Responses to selection and changes in combining ability after three cycles of a modified reciprocal recurrent selection in maize

This research reports responses to selection and changes in general (GCA) and specific (SCA) combining abilities after three cycles of a modified reciprocal recurrent selection procedure (MRRS) in EPB-4 and EPB-5 maize populations. In the MRRS procedure a cycle can be completed in 1 or 2 years depending on the availability of winter breeding nurseries. The original and the three selection cycles of the two populations per se (eight entries) and the partial diallel developed from the crosses between them (16 entries) were evaluated in six environments. Realized response to selection on the population cross was 7.25% cycle⁻¹ for grain yield, −13.63% cycle⁻¹ for plant lodging, and 11.93% cyle⁻¹ for prolificacy, whereas plant and ear heights remained unchanged. GCA estimates increased with selection cycles for both populations for grain yield and prolificacy, and decreased for plant lodging, indicating that the frequency of favorable alleles with additive effects for these traits increased with the MRRS cycles in both populations. SCA estimates increased for grain yield and prolificacy indicating that the frequency of favorable complementary alleles at loci with non-additive effects in the reciprocal populations increased with the MRRS cycles. For grain yield, SCA effects increased more than GCA effects with selection cycles, indicating that MRRS exploited more the non-additive effects than the additive effects for the improvement of this trait. The overall results showed that the MRRS procedure was highly effective in improving the population cross, exploiting both GCA and SCA effects.     

Changes in genetic diversity during seven cycles of recurrent selection for grainyield in oat, Avena sativa L.

Variability for desirable alleles within elite breeding populations is a requirement for long-term genetic improvement. Changes in genetic diversity were evaluated in a recurrent selection programme for enhanced grain yield in oat, Avena sativa L., using coefficient of parentage (r_p) and restriction fragment length polymorphism (RFLP) based genetic similarity (S) estimates. Both r_p and s increased during seven cycles of recurrent selection, indicating a reduction in the level of genetic diversity within the population. The relationship between genetic diversity and agronomic performance was also examined. Genetic similarity measures were poor predictors of the near-homozygous progeny performance and general and specific combining ability effects for three agronomic traits in the recurrent selection programme. In addition, there was no apparent trend for preferential selection of progeny from either more similar or more diverse parents within a given cycle of selection. The overall reduction in genetic diversity in this population has not affected selection response for grain yield. Variability remaining at important loci or de novo variation are possible explanations for the continued selection progress.     

不同轮回选择方法对玉米窄基群体的改良效果

利用轮回选择进行群体改良,是玉米种质扩增与改良的有效方法,能为选育优良自交系提供基本素材,进而提高选育自交系及杂交种的效率。本研究以玉米窄基群体P4C0及其经过不同轮回选择方法改良的10个群体为材料,通过多点田间表型鉴定和配合力测定,研究不同轮回选择方法对玉米窄基群体的改良效果,利用SSR分子标记分析不同轮回选择方法对群体遗传多样性的影响。结果表明,几种轮回选择方法都能有效改良群体的主要性状及其一般配合力(GCA)。以时间计算,控制双亲混合选择(MS)对群体P4C0主要性状及其GCA改良效果优于半同胞-S2:3(HS-S2:3)轮回选择,但在株高和穗位高的改良上,HS-S2:3选择效果较好。以轮次计算,开放改良对群体P4HSC1主要性状及其GCA的改良效果优于MS,但开放改良后,群体株高和穗位高有较大幅度的增加。不同轮回选择方法对群体遗传多样性和遗传结构的影响不尽一致。P4C0经过5轮MS后,在群体改良的低代,群体遗传多样性得到了较好的保持,而多代的改良导致群体遗传多样性下降;P4C0经过1轮HS-S2:3选择后,遗传多样性比P4C0有较大幅度的下降。P4HSC1经过1轮开放改良后,遗传多样性有较大幅度的增加。P4HSC1经过3轮MS改良后,群体遗传多样性呈增大的趋势,但每轮增加的幅度均较小。     

Recurrent selection of cocoa populations in Côte d'Ivoire: comparative genetic diversity between the first and second cycles

In Côte d'Ivoire, the cocoa breeding programme has been based on the creation of hybrids between different genetic groups. From 1990 onward, a reciprocal recurrent selection programme has been set up with the purpose of improving simultaneously the characteristics of the two main genetic groups: Upper Amazon Forastero (UA) and a mixture of Lower Amazon Forastero (LA) and Trinitario (T). Based on data obtained from 12 microsatellite primers, the genetic diversity and genetic distances of the parental populations used in the first and second selection cycles are presented. The results revealed that the diversity of populations UA0 and UA1 on the one hand and (LA+T)0 and (LA+T)1 on the other is similar. The genetic distances were small between the parental populations used for the first and second cycles. Genetic diversity was greater in the UA group than in the LA+T group. The number of rare and of private alleles was reduced for both genetic groups, as well as the number of the frequent alleles in the LA+T group.     

