MATERIALS AND METHODS

The investigation was conducted under irrigation conditions between 2000 and 2001 years on the experimental area of the Department of Field Crops, Faculty of Agriculture, Yüzüncü Yıl University in Van, Turkey. The soil of the experimental area was clay-loam, pH was 7.7, low in organic matter (1.0%), poor in available nitrogen (0.080 mg L^-1) and phosphorus content (27.5 kg ha^-1), rich in potassium and lime contents (524 kg ha^-1 and 12%, respectively) and at least in salt content (0.080%). The total rainfall was 234.60 and 137.5 mm in the experimental years, compared with the long-term (1965-1995) mean of 412.5 mm. The monthly average temperature (first year 10.3°C and second year 10.9°C) and relative humidity (first year 59.4% and second year 60.1%) means were similar to the long-term average (8.3°C; 65.1%).

Fifteen safflower cultivars were sown in a randomized complete block design with three replications. The plot size for each treatment during 2000 and 2001 was 11.25 m^-2 (5.0x2.25 m) and sown with 45x15 cm row spacing (app.13.3 plant m^-2) in the midts of May in the experiment years. Nitrogenous fertilizer (ammonium sulfate 21% and 150 kg ha^-1) and phosphorus fertilizer (triple super phosphate 42% and 100 kg ha^-1) were applied before sowing and all standard agronomic practices were applied. Samples were obtained in the second week of October during both the years. Agronomic characters were determined on ten plants randomly selected in the mid-rows all of plots. Seed yield (kg ha^-1), plant height (cm), primary branches/plant), heads/plant), branch height (cm), stem diameter (mm); head diameter (mm), seeds/head), 1000-seed weight (g), oil content (%) were measured. The collected data was analyzed through computer TARIST statistical package. In order to determine the relationships between seed yield and the other examined characters simple correlation coefficients were calculated. The path coefficients were separated by using seed yield as a dependent variable (Wright, 1960).

RESULTS AND DISCUSSION

Positive significant relationships were found between seed yield and plant height (r = 0.389**), heads/plant (r = 0.212*), Branch height (r = 0.355**), stem diameter (r = 0.355**), head diameter (r = 0.819**), seeds/head (r = 0.670**), (Table 1). These results showed that any positive increase in such characters will suffice the boost in seed yield. These findings were in similar with the results of Dingming et al. (1993), Omidi Tabrizi, (2005), Patil et al. (2004), Mahasi et al. (2006) and Alizadeh and Carapetian (2006). On the other hand, negative significant relationships were determined between seed yield and primary branches/plant (r = -0.466**), 1000 seed weight (r = -0.220*), oil content (r = -0.428**). On the contrary, Hamadi et al. (2001) reported that seed yield in safflower was a strong positive association between the primary branches. Zheng et al. (1993) stated that the high-yielding safflower varieties have taller individuals lower branches, more effective heads, fewer ineffective heads, lower weight of seeds, higher average heads/plant and longer flowering. Present results confirm the findings of Zheng et al. (1993). The results indicated that improvement of seed yield in safflower could be decrease oil content and 1000-seed weight due to negative association between these traits (Pahlavani, 2005). Bagawan and Ravikumar (2001) stated that oil content and seed yield are negatively correlated and thus an attempt to improve one results in the reduction of the other. These findings were in accordance with the results of our study.

The highest positive correlations were determined between plant height and stem diameter (r = 0.823**), seed yield and head diameter (r = 0.819**), plant height and branch height (r = 0.728**) and head diameter and seeds/head (r = 0.720**). Similarly, Johnson et al. (2005), Omidi Tabrizi (2005), Çamaş et al. (2005) and Yuhai et al. (1993) determined that positive and significant relationships between these components. The results indicated that among all traits that correlated with head diameter, the highest correlation was observed for seed yield (r = 0.819**) and seeds/head (r = 0.720**).

Seeds/head showed positive and considerable correlation with seed yield (r = 0.670**), plant height (r = 0.328**) and branch height (r = 0.508**) and considerable negative correlation with oil content (r = -0.423**), primary branches/plant (r = -0.403**) and 1000-seed weight (r = -0.254**). These results showed that head diameter or seeds/head or both traits could be responsible for high seed yield in safflower (Corleto et al., 1997; Zheng et al., 1993; Abel and Driscol, 1976).

