Bunch component performance of AVROS and Yangambi population

The IOPRI oil palm breeding program has developed a total of 11 varieties through two selection cycles since 1974, using the RRS scheme. The performance of the varieties from the first and second cycles in the AVROS and Yangambi groups was compared using data from progeny test bunch analysis. The AV-02 variety from the second cycle of AVROS line showed an increase in fruit/bunch (F/B) and oil extraction rate (OER) by 0.92% and 0.50%, respectively, compared to the AV-01 variety from the first cycle. The AV-03 variety increased mesocarp/fruit (M/F) to 4.92% and OER to 1.21%. However, M/F and OER enhancement was compensated by kernel/fruit (K/F) and shell/fruit (S/F) decrease by 2.35% and 2.57%. In the Yangambi group, the YA-02 variety from the second cycle had an improved average weight bunch (AWB) of 4.00 kg or 21.4% greater than the YA-01 variety from the first cycle. The YA-02 AWB improvement was followed by an increase in M/F and K/F by 0.87% and 0.30%. The YA-03 variety had better kernel characters with a rise in K/F and S/F of 0.54% and 0.44%, and OER still increased by 0.57%, which may be due to its oil/dry mesocarp (ODM) increase. CART® analysis showed that the F/B, ODM, and M/F were the three highest predictors for OER values for all varieties, except for YA-02 with F/B, S/F, and M/F, and YA-03 with M/F, F/B, and S/F. ODM is an important consideration in the assembly of high-yielding varieties, as it is one of the three major predictor characters that affect OER values which highly inherited.


Introduction
The IOPRI breeding program with the Reciprocal Recurrent Selection (RRS) scheme started in 1974.The first variety from the first cycle was released in 1985, and the second cycle varieties in 2003-2010.The total varieties produced are 11 varieties.Most of the varieties today are produced from AVROS and Yangambi groups.IOPRI has documented bunch analysis data of various progeny testing for the last three decades.Bunch analysis is part of the selection criteria to determine which parents will be selected as sources to assemble high-yielding oil palm planting materials.
The RRS scheme calculated the performance of the parental populations through the performance of their progeny produced by utilizing the heterosis trait in the crossed palms [1].In other words, commercial varieties performance is reflected by their progeny performance.Bunch analysis data from all progeny tests can be used to compare the performance of various varieties, even though a commercial varieties demonstration block has not yet been established.
The bunch component characters have varying heritability values.This heritability value will determine the direction and selection criteria for the next breeding cycle.Several bunch characters such as fresh fruit bunch, bunch number, palm and kernel oil yield, kernel/fruit, and kernel/bunch were reported to have high heritability values [2].Sitepu et al. [3] stated that characters of mesocarp/fruit (M/F), oil extraction rate (OER), kernel/fruit (K/F), kernel/bunch (K/B) and shell/fruit (S/F) were more dominated by their pisifera parents based on their heritability values.Suppose the heritability value of the characters obtained is low.In that case, the selection criteria can be carried out through pedigree analysis by calculating the actual breeding values of the parents [4,5].
This study compares the bunch component performance of each AVROS and Yangambi group based on their selection cycle.In addition, an analysis of the predictors of bunch components that most influenced OER was carried out.Data analysis was limited to bunch component data from progeny tests without involving heritability or breeding values of their parental sources.The results of the bunch components analysis are expected to describe the bunch characteristics of each variety produced from the AVROS and Yangambi group from the first and second cycles.

Materials and Methods
The data collected from bunch analysis of progeny trials DxP/T of AVROS and Yangambi group varieties from the IOPRI's RRS first and second cycles from 6 to 9 years after planting.Data were collected from various experimental locations in North Sumatra, Indonesia.The total was about 1,500 bunch analysis data per variety after normalization.The data then were tested for comparison of two samples with a confidence interval (CI) = 0.95%.Comparisons were made for the character values of average bunch weight (ABW), ratio percentage of fruit/bunch (F/B), mesocarp/fruit (M/F), kernel/fruit (K/F), shell/fruit (S/F), oil/dry mesocarp (ODM) and oil yield/oil extraction rate (OER).In addition, after data standardization, Classification and Regression Trees (CART®) analysis was carried out using the statistical software Minitab 21 to predict the bunch component that most influenced OER for each variety.

