Sixteen-year-old “Fuyu” trees (Diospyros kaki), grafted on D. kaki seedlings, were used in this experiment. The trees were spaced 6 × 5 m apart and grown on flat land (35˚28'29'' N, 128˚71'78'' W) with sandy loam soils at Gimhae, South Korea. Annual rainfall of the area ranges 1200 to 1500 mm and annual mean temperature 12.3˚C to 14˚C. Conventional fertilization and irrigation recommended by RDA (2013) [13] were applied to the trees. For this experiment, Three similar sized trees were selected and their flower buds were thinned in mid-May to leave one bud in a below 20-cm long shoot and two buds in an over 20-cm long shoot. Unmarketable fruits were removed from early July to harvest after physiological fruit drop had finished. Since physiological fruit drop differed depending on the position of the fruit in the tree canopy, the leaf-to-fruit ratio per 1-year-old branch varied from 10 to 20.

All fruits were harvested on November 12 and were divided into six size classes. These were 161 - 180 g, 181 - 210 g, 211 - 230 g, 231 - 250 g, 251 - 280 g, and 281 - 320 g. Ten fruits per size class were randomly sampled from each tree, and all 180 fruits from the three trees were used to measure individual fruit characteristics. The weights, lengths and diameters of the fruits were recorded first, and then skin color was measured as Hunter a value using a colorimeter (CM-2500d, Konica Minolta Sensing Inc., Japan). A fruit firmness tester (53,205, Turoni srl., Forlì, Italy) equipped with a 5-mm plunger was used to measure flesh firmness. Fruit juice was collected from sliced mesocarps using a hand presser, and the soluble solids were determined by a digital refractometer with an accuracy of ± 0.2% (PAL-1, Atago Co., Tokyo, Japan). Seeds were numbered from the cross section of the fruits. The dry matter ratios of the fruits were calculated from fresh weight of some flesh slices and dry weight of the slices dried at 80˚C for 48 h.

The dried samples used to measure the dry matter ratio were ground to pass through a 20-mesh. To determine total N, 0.2-g sub-samples were analyzed by a Kjeldahl instrument (Kjeltec 2300, Foss Tecator AB, Höganäs, Sweden) using the micro-Kjeldahl method [14] . P, K, Ca, and Mg were analyzed using an inductively coupled plasma emission spectrometer (ICPS-7510, Shimadzu Co., Tokyo, Japan) after digesting 0.5 g of the samples with HClO₄ and H₂SO₄ in a heating block.

Three single-tree replicates for each size class were arranged in a completely randomized design. Differences among the data were detected by Duncan’s multiple-range test using the SAS System for Windows V8 (SAS Institute, Inc., Cary, NC, USA). Regression analysis was also performed in order to determine the relationships between fruit size and other fruit characteristics and mineral nutrient concentrations.

Table 1 shows the characteristic of the fruits from the different size classes after

^zMean values in each column with the same letter are not significantly different by Duncan’s multiple range test at P ≤ 0.05.

they had been harvested on November 12. The L/D ratio and flesh firmness significantly decreased as fruit size increased. The L/D ratio gradually changed from 0.71 for 161 - 180 g size class to 0.66 for 281 - 320 g size and flesh firmness fell from 30 to 24.2 N, respectively. Large fruits had a redder skin color with more seeds than small ones. As fruit size increased from 161 - 180 g to 281 - 320 g, Hunter a value of the fruit skin increased from 32.9 to 37.1 and the number of seeds increased 2-fold more. Although soluble solids were higher in larger sized fruit, this difference was not statistically significant. Fruit size did not affect the dry matter ratio. Table 2 shows that there was a strong negative relationship between fruit size and the L/D ratio (R² = 0.741) and firmness (R² = 0.604). Skin color and the number of seeds were positively related to fruit size, especially skin color (R² = 0.771). Although not significant, there was a weak positive relationship between fruit size and soluble solids.

Previous studies suggested that increasing fruit size by reducing the fruit load could advance fruit coloration and maturity [1] since carbohydrate availability increased [4] [5] . However, the fruit characteristics in this study might be affected by other factors as well as fruit load. Different light interceptions within a tree canopy could change physiological fruit drop, which would alter fruit load on the branches [15] [16] . The different light environments might also affect growth, coloration, and soluble sugar concentration in fruits [15] [17] [18] . It has previously been shown that the number of seeds per fruit could change fruit growth and coloration, since fruit seeds enhance sink activity for photosynthates [8] [16] [19] .

Regardless of the various factors, large fruits in this study might mature faster than small ones, when their higher skin color and lower flesh firmness are considered. The lower L/D ratios of large fruits indicated that the active growth of diameter occurs with increasing fruit weight. Higher skin color in large fruit indicated that the large fruits were produced from branches with better light interception or with low fruit load. Lower firmness in large fruits could be directly related to earlier maturation at harvest [3] [10] and higher cell volumes [20] , compared with the small fruits. In addition, Johnson (1992) [2] observed that the increased proportion of intercellular spaces in large fruits was negatively related to flesh firmness. Insignificant change in soluble solids, despite a clear difference between fruit size and skin color, reflected that the soluble solids were more affected by other factors. The positive relationship between fruit size and the number of seeds (Table 2) confirmed that the number of seeds is positively correlated with fruit growth as previously reported [16] .

Concentrations of P, K, and Ca significantly decreased as fruit size increased (Table 3). The concentrations did not apparently change when fruit size class increased from 161 - 180 to 211 - 230 g. However, when fruit size increased from 211 - 230 g to 281 - 320 g, P decreased from 0.093 to 0.078%, K from 0.99 to 0.94%, and Ca from 0.092 to 0.82%. Therefore, there were highly negative relationships in the linear regression model between fruit size and fruit P (R² = 0.393**), K (R² = 0.446**), and Ca (R² = 0.417**) concentrations (Table 4). On the other hand, increasing fruit size did not affect concentrations of N and Mg in the fruit (Table 3) and there is no significant relationship between fruit size and concentrations of N and Mg.

In persimmon, contents of mineral nutrientsper fruit increase with fruit growth until harvest, but their concentrations gradually decrease as the fruit approaches the maturing stage because they are diluted by rapid fruit swelling [21] [22] . In contrast, there was an increase in N and P concentrations as apple fruit size increased [9] [23] . Wünsche and Ferguson (2005) [3] suggested that low concentrations of some mineral nutrients in large fruits under low fruit load are related to the increase in dry matter due to high fruit carbohydrate levels. In particular, rapidly growing fruits contain lower Ca concentration since the Ca accumulation cannot keep up with fruit expansion, whereas it is more likely to be accumulated in slowly growing fruits [3] [23] . Fallahi et al. (1984) [9] also reported that the flesh Ca concentrations decreased in large apple fruits produced

^NS, *, ** Non-significant or significant at P ≤ 0.05 and P ≤ 0.01, respectively.

^zMean values in each column with the same letter are not significantly different by Duncan’s multiple range test at P ≤ 0.05.

^NS, ** Non-significant or significant at P ≤ 0.01, respectively.

from low fruit load trees. Low Ca concentration in large fruits could reduce fruit storability [10] . Therefore, low flesh firmness in the large fruits (Table 1) could be related to their low Ca concentration. Non-significant changes in N and Mg concentrations of fruit (Table 3) indicated that the larger the fruits, the more they absorbed N and Mg, compared to the other nutrients. However, some previous reports have indicated that fruit N and Mg concentrations were not affected by fruit size [11] [23] .