基于SSR标记的平果金花茶的遗传多样性和遗传结构分析

平果金花茶(Camellia pingguoensis D.Fang)为濒危植物,分布于广西平果县和田东县。我们在其分布区采集了7个野生代表种群,共216个个体进行实验,目的是利用微卫星分子标记方法对平果金花茶的遗传多样性和遗传结构进行探究,并基于研究结果对平果金花茶提出保护策略建议。研究结果表明,8对微卫星引物共检测到104个等位基因,平均每对引物检测到13个。平果金花茶的平均等位基因(Na)、有效等位基因(Ne)、期望杂合度(He)和观测杂合度(Ho)分别为5.4、2.995、0.512和0.557。AMOVA结果显示74.69%的变异存在种群内,种群间的遗传分化值(Fst)在0.1331~0.3666之间。STRUCTURE结果显示K=7为最佳分组,即每个种群各自成组。平果金花茶具有中等程度的遗传多样性和高水平的遗传分化。因此,在制定保护策略时,应尽可能对多个种群进行保护。     

利用微卫星标记分析建鲤种质资源的研究

为了评估和合理利用建鲤种质资源。从建鲤繁育群体中随机选取185尾鱼,用20个微卫星位点进行遗传多样性分析。结果表明：20个微卫星位点在该群体中所检测到的等位基因片段长度在114~316 bp,共检测出156个等位基因和402种基因型,各位点等位基因数为5~13个,平均7.8个,基因型数10~44种,平均20.1种;各位点观察杂合度(Ho)、期望杂合度(He)分别在0.346~0.978和0.619~0.880之间,平均分别为0.6434和0.757;所检测的20个位点多态信息含量(PIC)在0.552~0.868之间,平均为0.7253,都属于高度多态位点(PIC>0.5)。实验结果表明,该建鲤繁育群体多态信息含量丰富,遗传多样性水平较高,具有较大的选育潜力。群体内平均固定系数(FIS)为0.1479,说明该建鲤群体存在杂合子缺失现象。根据个体间的遗传距离构建的聚类图可以清楚地显示每个个体之间的遗传差异,这可为建鲤保种和繁殖配组提供依据,避免近交现象。     

Genetic Diversity of Farmers’Preferred Sorghum Accessions and Improved Lines from ICRISAT Reveal a Disconnect Between Innovation and Technology Transfer

The genetic diversity of 65 accessions of sorghum [Sorghum bicolor (L.) Moench] collected from various farmers and germplasm lines from ICRISAT-Kenya were analyzed. Simple sequence repeats (SSR) markers were used in order to determine the extent and distribution of its genetic diversity. Twenty-nine (29) SSRs markers were polymorphic and a total of 192 alleles were detected which showed diversity. The number of alleles per primer ranged from 2 to 17, with an average of 6.62. The range of polymorphism information content (PIC) ranged from 0.03 to 0.86, with total average of 0.82. According to the results analyzed, estimates of the mean allelic pattern across the two populations was generated; expected heterozygosity (He; 0.45, 0.54), average observed alleles (Na; 3.40, 6.20), number of private allele (0.23, 3.03), and Shannon information index (I; 0.85, 1.13) for farmer and ICRISAT-Kenya germplasm, respectively. The expected heterozygosity (He) varied from 0 to 0.26 with an average of 0.05. The Neighbor-joining phenogram based on Nei’s genetic distance grouped the 65 accessions into three main groups. The analysis of molecular variance (AMOVA) revealed that 99% of the total genetic variation was within accessions in a population whereas the genetic variation among populations in accessions accounted for 1% of the total genetic variation. Genetic diversity in ICRISAT sorghum material compared to the farmer’s collection suggested little infiltration of improved germplasm to the farmers.     

