A SIMULATION STUDY OF MICROEVOLUTIONARY INFERENCES BY SPATIAL AUTOCORRELATION ANALYSIS

To explore tile extent to which microevolutionary inference can be mad e using spatial autocorrelation analysis of gene frequency surfaces, w e simulated sets of surfaces for nine evolutionary scenarios, and subj ected spatially-based summary statistics of these to linear discrimina nt analysis. Scenarios varied the amounts of dispersion, selection, mi gration, and deme sizes; and included: panmixia, drift, intrusion, and stepping-stone models with 0-2 migrations, 0-2 selection gradients, a nd mig-ration plus selection. To discover how weak evolutionary forces could be and still allow discrimination, each scenario had both a str ong and a weak configuration. Discriminant rules were calculated using one collection of data (the training set) consisting of 250 sets of 1 5 surfaces for each of the nine scenarios. Misclassification rates wer e verified against a second, entirely new set of data (the lest set) e qual in size. Test set misclassification rates for the 20 best discrim inating variables ranged from 39.3% (weak) to 3.6% (strong), far lower than the expected rate of 88.9% absent any discriminating ability. Mi sclassification was highest when discriminating the number of migratio nal events or the presence or number of selection events. Discriminati on of drift and panmixia from the other scenarios was perfect. A subse quent subjective analysis of a subset of the data by one of us yielded comparable, although somewhat higher, misclassification rates. Judgin g by these results, spatial autocorrelation variables describing sets of gene frequency surfaces permit some microevolutionary inferences. ( C) 1997 The Linnean Society of London.