Cowpea is a self-pollinating annual diploid with chromosomes 2n = 2x = 22 belong to the Fabaceae family. It is one of the improved legume extensively grown in India and Africa (Boukar et al., 2020). Asia’s top cowpea-growing countries include India, Bangladesh, Myanmar, Sri Lanka, China, Korea and Thailand. In India, cowpea is grown across various states, including Karnataka, which is one of the notable contributors to its cultivation. In Karnataka, the crop is cultivated over an area of 78,446 hectares, producing a total output of 35,759 tonnes, with an average productivity of 0.46 t/ha (Indiastat, 2022). It is resilient to tolerate a wide range of environmental stressors, making it the one of the most preferred crops for small-scale farmers. Considering that production of crops including cowpea in highly vulnerable small holding rainfed ecosystem is challenging, development and deployment of high yielding cowpea cultivars resilient to biotic and abiotic stresses with specific adaptation is critical for sustainable cowpea production by resource poor farms (Reddy et al., 2025). Cowpea not only has a limited genetic base but is also experiencing genetic erosion, posing additional challenges to conventional breeding efforts (Tripathi et al., 2019; Horn and Shimelis, 2020). Limits to our understanding of the type of genetic architecture underlying trait differentiation also include the relative contribution of cytoplasmic effects. Geneticists and breeders have recognized reciprocal effects as one source of genetic variability (Mann and Pollmer, 1981).
       
Segregating populations derived from planned crosses are most widely used sources of new pure line cultivars in predominantly self-pollinated crops with no exception of cowpea. Biparental crosses are most widely used for deriving segregating populations. Direction of crosses in terms of which of the two parents to be used as male/female is critical in maximizing useful genetic variability. We hypothesized that there exist significant differences in segregating populations derived from reciprocal crosses. To test this hypothesis, the objective of our study was to examine a possible influence of reciprocal cross effects on grain yield and some quantitative traits in cowpea.

The material consisted of four genotypes [IC488271, IC263015, IC608044, IC590843] with manifested substantial variability for quantitative traits (Table 1). These four genotypes were used to effect three sets of reciprocal crosses. Four parental genotypes were used to generate three pairs of reciprocal crosses were affected namely C1=IC488271×IC590843, C2=IC590843×IC488271, C3=IC263015×IC590843, C4=IC590843×IC263015, C5= IC608044×IC590843 and C6=IC590843×IC608044 during early kharif 2023. IC590843 was employed as the common parent in all crosses owing to its green bold seed size, which is a desirable trait for enhancing grain yield and market acceptability in cowpea. The F₁’s of three pairs of crosses were raised and hybridity was confirmed using male parent-specific traits in each cross like seed size and seed coat color. The true F₁’s of six crosses was selfed. The self-seeds were planted in subsequent season to raise three sets of F₂ populations during Rabi 2023-24 at the Experimental plots, University of Agricultural Sciences (UAS), GKVK, Bengaluru (13°05′16.4"N 77°33′55.2"E). After 15 days of planting, seedlings were thinned to maintain 0.2 m between plants within a row. All the recommended agronomic practices were adopted to raise a good crop of cowpeas. The size of three sets of F₂ populations ranged from 101 to 137 (Table 2). The data were recorded on 5 randomly selected plants of parents and all the F₂ plants of six crosses on pod length (cm), pods plant^-1, grain yield plant^-1 (g) and 100 grain weight (g).




 
Data analysis
 
Data recorded on individual F₂ plants were used to estimate descriptive first-degree statistics such as mean, absolute range (AR), standardized range (SR) and second-degree statistics such as absolute phenotypic variance (σ²_p), phenotypic standard deviation (σ_p) and usefulness index using the following formulae. All the analysis were implemented in ‘R’ statistical software, version 3.5.2 (R Core Team 2024). The distribution of F₂ populations were graphically plotted using R software.
 
Arithmetic mean

Standardized range (SR) was estimated as

The significance of trait means between the three sets of reciprocal crosses (F₂) were examined using two-sample ‘t’ test.

Where,
X₁ = Mean of direct cross.                
X₂= Mean of reciprocal cross.                    
n₁= Number of plants in direct cross.
s₁= Standard deviation of direct cross.
s₂= Standard deviation of reciprocal cross. 
n₂= Number of plants in reciprocal cross.
 
Transgressive segregation index (TSI)
 
TSI was estimated as the proportion of the difference between the parents and the corresponding range for a phenotype in the segregating (F₂) populations. Based on descriptive statistics, TSI was estimated (Koide et al., 2019) as:

Levene’s test for homogeneity of variances
 
Levene’s test (Levene, 1960) was used to test if σ²_p of reciprocal F² populations were comparable. 
 
Usefulness index (UI)
 
UI accounts for trait mean, phenotypic variability and broad-sense heritability and thus provide a more comprehensive and informative statistic in terms of genetic gain expected when different intensities of selection were imposed in segregating populations (Allier et al., 2019; Lehermeier et al., 2017; Bernardo, 2020).
 

Uc = Trait mean + (k × " σ²_g "/"σ_p")   

 
Where,
k= Standardized selection differential at different selection intensities.
k= 2.67 and 2.06 at 1 and 5 per cent selection intensities, respectively (Bernardo, 2020).
σ²_g = Genotypic variance.
σ_p = Phenotypic standard deviation.
       
The ‘σ_g’ was estimated as the square root of σ²_g; The σ²_g was estimated at σ²_p - σ²_e, where, σ²_e for F₂ populations was estimated as the average of phenotypic variance in non-segregating populations (F₁, P₁, P₂).