Genetic diversity evolution through participatory maize breeding in Portugal

Natural, and in particular, artificial (human) selection may pose a danger to the existing crop genetic diversity. Nevertheless, on-farm breeding systems seem to achieve phenotypic improvements even though preserving variability. Using SSR markers, we analysed several selection cycles, over a 20 years period, of a Portuguese on-farm participatory maize OPV-‘Pigarro’ breeding project. No significant differences in allelic richness (N_ar), observed heterozygosity (H_O), expected heterozygosity (or gene diversity; H_E) or inbreeding coefficient (f) were detected among the selection cycles. 58 out of 107 alleles were common to all the selection cycles studied. The analysis of molecular variance showed that the variation among selection cycles represented only 7% of the total molecular variation. However, the number of private alleles varied among the selection cycles, being the highest detected at the beginning of the selection project. These findings demonstrate that an allele flow took place during the on-farm selection process of ‘Pigarro’ but the level of genetic diversity was not significantly influenced. Since interesting phenotypic improvements were also achieved, on-farm breeding projects, like this one, should be valued as a way to preserve unique Portuguese maize landraces in risk of disappearing.     

Genetic diversity among populations and breeding lines from recurrent selection in Brassica napus as revealed by RAPD markers

Recurrent selection facilitated by dominant male sterility has been conducted to broaden the genetic basis for cultivar development in Brassica napus. This study aimed to evaluate the genetic variation in four base populations (C0‐C3) and breeding lines from two of the populations produced during recurrent selection by random amplified polymorphic DNA (Rapd) markers. Genetic variation in four populations declined gradually with the advance of selection cycles as measured by expected genetic heterozygosity (from 0.2058 in C0 to 0.1536 in C3) but the decline was not statistically significant. When compared with the average genetic distances for 21 germplasm collections with wide geographical and genetic origins (0.4712) and seven breeding lines from pedigree selection (0.2059), seven breeding lines selected from the C1 population and 11 from the C3 population had a larger average genetic distance (0.5339 and 0.5486, respectively). Clustering analysis indicated that the lines from recurrent selection had a much lower genetic similarity than lines from pedigree selection. Our results suggest that base populations derived from recurrent selection could provide a wider genetic variation for selection of breeding lines with more broad genetic bases.     

楸树EST-SSR标记开发及种质资源的遗传多样性分析

利用转录组数据开发SSR标记是一种经济高效的DNA分子标记开发策略。本研究利用高通量测序技术开发楸树(Catalpa bungei)EST-SSR标记,了解其在转录组序列中的分布类型,并利用41对多态性较好的引物对来自安徽(AH)、河南(HN)、湖北(HB)和山东(SD)的4个楸树居群的48个无性系的遗传多样性进行分析。结果表明:在大于1kb的14634条序列中,共鉴定出3999个SSR位点,580条序列包含2个及以上位点;单核苷酸重复(1957,48.94%)是最常见的SSR类型,其次是二核苷酸重复(1164,29.11%),三核苷酸重复(834,20.86%),四核苷酸重复(41,1.03%)和五核苷酸重复(3,0.08%);共扩增出243个等位基因,每个位点的等位基因数为3~13个,平均为4.85个,观测杂合度为0.14?1.00,平均为0.63,期望杂合度为0.42?0.84,平均为0.67,大部分位点偏离哈迪-温伯格平衡;AH和HN居群在有效等位基因数和期望杂合度的数值较高,表明其遗传多样性较丰富;分子方差分析(AMOVA)表明,遗传变异主要发生在居群内。本研究为楸树及同属其他树种种质资源鉴定、遗传多样性提供依据,且在指导楸树良种选育等方面具有重要的意义。     

Originality of M3S maize population and changes in allele frequencies revealed by SSR markers after two cycles of selfed progeny recurrent selection

Maksimir 3 Synthetic (M3S) maize population was developed at the Faculty of Agriculture University of Zagreb by intercrossing inbred lines, whose origins trace back to several open-pollinated varieties and local populations from different regions of the former Yugoslavia. The population was subjected to two cycles of selfed progeny recurrent selection for grain yield. The objectives of this study were: (i) to determine genetic distances among the parental inbred lines of the M3S population (M3S progenitors), the M3S population before and after two cycles of recurrent selection, and elite inbred lines representing the BSSS and Lancaster heterotic group; and (ii) to examine the effect of two cycles of recurrent selection on allele frequency changes in the population. Nine M3S progenitors, three BSSS lines, and three Lancaster lines were genotyped at 24 SSR loci, out of which nine randomly chosen loci were used for genotyping 96 individuals from both C0 (the M3S population before selection) and from C2 (M3S population after two cycles of selection). A total of 101 alleles were detected across 24 loci in the 15 lines, whereas 83 alleles were found in the nine M3S progenitors. Among the latter 83 alleles 31 were unique, i.e. found only in one of the progenitors. Mean genetic distance among nine M3S progenitors was 0.61 indicating a broad genetic base of the M3S population. High mean genetic distance was found between M3S progenitors and BSSS lines (0.69) and M3S progenitors and Lancaster lines (0.71). This indicates that the M3S population represents a germplasm source unrelated to both the BSSS and Lancaster germplasm. Mean genetic distance between the M3S population and BSSS as well as Lancaster lines decreased slightly after two cycles of recurrent selection suggesting the need to introduce testers from both groups in future selection in the M3S population in order to maintain heterotic complementarity of the M3S population to these groups. A test of selective neutrality identified several non-neutral loci in the population whose allele frequency changes from the C0 to the C2 cannot be explained by genetic drift. The majority of non-neutral alleles, whose frequency increased after two cycles of selection, were present in at least one line from the BSSS or Lancaster heterotic group.     

