A: Sketch of a neutral coalescent tree with tree size . B and C: A selective sweep in locus C leads to a tree of low height ( small). The selective sweep was initiated by a beneficial mutation at time . At some distance from C, a single lineage (circled branch in C) has “recombined away” leading to the unbalanced tree shown at locus B. Note that tree height between trees B and C changes drastically and that at locus C and at locus B. Multiple recombination events (indicated by the crosses at the bottom line) between loci A and B lead to essentially uncorrelated trees at A and B.

Shown are the values for independent realizations. -axis: values of (black circles) and (red squares) are determined for the subtrees originating at node , . The solid gray line shows the theoretical expectation according to .

Correlation based on simulations ( replicates) of the statistic of the true tree. Pearson's correlation coefficient is measured between and for pairs of trees at position and position . Three scenarios are compared: standard neutral model with constant population size (green), population bottleneck (blue) and selective sweep (red). Sample size , , and a recombinaton rate of is assumed. The bottleneck parameters are: , . The selective sweep has a strength of . The selected site is at position . Under standard neutrality, correlation is reached at position cM, corresponding to about bp.

A: estimation of by . B: estimation of by . First row: standard neutral model. Second row: Selective sweep; estimation of at distance from selected site. Third row: Selective sweep; distance from selected site. Parameters: ; (top and bottom row); (middle row); ; ; .

Plots in column A show the distribution of , i.e. when the tree topology is known. Plots in column B show the distribution of the estimate when the tree topology is unknown, but estimated from microsatellite polymorphism data. Each boxplot corresponds to one of marker loci located at the positions indicated on the axis. The regions spans kb in total. Symmetric step-wise mutation model with . Other parameters: , and recombination rate per bp (corresponding to 1 cM/Mb). First row: standard neutral model with constant . Second row: bottleneck model with severity and onset . Third row: Selective sweep at locus with which was completed time units ago. For comparison with the theoretical expectation, the leftmost boxplot in each panel shows the standard normal distribution (labeled ‘N’).

Parameters: level (solid) and . (dotted); selection coefficient ; time since fixation ; sample size ; mutation rate ; recombination rate . The axis shows positions to the left (negative values) and right (positive values) of the locus under selection at position . Scale is in cM x, corresponding here to kb.

of tests (stars), (circles) and (triangles) applied to a kb region sorrounding the pfmdr1 locus in P.falciparum. Shown are significant results on the 5% (open symbols) and 1% (filled symbols) levels. Positions of the examined microsatellite markers are indicated by arrows. Data from .