CORRELATION AND PATH ANALYSIS AMONG QUANTITATIVE TRAITS OF FORAGE SORGHUM GROWN IN THE SEMIARID REGION

– Forage sorghum ( Sorghum bicolor L.) has been grown in diverse climate variations, in warm and dry regions, or in regions with short drought periods. Thus, there is a need to select sorghum genotypes adapted to production systems under drought or rainy irregularities. This study evaluated the direct and indirect correlations among morphophysiological and productive characters of forage sorghum genotypes with Pearson’s correlation and path analysis. The experiment was designed under complete randomized blocks, with 25 treatments (genotypes) and three replicates. Analyzes of phenotypic correlation and the path analysis were done for plant height, panicle length, stem diameter, plant stand, green matter and dry matter yields, stem weight, plant weight, panicle weight, dry matter content, and forage mass. Plant weight presented the most significant direct effect on the forage mass (target variable), contributing to the increase in forage sorghum production. Therefore, heavier plants are more suitable to select forage sorghum genotypes indirectly, aiming to increase forage mass.


2
Forage sorghum (Sorghum bicolor L.) has been grown in a considerable climate variation, in warm and dry regions, or regions with short drought periods due to its morphological characteristics (Ribas, 2000). Considering the enormous adaptability of sorghum (Silva et al., 2022), this crop is suitable for development and expansion in regions with irregular rainfall, besides the successions of summer crops.
For the success of a breeding program of forage sorghum, it is crucial to know the correlations among variables to plan strategies for increasing genetic gains through the selection of many characters simultaneously (Souza et al., 2014). Thereby, in the initial phases of a breeding program, the indirect selection of characters with low heritability, correlated to others with great heritability, is indicated to obtain more significant genetic gains and efficiency in the breeding program (Rios et al., 2012).
In this sense, Pearson's correlation shows the magnitude and direction of the linear association between two characters (Entringer et al., 2014;Crispim Filho, 2018). The nature of this correlation may be phenotypic, genotypic, or environmental, although the genotypic correlation instigates a greater interest in the breeders because of its hereditary nature (Salla et al., 2015). Nevertheless, only the study of correlations does not allow conclusions about causeeffect ratios; that is, the correlation is only a simple measurement of linear association (Souza et al., 2014).
Aiming to overcome this limitation, the method of path analysis allows the breakdown of correlation coefficients into direct and indirect effects, estimated by a regression equation of the characters on a target variable, guarantying more excellent reliability on the choice of characters for selection (Rios et al., 2012;Faria et al., 2015). Based on this context, the study aimed to evaluate the association among quantitative characters of forage sorghum grown in a semiarid region.

Material and Methods
The experiment was carried out at the Experimental Farm from Embrapa Semiárido, located in Graccho Cardoso, SE, Brazil (10º13'06" S, 37º25'13" W, 291 m). According to Köppen's classification, the climate is BSsh, hot climate, Steppe type, and the rainy season occurs in the winter. The experiment was carried out from June 25 to December 9 in the agricultural year of 2019. Climate data were recorded from the meteorological station of Embrapa Semiárido ( Figure 1).
Twenty-five forage sorghum genotypes were assessed, within commercial and experimental ones (Table 1), under a randomized complete block design with 25 treatments and three replications, totalizing 75 experimental plots. Each plot was composed of two rows with five meters, spaced by 0.10 m between plants and 0.50 m between rows, totalizing 50 plants per row, besides a complete stand of 200,000 plants per hectare. The seeding was made manually after the opening of furrows and the distribution of one sorghum seed for each 0.10-m distance. In addition, two seeds were added at the beginning and the end of each row, aiming to avoid bias regarding the border effect.
Initially, the conventional tillage of soil was made with a disc-plow, used at 0.3-m depth.
Moreover, the row markings for the seeding were made using a 3-rod furrower. Then, the furrower regulation was made, aiming that the rods stayed at a 0.5-m distance and 0.05-m depth, approximately.
After that, the experimental units (plots) demarcation was made with ropes, the plots were identified, the Forming fertilization was made simultaneously with the seeding, according to the chemical soil properties (Table 2). First, commercial fertilizer was distributed into the furrows with a formulation of 8-28-16 (nitrogen-phosphorous-potassium) applied by 300 kg/ha. Then, after 40 days from the seeding, 60-g urea was manually applied to each plot row as a nitrogen source for the cover fertilization. Cultural practices were done according to the recommendations and requirements of sorghum cropping management (Ribas, 2000).
At the time of evaluations, the plots were tagged correctly to identify the genotypes and provide a good precision regarding the following variables:    All statistical analyses were made using GENES software (Cruz, 2013).

Results and Discussion
A significant difference among genotypes for all traits was observed for PL and DMY (Table 3).
A significant part of phenotypic variance occurred due to the genotypic one for LW, SW, PW, GMY, ST, PH, PNW, and DMY variables. Therefore, these characteristics could be attributed to genetic expression (Table 4).
= ( ) * 100     CVg did not indicate that these variables would not have genetic gains but suggest that their gains likely will take more time, precisely for the low variability (Araújo et al., 2014).
According to Yokomizo et al. (2016), the methodology developed by Vencovsky & Barriga (1992) points out that CVg/CVe ratio greater than one (1), or near to this, is favorable to obtain immediate

Concerning the coefficients of phenotypic (rP)
and genotypic (rG) correlations, the dissimilarity between some variable pairs was present considering the sign and the magnitude ( Table 5). Because of this dissimilarity, phenotypic correlations were analyzed in detail.
In 96.9 % of total variable pairs, the phenotypic correlations were equal to or higher than genotypic ones. Likewise, 96.9 % of the results were more significant than the environmental correlations. In addition, it was verified that in 66 possible combinations for 12 analyzed variables, no environmental correlations were more remarkable   thus, genotypes with heavier stems tend to be taller, more productive, and have lower dry matter content.
These results suggest that plants with heavier stems tend to have a greater moisture content, influencing silage fermentation (Basso et al., 2012).
The The PH presented a weak and negative correlation with the PNW (-0.25). Thus, how heavier the plant, how heavier the panicle will be. According to Araújo et al. (2014), the reduction of panicle size occurs due to the water deficit, mainly in the plants' flowering and maturity stages. The second and the third components, which had more significant total effects on the FM, were LW (0.79) and PH (0.55), respectively. Thus, taller genotypes with heavier leaves consequently will be