Evaluation of three cycles of full-sib reciprocal recurrent selection in two maize populations from the Northeast of Spain

In Europe a heterotic pattern commonly used in maize breeding is “American Dent × European Flint”. Maize breeding programs generally use only a small portion of the useful genetic variability present in the local open-pollinated varieties which, otherwise, have a poor performance that hampers their use. Two maize composites, EZS33 formed by open-pollinated flint varieties from dry or Mediterranean Spain, and EZS34 from USA dent populations, were developed in Zaragoza (Spain). Both were subjected to three cycles of full-sib reciprocal recurrent selection (RRS) for yield. The objective of our study was to evaluate after three cycles of selection the effect of RRS. The selection significantly increased yield in the population crosses (3.0 % per cycle) and in the crosses of the populations with testers of different origin. The correlated responses for the population crosses in other agronomic traits like flowering and early vigor were in the desired direction, but plant height was reduced. We propose that stover yield or related traits could be included in selection programs as selection criteria to prevent their decline. The adapted population increased the frequency of favorable alleles for yield, but the inbreeding depression counteracted their effect on the mean. RRS had a positive effect on other traits. According to our data, RRS can be useful to develop improved populations from which it would be possible to develop lines with improved both specific and general combining ability with different heterotic groups. We conclude that the heterotic pattern “Mediterranean Spain × US Dent” is potentially very useful for maize breeding for adaptation to Mediterranean conditions and an interesting source of cultivars for low-input agriculture.     

Development of a recurrent apple breeding programme in New Zealand: a synthesis of results,and a proposed revised breeding strategy

This paper describes an apple breeding population that has been developed in New Zealand since 1990 to broaden the genetic base for cultivar breeding. The population is now in its third generation, with recurrent selection for general combining ability being the underlying approach. During the 20 years of the programme, nearly 1,000 trees have been progeny tested. An early focus on selection within open-pollinated tests has been replaced by selection within control-pollinated tests. Genetic evaluation methods along with estimates of genetic parameters (e.g. heritability, genetic correlations, and genotype-by-environment interactions), genetic gain, and genetic diversity are summarised in this paper. The third-generation crossing programme involving nearly 350 parents is almost completed and seedlings from these crosses are in the nursery. In the revised breeding strategy, three hierarchal populations are envisioned to better integrate this broad-based population with the more commercially focused cultivar breeding programme: main breeding population (MBP), elite breeding populations (EBP) and cultivar development populations (CDP). A two-subline structure and a minimum of two crosses per parent will be implemented in the MBP. EBP will focus on developing parents for breeding goals of the CDP. Forward-selected (based on estimated breeding values) candidates from the main- and elite populations will be used as parents of the CDP. Cloning of seedlings in elite populations and also in the first-stage of cultivar development, will be investigated as a tool for enhancing predictability and shortening the cycle for delivering genetic gain.     

Divergent Selection for Heading Date in Barley

Divergent selection for heading time was performed in two F₂ barley population. Five populations were obtained for each cross: F₅ SSD (unselected control), 3E and 3L (from three cycles of selection for earliness or lateness), 2E1L and ZL1E (from two cycles of direct and one of reverse selection). These populations, together with corresponding parents and F₂ generations, were evaluated over two years. The response to selection was 5.6 and 6.5 days inane direction (earlier heading) and 7.7 and 6.7 clays in the other direction (later heading) in the two crosses, respectively. 3E and 3L populations were highly transgressive as compared to their parents. A sizeable amount of genetic variability was still present after two cycles of selection. Heading was probably controlled by a polygenic system with prevailing additive effects and alleles for earliness and earliness somewhat equally distributed in the parents. Selection fur heading time led to significant changes in plant height, yield and kernel weight. Early progenies were higher yielding than lace ones.     

