*2.5. Microsatellite Analysis*

Laboratory techniques for DNA extraction were performed as described by Peterson. Amplification reactions were carried out in 15 uL reaction volumes containing 30 mg genomic DNA, 1.0 μM each of SSR primers sequences, which were drawn from the following sources: BNL primers from the Research Genetics Co. (Huntsville, AL, USA, http://www.resgen.com, accessed on 7 April 2022); JESPR primers [37]; CIR primers [37]; and NAU primers [38,39], 100 uM each of dATP, dCTP, dGTP, and dTTP, 1 unit of Taq DNA Polymerase (Fermentas), 1x*Taq* Polymerase Buffer, and 2.5 mM MgCl2. PCR amplifications were performed as described [40] using a Peltier Thermal Cycler (MJ Research, Waltham, MA, USA) programmed as follows: an initial denaturation of 5 min at 94◦; 35 cycles of 94◦ for 1 min (denaturation), 55◦ for 1 min (annealing), and 72◦ for 2 min (extension). One additional cycle of 10 min at 72◦ was used for final extension. The amplified products were electrophoresed on a 10% non-denatured polyacrylamide gel using a DYCZ-30 electrophoresis apparatus (Beijing WoDeLife sciences instrument company, Beijing, China).

### *2.6. QTL Mapping*

Genetic mapping and QTL analysis were performed on each population separately and combined across populations. Linkage maps were constructed using MAPMAKER/Exp Version 3.0b software [41]. QTLs were identified by composite interval mapping [42] using Windows QTLs Cartographer 2.5 [43]. A LOD threshold of 3.0 was used [44]. Marker's order was confirmed with the "ripple" command. Recombination frequencies were converted into map distances (cm) using the Kosambi mapping function [45].

Tests for independence of QTLs were also conducted when 2 or more QTLs of a trait were located on the same chromosome [46]. QTLs were declared significant if the corresponding LR score were greater than 11.5 (equal to a LOD score of 2.5). The proportion of the phenotypic variation explained by each QTL was calculated as R2 (%) = Phenotypic variability explained by QTL/all of the variation in the population × 100. The total phenotypic variance explained together by all the putative QTLs for each trait was estimated by fitting a multiple-QTL model in the Mapmaker/QTL program.
