Map Construction

Maps are prepared using computer pro- grams to identify loci in linkage groups and then determine the order of the loci and dis-

tances between them. Linkage groups are generally identifi ed by computing pairwise tests for evidence of linkage at a range of criteria, usually the LOD score. The LOD score is a ‘log of odds’ to the base 10, so that a LOD score of 2 means that it is 100 times more likely that the data are explained by linkage at the most probable recombination fraction, than that they are explained by no linkage. A locus is considered a member of a linkage group if it shows signifi cant link- age (at the current LOD value) to any other member of the group. When the LOD score is low (say 1–2), all loci may fall into a single or a few linkage groups. Generally, with error-free data that meet the assump- tions of the analysis, a LOD score of 3–4 will identify a number of linkage groups equal to the number of chromosomes. The number of larger linkage groups will be fairly stable at this value, gradually increas- ing as the LOD score is increased. Linkage groups identifi ed at scores of 3 or 4 are gen- erally used for mapping, but higher values may be appropriate for some data sets.

Once linkage groups are identifi ed, the loci within each group are ordered and dis- tances between loci are computed. These calculations, and locus grouping, are per- formed by computer programs. Commonly used programs for construction of linkage maps are MAPMAKER (Lander et al., 1987) and JOINMAP (Stam and Van Ooijen, 1995). MAPMAKER performs multipoint linkage analysis, an approach with great power. The major limitation of this program for analysis of citrus maps is that it can analyse only a backcross or F2-type segregation.

This means that 1: 1: 1: 1 loci must be

recoded with a loss of information, and that maps that combine backcross and F2-type segregation patterns cannot be prepared. The MAPMAKER interface is command line (DOS or Unix), making it somewhat diffi - cult to use. JOINMAP uses multiple two- point linkage estimates, an intrinsically less powerful approach, but it allows analysis of all types of segregation patterns. JOINMAP also allows maps developed in different populations with shared markers to be merged. JOINMAP 3.0 has a modern

Windows interface and is considerably easier to use than MAPMAKER, but it is a commercial product that requires purchase of a licence, whereas MAPMAKER is free. It is probable that additional mapping programs will become available, so those conducting such projects should carry out an Internet search for linkage mapping programs to identify suitable programs.

Another issue that needs to be addressed in citrus mapping is marker phase. This refers to the actual physical chromosome on which allelic markers are located. For example, dominant markers can be in coupling (A B/a b) or repulsion (A b/a B) phase. If the grandparents of the mapping population are known, then the marker phase is also generally known. However, in many citrus crosses, the par- ents are not hybrids of known ancestry. In such situations, the phase must be inferred from the data. JOINMAP performs this analy- sis, while with MAPMAKER data of unknown phase must be entered in both possible phases, linkage groups identifi ed and one member of each pair of linkage groups dis- carded. If known, the phase can be speci- fi ed by the allele code: for a/b-type allele

codes, all a alleles are assumed to be on one homologue, and b alleles on the other. For some programs, allele position in the geno- type code indicates phase. For example, for two loci if the parents are specifi ed as ab ´ cd and ab ´ ab, then the chromosomes of the two parents are a a/b b and c a/d b.