Reciprocal recurrent selection in maize enhances heterosis and ears yield

The objective of the research was to evaluate recurrent selection strategies in maize (Zea mays L.) by determining the expected and realized responses to selection, as well as the alterations in heterosis, of hybrids from the cross between the original populations (C0) and the third cycle (C3). The genetic variances and covariances were obtained for intrapopulational (P₁₁ and P₂₂) and interpopulational (P₁₂ and P₂₁) half-sib progenies from populations 1 and 2 in cycles C0 and C3. One-hundred progenies of each type were evaluated [husked ear weight (g plant^?1)] at two separate locations by conducting 10 × 10 triple lattice design experiments contiguous for each cycle. Four different recurrent selection strategies were investigated: reciprocal recurrent selection (RRS), interpopulation testcross half-sib selection (THS 1 and THS 2) and intrapopulational half-sib recurrent selection (HSS). THS 1, which employed population 1 (with the lowest frequency of favorable alleles) as tester, was most efficient in the simultaneous improvement of the hybrid and the two populations per se. RRS showed the highest efficiency in improving the interpopulational hybrid but not in the parental populations. In order to estimate heterosis and realized responses to RRS, representative samples of seed from intra and inter-populational half-sib progenies from C0 and C3 were employed in randomized block design experiments with 10 repetitions carried out at two separate locations. RRS increased heterosis in the interpopulational hybrid from 12.3 % (C0) to 24.9 % (C3), and the productivity gain [husked ear weight (t ha^?1)] was 13.5 % over the three cycles.     

Levels and distribution of genetic diversity of coconut (Cocos nucifera L., var. Typica form typica) from Sri Lanka assessed by microsatellite markers

The coconut variety Typica, form typica, commonly known as Sri Lanka tall coconuts is the most widely exploited and grown variety in Sri Lanka. Under the coconut bio-diversity conservation programme, several Typica populations have been collected by island-wide surveys and planted ex situ. Thirty-three coconut populations were subjected to microsatellite assay with eight coconut-specific microsatellite primer pairs in order to study the levels and distribution of genetic variation of the collected materials for formulating future collection strategies and selecting parents for the breeding programme. A total of 56 alleles were detected ranging from 3 to 10 alleles per primer pair with an average of 7 alleles per locus. Overall a very high level of genetic diversity was detected (0.999) for all the populations studied ranging from 0.526 for population Debarayaya to 0.683 for population Dickwella. Only four introduced coconut populations, i.e. Clovis, Margeret, Dickwella, Mirishena and an embryo-cultured population were clearly separated from the resulting dendrogram. A very high level of within population variation (99%)accounted for native populations suggests a common history and a restricted genetic base for native Sri Lankan tall coconuts. Categorization of alleles into different classes according to their frequency and distribution confirmed the results of the dedrogram and concluded the adequacy of single large collection from the entire target area to represent the total genetic diversity in Sri Lanka. This study discusses useful information regarding conservation and breeding of coconut in Sri Lanka. This revised version was published online in August 2006 with corrections to the Cover Date.     

Genetic diversity and population structure of Indian soybean [<Emphasis Type="Italic">Glycine max</Emphasis> (L.) Merr.] revealed by simple sequence repeat markers

Genetic diversity in 90 Indian soybean cultivars was assessed using 45 SSR markers distributed on 20 soybean chromosomes. Forty-five SSR markers generated 232 alleles with an average of five alleles/locus. The observed frequencies of the 232 alleles ranged from 0.01 to 0.94 with an average of 0.19. The polymorphic information content (PIC) value of the SSR markers varied from 0.10 to 0.83 with an average of 0.61 and about 71% markers have a PIC value of >0.5. In this study, 54 rare alleles including 19 genotype specific alleles were also identified. The observed hetrozygosity for SSR markers ranged from 0 to 0.11 with a mean of 0.10. Cluster analysis grouped the 90 soybean cultivars into three major clusters and principal coordinates analysis (PCoA) results were similar to those of the cluster analysis. A combination of eight SSR markers successfully differentiated all 90 soybean cultivars. The population structure analysis distributed the 90 soybean genotypes into two populations with mean alpha (α) value of 0.1873. In AMOVA analysis, proportion of variation within population was high (88%), whereas only 12% occurred among populations. In cluster and structure analyses, most of the genotypes with similar pedigree were grouped together. Soybean cultivars DS228, MACS-13, LSb-1, Hardee, Improved Pelican, and Pusa-24 were the six most genetically distinct cultivars identified. The study reported a moderate genetic diversity in Indian soybean cultivars and findings would be useful to the soybean breeders in selecting genetically distinct parents for a soybean improvement program.     

