Sunflower (Helianthus annuus L.) occupies a prominent place among oilseed crops, as it contributes approximately 12% to global edible oil production. Water and nutrients play an important role in improving seed yield and oil quality of sunflowers [1] . Application of fertilizers substantially increases sunflower growth and yields, however, additions of nitrogen (N), phosphorus (P) and potassium (K) need to be optimized. In sunflowers, nutrient deficiency can result in up to 60% reduction in productivity [2] . The nutritional needs of sunflower are greater than many other crops like wheat, sorghum and corn, requiring higher amounts of N and other macronutrients [3] .

The number of achenes per head is a reflection of action of N in critical early stages of flowering in sunflower development. The potential number of flowers is determined very early, and subsequently affects the number of achenes and head diameter [4] . Head diameter is one of the morphological characteristics most affected by addition of N, showing increases even with small N doses (25 kg∙N∙ha⁻¹). However, this increase in head diameter does not continue with further increases of N. Sachs et al. [5] observed an increase of achenes productivity with N doses up to 55 kg∙N∙ha⁻¹, 41 kg ha⁻¹of K₂O and 46 kg∙ha⁻¹ of P₂O₅. The achenes’ oil content increased with the application of K₂O and P₂O₅ and the protein content decreased with increasing K₂O application.

Shortage of water, which is the most important component of life, limits plant growth and crop productivity, particularly in arid regions. Sunflower is commonly regarded as a plant that is tolerant to drought and it uses water efficiently. Nevertheless, the crop consumes a large amount of total water due to the fact that it produces high yields and a large vegetative bulk. It also has a long growing period coinciding with the warm months of spring and summer. Water stress on sunflower reduces plant height, root length, stomata number and causes early flowering, early maturity and seed yield reduction. Drought adversely influenced leaf area leaf area, days to maturity, leaf diameter, 1000-achene weight and achene yield per plant [6] .

The present research aimed to investigate the yield of sunflower affected by the interaction between NPK fertilization treatments and irrigation regimes under semiarid Brazilian conditions.

The experiment was carried out from March to July 2011 under greenhouse conditions at the Agricultural Engineering Department of the Federal University of Campina Grande, Paraiba State, Brazil.

The sunflower cultivar used was the Embrapa 122-V2000. A total of 44 treatments on a completely randomized design generated by the Baconian Matrix (Table 1) with four doses of N (0, 60, 80 and 100 kg∙ha⁻¹), four of P₂O₅ (0, 80, 100 and 120 kg∙ha⁻¹), four of K₂O (0, 80, 100 and 120 kg∙ha⁻¹), four available soil water (ASW) levels (55%, 70%, 85% and 100%) and three replicates resulting in 132 experimental units. Urea was used as a source of N; triple super phosphate as P and potassium chloride as K. In the soil in all the experimental units was applied, as pure solution, 2 kg∙B∙ha⁻¹ as boric acid.

Each experimental unit consisted of a plastic pot filled with 32 kg of an Alfisol with the following attributes using the procedures recommended by [7] : sand = 553.40 g∙kg⁻¹; silt = 117.30 g∙kg⁻¹; clay = 329.30 g∙kg⁻¹; pH (H₂O) = 6.6; Ca²⁺ = 1.45 cmol_c∙kg⁻¹; Mg²⁺ = 1.65 cmol_c∙kg⁻¹; Na = 0.17 cmol_c∙kg⁻¹; K⁺ = 0.21 cmol_c∙kg⁻¹; H⁺ + Al³⁺ = 0.79 cmol_c∙kg⁻¹; organic matter = 22.2 g∙kg⁻¹; Available P (Mehlich) = 8.1 mg∙kg⁻¹.

Soil water content was monitored daily at three depth intervals: 0 - 10, 10 - 20 and 20 - 30 cm, using a Frequency Domain Reflectometry (FDR) segmented probe, inserted into the soil through an access tube installed in the pots. The volume of water required to maintain ASW for each treatment was calculated based the difference between field capacity and the permanent wilting point, and the FDR measurements. Irrigation was performed daily.

Five sunflower seeds were sown directly in the pots at a 2 cm depth. Twenty days after sowing, seedlings were thinned to one plant per pot.

When the experiment was finalized, plants were harvested and measured for dry matter (grams) of the head (DMH), total number of achenes (TNA), total weight (grams) achenes (TWA) and 1000 achenes weight (W1000A).

The results were analyzed statistically through the analyses of variance (ANOVA) described by [8] , using SAEG software [9] .

^*Reference level used by the sunflower growers of the region.

The ANOVA results are presented in Table 2. The results indicate that N, P and ASW had a significant effect on dry matter of the head (DMH) at 1% level of probability (Table 2) corroborating [10] .

