.80.323.040.317.856.0gumbelGR5.40.318.131.814.044.1normalML3.40.211.520.28.928.0normalMO3.30.211.019.38.526.8normalGR2.60.18.615.16.620.9logisticML2.70.19.216.17.122.3logisticMO3.40.211.620.38.928.1logisticGR2.30.17.913.86.119.1triangularML4.30.214.525.411.235.3triangularMO3.10.110.318.18.025.2triangularGR2.70.19.015.86.921.9burrML5.00.216.929.513.041.0VII. Comparison of Predicted and Tested ToxicityTo further compare SSDs fit to predicted values to SSDs generated from test results I cross-tabulated the taxa from each set of predicted values with corresponding taxa for which empirical test results were available (Tables 24 – 28). These tables also report the discrepancy as absolute value of the difference between predicted and tested toxicity value, scaled to the empirical toxicity value (e.g., the discrepancy for Lepomis macrochirus when Oncorhynchus mykiss is used as a surrogate is |108.2 – 69.3|/69.3 = 0.56) and the toxicological distance as the absolute value of the difference between the tested toxicity value of the surrogate test species and that of every other species. For example, the distance between Oncorhynchus mykiss and Lepomis macrochirus, when Oncorhynchus mykiss is used as the surrogate is |73.5? 69.3|/69.3 = 0.06. Two correlations were examined for each of these sets of crosstablulated data, between the predicted and tested toxicity values and between the discrepancy statistic and the distance statistic. The latter test was done to examine whether predicted values get worse as the distance (measured as relative sensitivity to the chemical) increases between taxa. Results of these correlations are presented in Table 28. In all cases, the predicted and tested values were significantly correlated. However, there was also a significant correlation between the discrepancy and distance statistics, suggesting that an estimated toxicity value would be less accurate, as the difference in sensitivity between surrogate and predicted species grows.Table 24. Crosstabulated toxicity values for Web ICE predictions based on rainbow trout regressions.TaxonWeb ICEGeomeanDiscrepancyDistanceMicropterus salmoides57.50266.930.780.72Salmo salar66.24313.600.790.77Perca flavescens68.09263.000.740.72Sander vitreus70.1064.000.100.15Salmo trutta71.43101.000.290.27Oncorhynchus mykiss73.5273.520.000.00Salvelinus namaycush84.62103.880.190.29Oncorhynchus clarkii85.07235.020.640.69Oncorhynchus kisutch99.05173.460.430.58Lepomis macrochirus108.2469.280.560.06Lepomis cyanellus138.81163.160.150.55Ictalurus punctatus183.148267.490.980.99Pimephales promelas212.5711029.170.980.99Cyprinus carpio229.074941.000.950.99Carassius auratus334.245161.090.940.99Poecilia reticulata378.953069.000.880.98Ameiurus melas392.6312285.360.970.99Table 25. Crosstabulated toxicity values for Web ICE predictions based on bluegill regressions.TaxonWeb ICEGeomeanDiscrepancyDistanceSalmo salar26.3313.60.920.78Salmo trutta36.7101.00.640.31Perca flavescens53.6263.00.800.74Micropterus salmoides56.8266.90.790.74Oncorhynchus kisutch58.5173.50.660.60Oncorhynchus mykiss63.773.50.130.06Lepomis macrochirus69.369.30.000.00Salvelinus namaycush86.8103.90.160.33Oncorhynchus clarkii101.1235.00.570.71Lepomis cyanellus130.7163.20.200.58Pimephales promelas167.411029.20.980.99Cyprinus carpio179.44941.00.960.99Ictalurus punctatus219.18267.50.970.99Poecilia reticulata234.03069.00.920.98Carassius auratus274.55161.10.950.99Table 26. Crosstabulated toxicity values for Web ICE predictions based on fathead minnow regressions.TaxonWeb ICEGeomeanDiscrepancyDistanceSalvelinus namaycush1374.7103.912.2105.2Oncorhynchus kisutch1740.1173.59.062.6Perca flavescens2351.2263.07.940.9Micropterus salmoides2728.5266.99.240.3Oncorhynchus clarkii2730.3235.010.645.9Oncorhynchus mykiss4615.073.561.8149.0Lepomis macrochirus6234.669.389.0158.2Cyprinus carpio6520.44941.00.31.2Poecilia reticulata10520.73069.02.42.6Pimephales promelas11029.211029.20.00.0Carassius auratus11101.75161.11.21.1Ameiurus melas11480.512285.40.10.1Oryzias latipes12110.69700.00.20.1Ictalurus punctatus15639.98267.50.90.3Table 27. Crosstabulated toxicity values for Web ICE predictions based on all three surrogates.TaxonWeb ICEGeomeanMicropterus salmoides56.8266.9Salmo salar66.2313.6Perca flavescens68.1263.0Lepomis macrochirus69.369.3Sander vitreus70.164.0Salmo trutta71.4101.0Oncorhynchus mykiss73.573.5Salvelinus namaycush84.6103.9Oncorhynchus clarkii85.1235.0Oncorhynchus kisutch99.1173.5Lepomis cyanellus130.7163.2Ictalurus punctatus183.18267.5Cyprinus carpio6520.44941.0Poecilia reticulata10520.63069.0Pimephales promelas11029.211029.2Carassius auratus11101.75161.1Ameiurus melas11480.412285.4Oryzias latipes12110.69700.0Table 28. Correlations between cross-tabulated toxicity valuesSurrogate Speciescorrelation between predicted and testedCorrelation between discrepancy and distancerainbow troutρ = 0.74, p = 7.5*10-4ρ = 0.88, p = 2.6*10-6bluegillρ = 0.72, p = 2.7*10-3ρ = 0.90, p = 4.6*10-6fathead minnowρ = 0.84, p = 1.8*10-4ρ = 0.89, p = 1.7*10-5all threeρ = 0.80, p = 1.4*10-4n/a ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download