基于果实品质与EST-SSR标记的刺梨居群遗传多样性分析及核心种质构建

通过分析评价野生刺梨(Rosa roxburghii Tratt.)资源居群遗传多样性和遗传结构,同时构建核心种质,为刺梨资源的保护和发掘利用提供科学依据。本研究利用10对EST-SSR引物和9个果实品质性状指标对收集的12个刺梨自然居群(共102份种质)的遗传多样性及遗传结构进行分析,同时结合原贵州省32份初级核心种质,采用位点优先取样策略进一步构建西南地区核心种质。结果表明,黔西(QX)居群拥有最高的Shannon信息指数I=0.6965,基因多样性指数h=0.3935,多态位点百分率p=84.62%;而古丈(GZ)居群则无论在分子数据还是表型数据都表现为遗传多样性最低;AMOVA分析表明刺梨居群内遗传变异在87%以上,居群间基因流Nem在3.5以上、平均Nei’s遗传距离(GD)0.223。构建的19份核心种质等位基因保留率和稀有等位基因保留率均为100%,能够代表原种质的遗传多样性。西南地区野生刺梨的遗传变异主要发生在居群内,居群间具有基因交流频繁、Nei’s遗传距离小等特点,构建的19份核心种质从等位基因保留率、稀有等位基因保留率及地理分布均能够较好地代表原种质的遗传多样性。自然居群以黔西(QX)居群遗传多样性最高。因此,野生刺梨的保护策略可采用就地保护黔西(QX)居群与迁地保护19份核心种质相结合的方法进行。     

Improvement of maize populations for resistance to downy mildew

Upgrading levels of disease resistance are a primary objective of maize breeding programmes. Efficacy of S₁ recurrent selection in improving levels of resistance to downy mildew (DM) infection was assessed in Nigeria from 1997 to 2000 in six maize populations. Improvement procedures consisted of evaluating S₁ progenies under artificial infection with DM spores and in disease‐free environments and using a selection index to combine selection for reduced DM infection with appropriate agronomic characters from more than one environment. Three to four cycles of selection were completed in each of the populations. Products from the different cycles of selection were evaluated and data collected on DM infection parameters and agronomic traits. Result obtained showed 3–4 cycles of selection were adequate to reduce DM infection levels significantly and increase grain yield. Downy mildew infection decreased by between 58 and 100% while grain yield increases ranged from 10 to 98% for the 2‐4 cycles of selection relative to the C0 (original). Selection increased grain yield with acceptable changes in plant height while maintaining maturity in disease‐free environments.     

Genetic divergence in two tropical maize composites after four cycles of reciprocal recurrent selection

Two tropical maize composites were subjected to four cycles of reciprocal recurrent selection to develop divergent inbred lines with good combining ability. This study was conducted to examine the extent of genetic diversity, changes in allele composition and genetic structure, of 100 randomly selected S₁ lines each from the original (C₀) and advanced (C₄) selection cycles of TZL COMP3 and TZL COMP4, genotyped using single nucleotide polymorphism (SNP) markers. Results revealed that the proportion of alleles at both low and high frequencies decreased from C₀ to C₄, whereas those at intermediate frequencies increased at C₄ in the two composites. More unique and other alleles were lost at C₄ in TZL COMP3 relative to those in TZL COMP4. The changes in different measures of genetic diversity were either small or negligible with selection in the two composites. The proportion of markers departing from Hardy–Weinberg equilibrium (HWE) decreased with selection, whereas the total number of pairs of markers in linkage disequilibrium increased with selection in the two composites. Examination of changes in population structures using a model‐based approach as well as cluster and multivariate analyses found a high degree of concordance in stratifying the 400 S₁ lines into four non‐overlapping groups corresponding to the two selection cycles each within the reciprocal composites. The observed molecular‐based divergence between cycles within the same composite and the clear differentiation between the complementary composites highlight the importance of reciprocal recurrent selection for preserving genetic diversity for long‐term selection. This increases the potential of the advanced selection cycles to sustain genetic gain in productivity of hybrids adapted to the savannas in West and Central Africa.