Introduction
The fruit tree industry is one of the most important agricultural sectors in Tunisia, with more than 2 million hectares planted mainly with olive (1.5 million ha), almond (257,000 ha), pistachio (44,000 ha) and palm date (26,000 ha). Citrus and stone fruits are also economically important crops (Table 1). Owing to water scarcity, the sector is characterized by significant disparities between regions and growers. For example, olive tree orchards have been found to vary from very low density (17 trees/ha in the center and south of the country) to very high density with more than 1,250 T/ha, the difference depending mainly on water availability for irrigation. Water scarcity is the main limiting factor for Tunisian agriculture.

Mineral nutrition is one of the major tools to optimize fruit yield and quality (Tagliavini and Marangoni, 2002). Potassium is known not only to play an important role on olive yield and quality but also on water-use efficiency (Arquero et al., 2006), which indicates its importance for the Tunisian fruit sector where trees have been traditionally grown on calcareous soils under rainfed conditions.

Nutrient uptake depends on nutrient supply to the root system i.e. nutrient availability and the nutrient requirement level and the uptake period (Chapin, 1991). Fine textured soils are characteristically K⁺ fixing soils, where soil-surface application is almost without effect (Mengel, 2002). Fertigation relative to broadcast application has been shown to increase K mobility in the soil (Nielsen et al. 1999; Uriu et al. 1980; Southwick et al. 1996). Weinbaum et al. (1994) suggested that this effect of K fertigation in significantly improving K uptake resulted from increasing rhizosphere soil solution K concentration. However, limitation in tree capacity to take up K during the high demand period can present a problem (Weinbaum et al. 2002) especially in K⁺ fixing soils.

Foliar application of nutrients is in general helpful to satisfy plant requirement and has a high efficiency (Inglese et al. 2002). Potassium is particularly well adapted to this form of fertilization because soon after foliar spraying takes place it is rapidly translocated from the leaves (Mengel, 2002). Foliar application is thus an attractive means of K application to trees especially in arid zones where a lack of water under low rainfall conditions in summer drastically depresses absorption of soil nutrients.

In 2003, IPI coordinating research in the WANA region working jointly with the Fruit Tree laboratory of the Institut National Agronomique de Tunisie (INAT) began a project to evaluate the effect of potassium foliar spraying on different fruit crops (olive, citrus, pistachio, peach and plums) under different growing conditions. The purpose of this paper is to present results obtained from the olive experiment under rainfed conditions. Results from fertigated orchards of peach and plum will be reported in another publication.

Materials and methods
This experiment was carried out over a period of five years beginning 2003 on a fine sandy soil at the Olive Institute experimental station which is located 26 km north of Sfax in the center of Tunisia. The zone is characterized by a semi-arid climate with an annual precipitation of 200 mm. A commercial olive orchard growing the cultivar Chemlali (the most important olive oil cultivar in Tunisia) was used for this experiment. The trees 24 x 24 m apart were grown using standard cultural practices of the Sfax region, i.e. under rainfall condition and without any fertilization.

At the beginning of the season, an estimate of potassium requirement was made based on the removal of nutrients contained in the yield (200 kg/tree) and the pruned wood. The total tree removal was estimated as 26.71 kg K₂O/ha (1.57 kg K₂O/tree). Potassium was applied by two different methods and on two quantities as indicated in Table 2. The same procedure was carried out annually from 2003-2007.

Potassium was applied as potassium sulfate (K₂SO₄).

The soil spreading treatments were carried out by one application made during the period of flower bud swelling.

The foliar fertilization treatments were applied using a 100 liter/tree spray, with K concentration of a maximum of three per cent as follows:

• 30% of K total tree removal during the flower bud swelling
• 40% of K total tree removal during the second fruit development stage
• 30% of K total tree removal just at the beginning of the fruit color change

Four single tree replications were used for each treatment. The vegetative growth of the shoots was measured once a month. Every 15 days after fruit set, fruit weight was measured on a sample of 200 fruits (50 per tree). At harvest yield per tree was determined and the oil extracted using an oil mill on four samples of 2.5 kg olives per treatment.

Leaf samples were collected in July each year from all treatments. About 40 fully expanded mature leaves per each tree were collected. In all cases, leaves were taken from the middle portion of the current season's terminal shoot growth. Leaf samples of the same treatment from each block were combined as samples of 100-120 leaves for determining leaf K concentration.

Results and discussion

Vegetative growth

No significant effect of the different potassium fertilization treatment was observed on vegetative growth (Fig. 1).

Fruit maturity and quality

The pattern of fruit growth was also not influenced by fertilization treatment (Fig. 2). Although fruit growth was higher during stage three for the 100 per cent foliar treatment, this increase was not statistically significant. Similar results were observed by Inglese et al. (2002).