Bunch components performance of each AVROS and Yangambi varieties.
The higher ABW for the AVROS group first cycle variety became a prominent characteristic compared to the second cycle varieties.However, in second-cycle varieties from the AVROS group, the oil yield characters were higher by 0.5 and 1.21% compared to the first-cycle varieties.AV-03, the most recently released from the second cycle variety, had a significantly higher M/F character of 4.92%, even though it had the smallest ABW (table 1).The high value of M/F was compensated by a decrease in the value of K/F and S/F, which reached > 2%.Higher OER and M/F mean that the character varieties released in the second cycle focused on higher oil yield values than their predecessors.However, ABW was not as large as ABW in first-cycle varieties.Varieties from the Yangambi group showed higher kernel and shell percentage characteristics than varieties from the AVROS group.Improvement on the second cycle aims to maximize ABW (YA-02) and CPO and PKO values (YA-03).The YA-02 variety had 4.00 kg heavier or 21.4% higher than the YA-01 variety from Yangambi's first cycle (Table 2).YA-02 variety also increased M/F and K/F by 0.87% and 0.30%, even though the OER value decreased by 1.09%.This AWB character is very much liked by oil palm smallholders who are indeed focused on AWB.Nonetheless, for oil palm business stakeholders focusing on CPO and PKO products, the YA-03 variety is assembled by considering its K/F, S/F, and OER values.Even though it is lighter in AWB, the YA-03 variety has significantly better K/F, S/F, ODM, and OER values compared to its predecessor YA-01 variety, by an average of 0.5%.The results of the CART® analysis showed that for the AVROS and Yangambi group varieties, almost all OER values were most influenced by F/B values, except for AV-03, whose OER values were most affected by M/F.The three highest predictors for OER values for all varieties were F/B, ODM, and M/F characters, except YA-02, where the three highest predictors were F/B, S/F, and M/F and YA-03, with the three highest predictors namely M/F, F/B and S/F.In the AVROS group varieties, relative percentages of the tested predictor characters were <60% to the main predictor characters (Figure 1).Meanwhile, in the Yangambi group, the relative percentage of predictors was > 60% of the main predictor characters (Figure 2).This shows that in the AVROS group varieties, the value of the main predictor character strongly influences the OER value.

Discussion
The purpose of the progeny test in the RRS scheme for oil palm breeding program is to evaluate the general combining ability (GCA) and specific combining ability (SCA) of crossings between parents.The GCA and SCA values determine the parents used as mother palms for breeding superior oil palm plant materials.Selection progress can be assessed by involving performance value on the parental test.In this study, only the progeny testing performance was analyzed, assuming that this value represents the DxP performance of the commercial variety.
AV-01 and AV-02 varieties are AVROS/SP540T lineage selected based on the best GCA value.AVROS/SP540T is known to have good GCA in the characteristics of bunches and oil production [6].The assembly of these two varieties has a number of dura and pisifera lines since it is based on GCA.Value of F/B and OER traits in the VA-02 variety was significantly increased by the second cycle's selection of dura and pisifera lines (Table 1).AV-03 variety is an AVROS group variety with the best SCA value.This variety has a distinctive character with high M/F and OER values compared to AV-01 and AV-02 (Table 1).The effect of dominance, epistasis, or both [7,8] of M/F character is significant in the AV-03.
YA-02 and YA-03 are Yangambi group second-cycle varieties assembled based on SCA with certain characters.YA-02 has significantly high bunch weight characters, and YA-03 has suitable kernel and shell character values.This YA-02 large bunch character is favored by oil palm smallholders who depend on ABW for their income.However, for oil palm industries stakeholders focusing on their CPO and PKO products, the YA-03 is chosen considering their K/F, S/F, and OER values.The pisifera characters used to assemble these two varieties also show similar characteristics.Sitepu et al. [3] mentioned that the influence of the pisifera character contributed significantly to these two varieties.Judin et al. [9] also reported that the Yangambi progeny tested had the highest bunch weight and oil in FELDA.
OER, or ratio of oil to bunches, is the main component used as a reference in assessing productivity, in addition to the number and weight of bunches [10].OER values are influenced by intrinsic factors, namely parental genetic characteristics [3,[11][12][13] and extrinsic factors such as weather, rainfall and rainy days, water deficit and processes at the mill [14][15][16][17].Woittiez et al. [18] summarized this well in their published review of the factors influencing palm oil yields.
In this study, the intrinsic characteristics of the bunches were analyzed for all AVROS and Yangambi group varieties to see which characters strongly influenced the OER score.The F/B character is essential in influencing the OER value (Figures 1 and 2).This is in line with research by Harun and Noor [19], which states that oil yields increase as F/B increases and have a positive correlation.Although the strength varies for each variety, the M/F and ODM characters are generally enough to determine the OER value.In contrast, the ABW contribution was insignificant to the OER value for each variety.
The F/B value is closely related to the fruit set of bunches.The fruit set value is influenced by several extrinsic factors, such as water [14], nutrition ,and pollination/pollinator [20,21].Genetically, the F/B value has a low heritability value based on SNP and SSR marker tests compared to ODM and S/F values [22].Therefore, the ODM value is an essential consideration in the assembly of high-yielding varieties because ODM is one of the major predictor characters that affect OER values.

Conclusion
The second cycle varieties from the AVROS group have a significant increase in OER of 0.5 and 1.21% compared to the first cycle.The distinctive characteristic of the second cycle of AVROS is increased mesocarp thickness to fruit.The second cycle variety from the Yangambi group has distinctive characteristics ABW value is 21.4% higher than the first cycle variety (YA-02).The K/F, S/F, and OER values are 0.5% higher (YA-03).The bunch component that acts as the main predictor of OER is F/B.The main predictors for OER scores are F/B, M/F, and ODM.The ABW value has the lowest value for the OER predictor.

Figure 1 .
Figure 1.The relative importance of the OER value predictor characters to the main predictor characters in AVROS first and second cycle group varieties.Note:When a predictor is split, character importance measures model advancement.As a percentage improvement compared to the top predictor, relative relevance is defined.

Figure 2 .
Figure 2. The relative importance of the OER value predictor characters to the main predictor characters in Yangambi first and second cycle group varieties.

Table 2 .
Comparison of bunch components of Yangambi line varieties from first and second cycles 3.1.2.Predicting bunch component that affects OER value.