Genetic architecture of cashew germplasm accessions

The variability and genetic architecture was assessed deploying 13 important quantitative characters of 478 cashew germplasm accessions evaluated and conserved in National Cashew Field Gene Bank, Directorate of Cashew Research, Puttur in India. Considerable variability was observed for all characters and the highest co-efficient of variation (CV) was observed for sex ratio followed by cumulative yield per plant and apple weight. The lowest CV was observed for shelling percentage followed by shell thickness. Frequency distribution patterns showed highly positively skewed distribution for characters such as nut weight, sex ratio, apple weight and apple to nut ratio. Genetically, it is evident that decreasing alleles are in excess and dominant for these characters. Tree spread, kernel weight and cumulative yield per plant showed moderately positively skewed distribution indicating decreasing alleles are in slight excess and dominant. Flowering intensity showed moderately negative skewed distribution indicating the presence of increasing alleles in slight excess and their dominant nature. Tree height, shell thickness, flowering duration, shelling percentage and leaf area showed approximately symmetric distribution indicating increasing and decreasing alleles are in equal proportion and the dominance is ambi-directional. Significant positive correlations with cumulative yield per plant were observed for tree height, tree spread, sex ratio, flowering duration, apple to nut ratio, shelling percentage and leaf area and significant negative correlation for shell thickness. The present germplasm collection represents sufficient number of accessions for both quantitative and qualitative characters in desired direction. However, based on the frequency distribution patterns, it is imperative to collect germplasm with dwarfness, less tree spread, high nut weight, apple weight and high yield.


Introduction
Germplasm in any crop is the basic resource for a breeder to develop improved varieties and hybrids. Hence, robust germplasm collection representing wide array of variability is essential for any meaningful improvement in terms of yield and quality parameters. In many crops, considerable efforts towards germplasm collection and conservation have been documented (Singh and Srivastava, 2004). However, comprehensive evaluation and characterization of this germplasm needs greater attention and it is often discussed in scientific circles that proper assessment of the worth of the germplasm and its subsequent utilization in crop improvement program is very crucial. This would also facilitate systematic exploration for trait specific germplasm.
The objective of this study was to assess variability and the genetic architecture of cashew germplasm conserved in National Cashew Field Gene Bank, Directorate of Cashew Research (DCR), Puttur, Karnataka, India. Since establishment of National Research Centre for Cashew (NRCC) at Puttur in 1986 and its up gradation to DCR during 2009, vegetatively propagated material of cashew has been collected through joint surveys by teams consisting of scientists of DCR and the All India Coordinated Research Project (AICRP) on Cashew centers of the states namely Maharashtra, Goa, Karnataka, Kerala, Tamil Nadu, Andhra Pradesh, Orissa and West Bengal (Swamy et al., 1997(Swamy et al., , 1998(Swamy et al., , 2000 and eventually conserved in the National Cashew Field Gene Bank (NCFGB). This gene bank serves as a repository of cashew germplasm to be used for breeding and other purposes.

Material and methods
Evaluation and characterization was undertaken in 478 accessions (up to 2003 planting) after six annual harvests as per Cashew Descriptors (Anonymous, 1986) using 27 quantitative and 41 qualitative characters. The evaluated accessions are active collections in the field gene bank with 4 softwood grafts per accession planted at 4m x 4m spacing under rainfed conditions by adopting recommended package of practices. Colour of mature cashew apple was recorded by referring "RHS colour chart" (Anonymous, 1995). In the present paper, 13 important quantitative characters such as tree height, spread, nut weight, sex ratio, apple weight, shell thickness, flowering duration, apple to nut ratio, shelling percentage, kernel weight, leaf area, cumulative yield per plant and flowering intensity, were recorded following experimental manual on cashew (Thimmappaiah et al., 2005) and used for analysis. The mean, standard deviation, skewness and kurtosis were calculated and frequency distribution was arrived for the selected characters using Descriptive Statistics option of IBM SPSS Statistics version 20. Further, correlation among these quantitative characters was worked out using 'correlate' option in the same software.
An attempt was made to identify accessions having desirable magnitude of these quantitative characters. The number of accessions in each category of 7 desirable qualitative characters was identified. The desirability and importance of the selected characters were decided based on previous experience in cashew breeding and the requirements for further improvement.

