The development of efficient methods for amplifying random DNA sequenc
es by the polymerase chain reaction has created the basis for mapping
virtually unlimited numbers of mixed-phase dominant DNA markers in one
population. Although dominant markers can be efficiently mapped using
many different kinds of matings, recombination frequencies and locus
orders are often mis-estimated from repulsion F-2 matings. The major p
roblem with these matings, apart from excessive sampling errors of rec
ombination frequency (theta) estimates, is the bias of the maximum-lik
elihood estimator (MLE) of theta (theta(ML)). theta(ML) = 0 when the o
bserved frequency of double-recessive phenotypes is 0 and the observed
frequency of double-dominant phenotypes is less than 2/3 - the bias f
or those samples is - theta. We used simulation to estimate the mean b
ias of theta(ML). Mean bias is a function of n and theta and decreases
as n increases. Valid maps of dominant markers can be built by using
sub-sets of markers linked in coupling, thereby creating male and feam
le coupling maps, as long as the maps are fairly dense (about 5 cM)- t
he sampling errors of theta increase as theta increases for coupling l
inkages and are equal to those for backcross matings when theta = 0. T
he use of F-2 matings for mapping dominant markers is not necessarily
proscribed because they yield twice as many useful markers as a backcr
oss population, albeit in two maps, for the same number of DNA extract
ions and PCR assays; however, dominant markers can be more effeciently
exploited by using doubled-haploid, recombinant-inbred, or other inbr
ed populations.