For plants treated with N the DMH increased with increasing N rate whose data were fitted to a quadratic regression model (Figure 1(a)). It is observed that the highest value of this variable (28 g) was registered for the highest dose of N (100 kg∙ha⁻¹) 25% higher than the DMH found for the control treatment. The DMH was also influenced significantly by increasing P doses, and these results were fitted to a quadratic regression model (Figure 1(b)). The highest value of this variable (31 g) was obtained with the highest dose of P (120 kg∙ha⁻¹) showing a superiority of 556% when compared with the reference level. The results corroborate [11] and [12] who found an increase of the DMH of sunflower cv. Embrapa 122/V-2000 with increasing N doses and ASW.

Potassium rates did not affect the DMH values (Table 2). These results disagree with those results found by [6] and [12] who found a positive significant effect of K on DMH.

The DMH increased linearly with increasing ASW from 18 to 24 g, for the lowest (55%) to the greatest ASW doses (100%), respectively (Figure 2). Guedes Filho et al. [11] observed an increase of 43% in DMH when

ASW was increased from 55% to 100%. Similarly [13] observed that the level of ASW influenced the DMH linearly at 1% probability. The higher dry matter accumulation may be a reflection of a greater ion absorption in soil, since the increase in soil moisture on the development of sunflower crop can be significant in nutrient uptake by plants.

The total number of achenes (TAN) was significantly affected, at 1% level of probability, by the increase of N, P and K doses and available soil water. The interaction among N and ASW at 1% level of probability was also significant (Table 2) corroborating [12] .

The increase of the number of achenes with the N doses was fitted to a linear model as observed in Figure 3(a). The greatest TAN was obtained, thus, with the highest N doses.

The effect of P on the TAN was fitted to a quadratic model (Figure 3(b)) verifying the greatest achene number (416) at 80 kg∙P∙ha⁻¹. This kind of adjustment was also reported by [14] and [15] . The response of achene number to the K was also fitted to a quadratic model (Figure 4(a)) verifying the greatest achene number (506) for the 120 kg∙K∙ha⁻¹.

The increase in available soil water content increased, in a quadratic manner, the achene number from 301 to 448 when water increased from 55% to 100% ASW, an increase of 52.30% (Table 2). Otherwise, [11] found a quadratic effect producing above 600 achenes with 100% level available soil water.

The effect of ASW on the sunflower is similar to other studies [12] [16] - [18] . Figure 4(b) shows an increase of the TAN with the ASW in a quadratic model, obtaining the highest number of achenes for the highest level of water.

The interaction among N and the available soil water content on TAN was significant to the 1% level of probability (Table 2). Figure 5 presents the cited interaction. It is observed that in general, the TAN increased

with the N application and with the ASW. Thus the highest total number of achenes (844) was obtained with the highest N application (100 kg∙ha⁻¹) and the highest ASW (100%).

The total achenes’ weight was significantly influenced by increasing N, P, K rates and the ASW (Table 2). Total achenes’ weight increased with N and P treatments, the data was adjusted to a linear and quadratic regression model, respectively (Figure 6(a) and Figure 6(b), respectively). Analyzing Figure 6(a), it is observed that the highest total achene’ weight (16 g) was obtained with the highest N dose (100 kg∙ha⁻¹). Figure 6(b) shows that the highest total achene’ weight (16 g) was obtained with the highest P doses (96 kg∙ha⁻¹). These highest weights were 136% and 728% superior to the control treatment, respectively.

The TWA increased linearly with increasing K (Figure 7(a)) verifying the greatest achene weight (15.67 g) for the 120 kg∙P∙ha⁻¹. The ASW treatments increased the total achenes weight linearly obtaining 8.94 and 15.22 g for the lowest and highest treatment, respectively. There was an increase of 70.25% between the lowest and highest water treatments.

The 1000 achenes’ weight of sunflower was only affected by the P application, at the 1% level of probability. The regression was adjusted to a quadratic model (Figure 8), verifying an increase of the 1000 achenes’ weight with P application and therefore, the greatest value for the treatment of 120 kg∙P∙ha⁻¹ (60 g).

The dry matter of the head, the total achenes’ number and the total achenes’ weight increased significantly with the nitrogen doses applied.

All the sunflower variables studied increased significantly with the phosphorus doses applied.

The total achenes’ number and the total achenes’ weight increased significantly with the potassium doses applied.

With the exception of the 1000 achenes’ weight, whose effect was not significant, all the production components of the sunflower increased with the available water in the soil.

The interaction between N and the available soil water content was significant only for the TAN. It increased with the N application and with ASW.

The highest rates of N, P, and K (100, 120 and 120 kg∙ha⁻¹, respectively) and the 100% available soil water produced the highest production.

Thanks to the Coordination of Improvement of Higher Education (CAPES) for the scholarship award to the second author at the Graduate School.