Criteria to assess the reciprocal cross differences in F₂ segregating populations
 
Significant trait means, standardized range, variance and UI between reciprocal F₂ populations derived from three crosses were interpreted as significant reciprocal cross differences. 

The success of plant breeding hinges on the level of genetic variability within the working germplasm and/or in the segregating generations derived from carefully chosen parents. Although biparental crosses are commonly employed to develop segregating populations, the number of parents involved can vary. We hypothesize that even the direction of both bi- and multiparent crosses could markedly influence the resulting genetic variability. In this manuscript, we refer to segregating populations as breeding populations (BPs). From a plant breeding perspective, the most promising BP is one that shows both a high mean for the target trait and substantial genetic variance. A higher mean ensures a head start, while greater variance provides scope for effective selection. In this background, the findings of the present investigation are discussed.
       
It is evident from the results that straight and reciprocal crosses were comparable for means of all the four traits, though trait means per se varied with the cross combinations (C1/C2, C3/C4 and C5/C6). Further, means of all the four traits of cross combination namely C5/C6 were significantly greater than the other two cross combinations (Fig 1).  These results suggest that there were no significant reciprocal cross differences for trait means. These findings are consistent with those reported by Kovačević et al. (2022) in maize, where the maternal inbred line did not exhibit statistically significant differences between the two variants, indicating a minimal reciprocal effect.


       
As is true with respect to trait means, estimates of standardized range (SR) of all the four traits were also comparable between straight and reciprocal crosses of all the cross combinations (C1/C2, C3/C4 and C5/C6). However contrary to trait mean, estimates of SR of the cross combination namely C5/C6 was significantly lower than other two cross combinations (Table 3). Considering that without transgressive segregation (TS), plant breeding does not work; plant breeding does work and because TS occurs (Mackay et al., 2020). The estimates of TSI were relatively close between members of each reciprocal cross pair (C1/C2, C3/C4 and C5/C6), indicating comparable TS for all the traits. While C1/C2 displayed higher TSI than other crosses for grain yield, C3/C4 exhibited higher TSI for pods plant^-1; C5/C6 showed higher TSI for pod length and 100 grain weight (Table 3). Theoretical investigations have indicated that TS results from dispersion of favourable complementary alleles between the parents from which BPs are derived (Bernardo, 2020; Mackay et al., 2020). These theoretical studies suggests that alleles that increase grain weight plant^-1 are dispersed between parents (IC488271 and IC590843) of C1/C2 cross, while those that increase pods plant^-1 are dispersed between parents (IC263015 and IC590843) of C3/C4 cross; those that increase pods length and 100 grain weight are dispersed between parents (IC608044and IC590843) of C5/C6 cross.


       
Quantifying the variability for target traits is critical in addition to means to make a better-informed choice about the selection of BPs (Vaggar et al., 2022). Crop breeders target highly variable segregating populations as they offer ample opportunities for selection and genetic improvement (Ongom et al., 2021). The estimates of absolute phenotypic variance of all the traits were comparable between all the cross combinations (Table 4). Further, C5/C6 were more variable than C1/C2 and C3/C4 for pods plant^-1 and grain yield plant^-1. These results are comparable to those observed in maize (Pollmer et al., 1979; Arellano-Vázquez et al., 2023). However, the proportion of genetic variance attributable to reciprocal differences was relatively small (Melchinger et al., 1985; Carena, 2005; Easterly et al., 2020; Dermail et al., 2023).


       
The estimates of UI were also comparable between straight and reciprocal cross combinations at 1% and 5% selection intensities (Fig 2). From the aforesaid discussion, it is evident that best BPs differed with the criteria. For example, based on SR and variance, C1/C2 and C3/C4 were better than the other cross. On the other hand, the cross C5/C6 was better than the other two cross combination based on mean and UI. Thus, the cross combination C5/C6 could be regarded as promising as per as probability of deriving pure-lines better than their parents is concerned. Further, it appears that direction of the cross is not significant for deriving BPs. Thus, the present results did not support our hypothesis of significant differences between reciprocal crosses.

Reciprocal crosses between the populations created hybrid seed with the same nuclear genetic composition, but different cytoplasmic genes. The genetic analysis of F2 pairs showed no statistically significant reciprocal effects for key grain yield traits. This suggests that either direction of the cross could be used to generate variability for grain yield traits and any parental line can serve as either the maternal or paternal component in a cross combination. For that reason, breeders should take the reciprocal effects in consideration in planning crossings strategies. This knowledge might be useful in deciding about the sequence of parents to exploit the reciprocal differences to an advantage. However, if no significant reciprocal differences are detected in cowpea, hybridization programmes should instead emphasize other critical factors such as the genetic divergence between parents, combining ability for yield and adaptive traits, heterosis potential, heritability estimates and stability of performance across environments. 

The senior author gratefully acknowledges Council of Scientific and Industrial Research (CSIR), New Delhi, India for providing Junior Research Fellowship (JRF) vide No. 09/0271(17220)/2024-EMR-I for pursuing PhD degree program at University of Agricultural Sciences, Bangalore.
 
Disclaimers
 
The views and conclusions expressed in this article are solely those of the authors and do not necessarily represent the views of their affiliated institutions. The authors are responsible for the accuracy and completeness of the information provided, but do not accept any liability for any direct or indirect losses resulting from the use of this content.

The authors declare that they have no conflict of interests or personal relationships that could have appeared to influence the work reported in this paper.