Correlations among morphological traits of sweet chestnut (Castanea sativa Miller) from Bosnia and Herzegovina

In this research, the following traits of sweet chestnut fruits were measured: fruit length (FL; in mm), fruit width (FW; in mm), fruit thickness (FT; in mm), hilum width (HW; in mm), hilum length (HL; in mm) and fruit weight (FWG; in g). The following traits were assessed according to the international descriptor for sweet chestnut (UPOV, 1989): fruit shape (FS); fruit embryony (FE) and penetration of seed coat into embryo (PSCiE). The following parameters were calculated (according to the method given by Furonez-Pérez and Fernández-López 2009): FL/FW ratio, relative size of the hilum (RSH): HW × HL.

FWG was measured in a fresh condition immediately after fruit collection, using an analytical balance with an accuracy of 0.01 g. Other traits were measured using a digital caliper with an accuracy of 0.01 mm.

The following traits of sweet chestnut leaves were measured: leaf blade length (LBL; in mm), leaf blade width (LBW; in mm) and leaf petiole length (LPL; in mm). The following traits were assessed according to the international descriptor for sweet chestnut (UPOV, 1989): leaf blade shape (LBS) (sharp, blunt, serrated) and leaf edge shape (LES) (mucronate, dentate). The following parameters were calculated according to method given by Serdar et al. (2011): total leaf length (TLL) (TLL = LBL + LPL), LBW/LBL ratio and LBW/TLL (blade + petiole).

Measurements were performed using a graph paper with an accuracy of 1/10 mm.

The following traits of sweet chestnut cupules were measured: cupule width (CW; in mm) and cupule height (CH; in mm). The following traits were assessed according to the method given by Pereira-Lorenzo (1996): needle length (NL) (long, medium, short) and number of fruits in cupule (NfiC). The following parameters were calculated according to the method given by Serdar et al. (2011): relative size of cupule (RSC) (CW × cupule height CH).

The measurement of these traits was performed using a digital caliper with an accuracy of 0.01 mm.

A linear correlation was calculated using Pearson correlation coefficient r. All investigated traits of fruits, as well as all investigated traits of leaves and all investigated traits of cupules were mutually compared. Data processing was done using the packages XLSTAT 2011 and STATISTICA 10.

The correlation coefficient (r) was determined by mutually comparing each investigated fruit trait with each The obtained results are shown in Table 2. Statistically significant and highest values of the correlation coefficient are marked in red. The values of the correlation coefficient for the fruit traits range from r = −0.521 to 0.923 for individual compared traits. Statistically significant positive and high values of the coefficient of variability were found in all measured fruit traits (FL, FW, FT and FWG, as well as HW and HL), which is very important in the breeding of sweet chestnut. In contrast to the measured traits, the assessed trait of FE showed a statistically significant positive correlation only with the trait FS (r = 0.070) and did not represent a significant trait for sweet chestnut selection and breeding. The highest values of correlation coefficient were found in most of the investigated fruit traits. For the compared traits, the highest values of the correlation coefficient were found for the following: relative hilum size and HW (r = 0.923) and relative hilum size and HL (r = 0.915). Statistically significant negative correlations with the highest value of the correlation coefficient were found in the following compared traits: FS with FW (r = −0.521) and FS with HL (r = −0.403). Traits that were neither in positive nor in negative statistically significant correlation were: FS and HL (r = −0.009), FE and FL (r = 0.007), FE and FW (r = −0.014), FE and FT (r = −0.031), FE and HW (r = −0.012), FE and HL (r = −0.012), FE and fruit mass (r = −0.023), PSCiE and FL (r = −0.006), PSCiE and FW (r = −0.029), PSCiE and HL (r = −0.028), FS and FE (r = 0.030) and relative hilum size and FE (r = −0.015).