Leaf Analysis

For each of the four years of the experiment the leaf concentration of K was higher for foliar treatments at F100 than the control (Fig. 3). However, the K concentration varied between "on" and "off" years for all the treatments: the values were higher during the off year and were above the sufficiency threshold of 0.8 per cent for olive (Freeman et al., 1994) and vice versa. Fernandez-Escobar et al. (1999) observed the same results. These workers concluded that the high leaf K accumulation following the "off" year and the rapid decline after March of the "on" year suggest a large K demand by the reproductive structures of olive.

Yield and quality

After five years of experiments, potassium applied either as foliar or in the soil increased yields of olives. A significant difference in the cumulated yield was observed between foliar spray at 100 per cent nutrient replacement for K removal (F100) and the control (Table 3). The increase in yield is higher than 200 kg/tree, representing an almost additional yearly harvest yield for this very alternating cultivar. In other work, Restrepo-Diaz et al. (2008) did not observe any effect of potassium foliar fertilization on olive under rainfed conditions. However they stopped their experiments after only three years.

We assume that since F100 was applied in a split application (three times as compared to only one application to the soil) and the fact that the foliar applied K was more available to the tree, that these advantages contributed to the significant yield increase resulting from this form of application.

Potassium application to the soil and in foliar form significantly increased fruit weight and Flesh to Pit Ratio (Table 4). Foliar application of K (F100) proved to be best treatment with an increase of approx. 33 per cent in fruit weight and 24 per cent in Flesh to Pit Ratio, over the control. This beneficial effect may possibly have arisen as a consequence of a higher availability of assimilates.

At the same time, potassium application significantly decreased oil content in the fruit. As the oil content is negatively correlated with fruit weight and yield, we assume that the decrease in the oil content is due to dilution caused by the higher yields.

Fig. 4 shows the oil yield achieved in the different treatments (i.e. the product of yield and oil content). Clearly, the treatment of F100 (foliar application of potash at a rate to fully compensate for the removal of K in crop and buds) yielded approx. 50 per cent more oil than the control, thus offering the farmers a significant additional income.

Conclusions
This study shows the importance of potassium fertilization in increasing the yield and oil yield of rainfed olive. The effect of split foliar application of potassium was superior to soil application which implies that this technology is preferable. Restrepo-Diaz et al. (2008) recommended foliar application in orchards growing under rainfed conditions because the lack of moisture in the soil during the growing period that could limit K uptake by the roots, which is the case for the majority of Tunisian olive orchards. Our calculations for potassium application rates based on removal of K in fruit and tree buds seem to be in good agreement. More work is still required in order to assess whether this balance is sustainable and profitable over longer periods. The results of this experiment demonstrate the need for a long term approach in such conditions, as both the fruit type (olive) and the climatic conditions pose a challenge in the analysis of data from field experiments.

Acknowledgements
Special thanks go to the International Potash Institute (IPI) for its support to this research.

Literature cited

• Arquero, O., Barranco, D., Benlloch, M. 2006. Potassium starvation increases stomatal conductance in olive trees. HortScience 41, 433- 436.
• Chapin F.S. 1991. Effects of multiple environmental stresses on nutrient availability and use. In H.A. McConey, W.E. Winier and E.J. Pell (Eds). Response of plants to multiple stresses. Acad. Press, Inc.
• Fernandez-Escobar R., Moreno R. and Garcia-Creus M. 1999. Seasonal changes of mineral nutrients in olive leaves during the alternatebearing cycle. Sci. Hort. 82:25-45.
• Freeman, M., K. Uriu, and H.T. Hartmann. 1994. Diagnosing and Correcting Nutrient Problems. In Olive production manual. University of California. Division of Agriculture and Natural Resources. Pub 3353:77-86.
• Inglese, P., Gullo G., and Pace L.S. 2002. Fruit growth and olive quality in relation to foliar nutrition and time of application. Acta Hort. 586:507-509.
• Mengel K. 2002 Alternative or complementary role of foliar supply in mineral nutrition. Acta Hort. 594:33-47.
• Nielsen G.H., Nielsen D. and Peryea F. 1999. Response of soil and irrigated fruit trees to fertilization or broadcast application of nitrogen, phosphorus, and potassium. Hort Technology 9:393-401.
• Restrepo-Diaz H., Benlloch, M. Navarro C. and Fernandez-Escobar R. 2008. Potassium fertilization of rainfed olive orchards. Sci. Hort. 116:399-403.
• Tagliavini, M. and Marangoni, B. 2002. Major nutritional issues in deciduous fruit orchards of north Italy, Hort Technology.12:26-31
• Weinbaum S.A., Brown P.H. and Johnson R.S. 2002. Application of selected macronutrients (N, K) in deciduous orchards: physiological and agrotechnical perspectives. Acta Hort. 594:59-64.
• Weinbaum S.A., Niederholzer F.J.A., Pochner S., Rosecrance R.C., Carlson R.M., Whittlesey A.C. and Muraoka T.T. 1994. Nutrient uptake by cropping and defruited field grown "French" prune trees. J. Amer. Soc. Hort. Sci. 119:925-938.