Table 2.

Performance of JoinMap 3.0 with high-density linkage data.^a.

Dataset	LOD threshold	Unincorporated markers	Map inflationb	Split linkage groups
Complete	1.0	15 ± 14% (range: 0–54%)	70 ± 76% (range: 1–256%)	0%
Complete	2.0	14 ± 12% (range: 1.2–36%)	61 ± 82% (range: 5–279%)	38%
Binned	1.0	0.8 ± 1.8% (range: 0–7%)	43 ± 71% (range: 1–269%)	0%
Binned	2.0	0.2 ± 0.2% (range: 0–0.2%)	2.8 ± 2.3% (range: 0–7.5%)	38%

^a All values reported are means ± SD across 21 linkage groups (7 chromosomes × 3 populations) or a subset of them in case there was insufficient linkage information for map construction. The populations used for this test were Clipper/Sahara, Dayton/Zhepi2 and Steptoe/Morex. Only DArT markers were used, in order to disentangle the potential effect of (hypothetical) DNA sample tracking errors from software performance. The LOD threshold was varied, while the other program settings were held constant (REC = 40; ripple after each locus; threshold for marker removal = 5). Decreasing the REC threshold had a similar effect as increasing the LOD threshold (data not presented).

^b Map inflation values are the percent increase in SARF for the locus order reported by JoinMap compared to the order reported by RECORD. If JoinMap did not incorporate all loci into a map, only the set of incorporated loci were analyzed with RECORD. For unknown reasons, heavily inflated maps with a large number of misplaced loci were often reported by JoinMap to be shorter than maps constructed from the same datasets using JoinMap settings that produced a close-to-optimum locus order.