The correlation coefficient (r) for leaf traits was determined by comparing the investigated leaf traits with each other, and the obtained results are shown in Table 3. Statistically significant and the largest positive and negative values obtained are marked in red. The values of the correlation coefficient for the leaf ranged from −0.621 to 0.989 for individual compared leaf traits. The highest positive and statistically significant value of the correlation coefficient was shown by the traits of LBL and LBW (r = 0.616), and LBW/LBL and LBW/TLL, where the correlation coefficient was 0.989. The highest negative value of the correlation coefficient (−0.621) was shown by the traits of LBW/TLL and LPL.

Leaf traits which had a low value of the correlation coefficient (around zero) were LES and LBL (r = −0.001), LES and LBW (r = −0.012), LES and TLL (r = −0.019), LBW/LBL and LES (r = −0.022) and LBW/TLL and LES (r = −0.001).

Table 4 shows the correlation coefficients (r) among all investigated traits of the cupule. Statistically significant correlations are marked in red. Statistically significant and high coefficients of positive and negative correlation of the compared traits of the cupule are shown in Table 4 and marked in red. The value of the correlation coefficient for the investigated traits of the cupule ranged from r = −0.069 to 0.935. All compared traits showed positive values of the correlation coefficient, which indicates that the increase of one variable affects the increase of another variable, for example, the size of the cupule is correlated with CW, CH and NfiC. The exception is the trait of NL, which was negatively correlated with CW and NfiC. The largest positive and statistically significant values of the correlation coefficient were found for the following traits: cupule size and CH (r = 0.935) and cupule size and CW (r = 0.916). The trait of NL showed a statistically significant negative correlation with NfiC (r = −0.069). The correlations between NL and CW (r = −0.042) and cupule size and NL (r = 0.007) were not statistically significant.

The correlation coefficient (r) for fruit and leaf was determined by comparing the investigated traits of fruit and leaf with each other, and the obtained results are shown in Table 5. Statistically significant and highest values of the correlation coefficient are marked in red. The values of the correlation coefficient for fruit and leaf range from r = −0.624 to 0.989 for individual compared traits. The highest statistically significant values of the correlation coefficient were found between the measured fruit traits (FL, FW, FT, FWG, HL and HW) and measured traits of the leaf (LBW, LPL), while the shape of the leaf base showed a statistically significant correlation only with FS and FE. Statistically significant and highest positive correlations were found in the measured fruit traits and LPL, while statistically significant negative values of the correlation coefficient were shown by the measured fruit traits and LBW. Statistically significant negative correlations with the highest value of the correlation coefficient were found in the compared traits: HL and LBW (r = −0.187), FW and LBW (r = −0.186), relative hilum size and LBW (r = −0.182), FL and LBW (r = −0.178) and FWG and LBW (r = −0.170). A trait that did not show a statistically significant correlation for most of the compared fruit traits was the leaf base shape, except with the FS (r = −0.053) and FE (r = −0.037).

Table 6 shows the correlation coefficients (r) between the compared traits of fruit and cupule, with statistically significant correlations marked in red. The value of the correlation coefficient for the compared traits of the fruit and the cupule ranges from r = −0.138 to 0.210. The highest statistically significant values of correlation with the investigated traits of fruit showed the trait of the cupule NL, except with the trait of FE where no statistical significance was observed. The CH trait did not show statistically significant correlations with the investigated fruit traits, except for the penetration of the seed coat into the cotyledon and the FL/FW ratio. The relative size of the hilum showed a statistically significant negative correlation (r = −0.064) only with the FL/FW ratio, while no statistically significant correlation was observed with other fruit traits.

The correlation coefficients (r) for leaf and cupule obtained by comparing each investigated leaf trait with each investigated cupule trait are shown in Table 7, with statistically significant correlations marked in red. The traits of LBL, leaf base shape, LES and TLL did not show a statistically significant correlation with cupule traits. Statistically significant and positive correlations were found for LBW with NfiC (r = 0.064) and LBW/TLL with NfiC. The trait LPL showed a negative and statistically significant correlation coefficient with CW, NfiC, cupule NL and relative hilum size.