Table 1:	Correlation coefficients among ten characters of safflower
ns: Not significant, *p<0.05, **p<0.001 1SY: Seed yield (kg ha-1), PH: Plant Height (cm); PBN: Primary Branch No. (No./plant); HN: Head No. (No./plant); BH: Branch Height (cm) ST: Stem Diameter (mm); HD: Head Diameter (mm), SN: Seed No. (No./head); SW: 1000-Seed Weight (g); OC: Oil Content (%)

Plant height exhibited a significant and positive correlation with heads/plant (r = 0.234*), but a significantly negative correlation with oil content (r = -0.389**). Present results confirm the findings of Çamaş et al. (2005) for oil content. Seeds/head reveled a insignificant association with heads/plant but expressed a highly negative correlation with 1000-seed weight (r = -0.254*). Mahasi et al. (2006) reported that negatively and significant correlation between seeds/capitulum and 100-seed weight.

Gencer et al. (1987) stated that seed yield and oil content are the primarily selection criteria for safflower breeding. Among the all traits that correlated with oil content, the lowest correlation was observed for heads/plant (Table 1). The oil content was negatively correlated with seed yield, plant height, branch height, stem diameter, head diameter and seeds/head whereas it was positively correlated with primary branches/plant. Pahlavani (2005) found the similar results for seed yield and oil content. Also, Çamaş et al. (2005) found similar results for plant height, branch height and primary branches/plant.

In order to get a clear picture of the interrelationships between different traits, the direct and indirect effects of different characters were worked out using path coefficient analysis in respect of seed yield (Sing et al., 2004). The path coefficient analysis based on seed yield as a dependent variable revealed that all traits, except primary branches/plant and oil content, showed positive direct effects (Table 2). Compared to the simple correlation analysis, path analysis of seed yield and its traits demonstrated that head diameter, heads/plant, branch height and seeds/head evolved the highest direct influence, 46.0, 34.2, 27.4 and 21.8%, respectively. Conversely, primary branches/plant had a negative and high direct effect (43.3%) with an indirect effect via head diameter (21.5%) and seeds/head (9.8%) on seed yield, but a positive indirect effect via heads/plant and branch height. On the other hand, the indirect effects of branch height and seeds/head via head diameter were stronger than its direct effects. These analyses indicated that head diameter, heads/plant, branch height and seeds/head were the main characters to seed yield. For this reason, these traits could be used more significantly for safflower improvement. The present findings of the importance of the direct effects on seed yield are not in agreement with the findings of Patil (1998), who showed the importance of the direct effect of 100-seed weight, heads/plant and primary branches/plant. Singh et al. (2004) also reported the importance of the direct effect of seed weights/head and oil percent. Similar research results with present study were published by others (Patil, 1998; Gupta and Singh, 1997; Iqbal et al., 2006; Hamadi et al., 2001; Corleto et al., 1997; Abel and Driscol, 1976; Omidi Tabrizi, 2001; Mahasi et al., 2006; Ashri et al., 1976).

Although stem diameter and plant height had small positive direct effects on seed yield with 20.1 and 5.2%, respectively, but a great indirect effect via head diameter (respectively, 34.4 and 32.6%). In addition to, the direct effect of 1000-seed weight (10.7%) was positive and small, the indirect effect of this trait, via head diameter (34.5%) and primary branches/plant (23.5%), was negative and high on seed yield. Conversely, oil content had a negative and small direct effect on seed yield, but a great indirect effect through head diameter, primary branches/plant and seeds/head. Similar results were published by Çamaş et al. (2005), Dingming et al. (1993) and Zheng et al. (1993).

It was analyzed that a higher indirect effect was exhibited on head diameter, heads/plant and seeds/head by most of the yield traits. And these traits correlated to seed yield could be given the primarily selection criteria for safflower breeding. Our results are consistent with Patil (1998), Gupta and Singh (1997), Gencer et al. (1987), Ghongade et al. (1993) and Omidi Tabrizi (2005).

Table 2:	The direct and indirect effects of nine yield components to seed yield in safflower (path analysis)1
*p<0.05, **p<0.01, 1values in parenthesis show percentage, 2SY: Seed yield (kg ha-1), PH: Plant Height (cm); PBN: Primary Branch No. (No./plant); HN: Head No. (No./plant); BH: Branch Height (cm) ST: Stem Diameter (mm); HD: Head Diameter (mm), SN: Seed No. (No./head); SW: 1000-Seed Weight (g); OC: Oil Content (%)

CONCLUSIONS

The relationships were determined between the seed yield and some agronomic components of safflower with simple correlation coefficients and path analysis. According to the results of the correlation analysis, seed yield was significantly correlated that all traits. Improvement of some characters can increase safflower seed yield., In this study, path analyses of seed yield showed that head diameter, head number/plant and seed number/head have the highest direct effect. This components had main components to seed yield.