Mean, skewness, kurtosis and frequency distribution
The mean represented in the Table 1 are values generally expected in cashew for respective characters. Considerable variability for all 13 characters was evident by the respective ranges and CVs. Highest CV (52.2%) was observed for sex ratio followed by cumulative yield per plant (49.2%) and apple weight (37.8%). The lowest CV was observed for shelling percentage (15.2%) followed by shell thickness (16.8%). The higher CV observed for sex ratio and yield indicated that there is wide variation for these characters in the germplasm collection. Hence, it is possible to make selections for these characters as desired level of the character can be chosen. Lowest CV observed for shelling percentage and shell thickness indicated that selection will not considerably improve these characters though they are important for cashew industries.  The frequency distribution patterns of some characters are depicted in figures 1-8. Highly positively skewed distribution (Skewness value >1.0) was observed for characters such as nut weight, sex ratio, apple weight and apple to nut ratio. The characters such as tree spread, kernel weight and cumulative yield per plant showed moderately positively skewed distribution (Skewness >0.5 and <1.0). In the present study, flowering intensity showed moderately negatively skewed distribution (Skewness >-0.5 and <-1.0). For characters such as tree height, shell thickness, flowering duration, shelling percentage and leaf area, the distribution was approximately symmetric (Skewness between -0.5 and 0.5). None of the characters showed highly negatively skewed distribution. Positive kurtosis was observed for most characters except tree height, flowering duration and leaf area, which showed negative kurtosis. The kurtosis for flowering intensity was found to be zero.
Highly positively skewed distribution for characters such as nut weight, sex ratio, apple weight and apple to nut ratio indicated that majority of accessions are having less magnitude of these characters. In other words, it is difficult to find accessions with required higher values for these characters. Based on the skewness patterns, it is possible to infer about the magnitude and nature of alleles governing the character (Fisher et al., 1942). Accordingly, for characters which are highly positively skewed, it is evident that decreasing alleles are in excess and dominant. For characters with moderately positively skewed distribution i.e., tree spread, kernel weight and cumulative yield per plant, it implies that decreasing alleles are in slight excess and dominant. Cruz and Fletcher (2001) observed similar positive skewed distribution of nut yield and nut weight. Further, for flowering intensity, moderately negatively skewed distribution was observed which indicates the presence of increasing alleles in slight excess and their dominant nature. However, tree height, shell thickness, flowering duration, shelling percentage and leaf area showed approximately symmetric distribution. This implies that increasing and decreasing alleles are in equal proportion and the dominance is ambi-directional for these characters. There was no character with highly negatively skewed distribution indicating the lack of presence and hence dominance of increasing alleles. Most characters except tree height, correlated with canopy spread and leaf area. This indicates that improvement in positively correlated yield component can result in improvement of the yield. However, negative relationship between tree canopy and nut yield was observed by Aliyu (2006), Masawe (1994) and Masawe et al. (1998). Further, significant negative correlation was observed between nut yield and apple to nut ratio by Mendez-Natera (2003), flowering duration and nut yield by Rao et al. (2002).
Shell thickness showed significant negative correlation with cumulative yield and there was no significant correlation of characters such as nut weight, apple weight, kernel weight and flowering intensity with cumulative yield. Aliyu (2006) and, Fletcher (1997, 2001) also observed insignificant association between nut weight and nut yield as obtained in this study. However, Northwood (1966) observed negative significant association between these two characters. Ushavani and Jayalekshmy (2009) observed significant negative relationship between apple weight and yield. However, these discrepancies in association between nut yield and other characters may be attributed the population under study in addition to pleiotrophic gene effect and nature of past selection methods (Falconer, 1972).
Significant positive correlation was observed among many other characters of importance. Akin to this study, positive correlation was observed between plant height and plant spread by Azevedo et al. (1998), Tavares et al. (2011) and Samal et al. (2001, nut weight and kernel weight by Rao and Hassan (1956) and Rao et al. (2002), apple weight and nut weight by Mendez-Natera (2003) and

Correlation
Significant positive correlations with cumulative yield per plant were observed (Table 2) for characters such as tree height (0.39), tree spread (0.5), sex ratio (0.17), flowering duration (0.09), apple to nut ratio (0.11), shelling percentage (0.31) and leaf area (0.16). Similar results were obtained where correlation of number of hermaphrodite flowers per panicle with nut yield was positive as shown by Aliyu (2004Aliyu ( , 2006, Rao (1974), Lenka et al. (1999) and Murthy et al. (1984). Rao et al. (2002) observed that nut yield was positively  Samal et al. (2001), canopy spread and apple weight by Samal et al. (2001) apple weight and apple to nut ratio by Mendez-Natera (2003) and, nut weight and shell thickness by Aliyu and Yahaya (2001). Tree spread, nut weight and kernel weight have shown significant correlation with leaf area. Similar positive correlation of leaf size with nut weight was observed by Aliyu (2006). However, tree spread and flowering duration, nut weight and shelling percentage, sex ratio and flowering intensity, shell thickness and shelling percentage, flowering duration and leaf area and, apple to nut ratio and nut weight have shown significant negative correlation. But contrary to current study, Aliyu (2006) reported positive and significant correlation of weight of the whole fruit with nut weight.

Characters to be considered while exploration
Upon classification, it is evident that cashew germplasm collection represents sufficient number of accessions for most characters in the desired direction when one quantitative character is considered (Table 3). Germplasm with some combinations of two and three desirable quantitative characters have been identified (Table 4 and 5). The number of accessions under some qualitative characters has been depicted in Table 6. This will help in selecting germpalsm accessions for specific combination of characters based on breeding objectives. It is evident that few characters to be considered during germplasm collection include dwarfness and less tree spread as there are only five dwarf accessions (<2.5 m height) and four accessions having less than three meters tree spread as per Cashew Descriptors. Further, based on frequency distribution patterns, the germplasm with high nut weight, apple weight, apple to nut ratio, yield, sex ratio, kernel weight, should be given preference while collection. However, identification of trees based on characters such as kernel weight and sex ratio is quite difficult at the field level. Hence, collection based on dwarfness, less tree spread, high nut weight, apple weight, apple to nut ratio and yield is practical and useful.