Results of ecological niche modeling using a C-SVM based on Korarchaeota abundance and bulk geochemistry data.

aSVMs were created using a radial basis kernel for all single and two analyte combinations. Analytes and analyte combinations that had not appeared in any classifier scoring over 80% accuracy with 5 bootstraps were dropped from the final training sets to reduce the computational burden of additional testing. These reduced data sets were subjected to the same analysis as previously, using the radial basis kernel function and 100 replicates to yield accuracy, precision, and sensitivity measurements for each classifier. Models were ranked by accuracy and the most accurate classifiers are shown with the results of pH and Temperature-based classifiers for comparison.bAccuracy = [tp+tn]/ [tp+tn+fp+fn], where tp is true positives, tn is true negatives, fp is false positives, and fn is false negatives.cPrecision = tp/[tp+fp].dSensitivity = tp/[tp+fn] (sometimes termed ‘Recall’).eNA indicates that the precision or sensitivity cannot be calculated due to the absence of positive calls.