Rice is the staple food of the Indonesian people, and it accounts for more than half of their caloric intake. Rice production is a source of livelihood for more than 200 million people in the country, especially in the lowlands. The total land area cultivated to rice is close to 12 million hectares, and in 1987 irrigated rice covered 58% of the total cultivated rice area. Java is the main producer of rice. With increasing population and decreasing area for rice cultivation in the fertile lowlands due to urbanization and industrialization, it is important that the long-term sustainability of the irrigated rice production system in Indonesia be assessed. The Long-Term Fertility Experiment (LTFE) at the Indonesian Center for Rice Research (ICRR), Sukamandi, West Java was initiated in 1995 to assess the long-term changes in soil nutrient supply, nutrient balance, nutrient use efficiency, yields, and overall sustainability of a double rice cropping system.

Results for 21 cropping seasons (or 10.5 years of intensive cropping) indicate that with balanced fertilization of N, P, and K grain yield averaged 5.5 t ha^-1 in the dry season and 6.5 t ha^-1 in the wet season. The accumulated loss in grain yield without application of fertilizer N was 40 t ha^-1 across the 21 cropping seasons (Fig. 1). This corresponded to an average grain yield loss of 2 t ha^-1 in each season if fertilizer N was not used. Thus, the use of fertilizer N with appropriate amounts of fertilizer P and K ensured an average additional grain yield of 2 t ha^-1 in each season.

Based on the site-specific nutrient management (SSNM) approach for fertilization of rice (IRRI 2006), a rice crop requires about 50 kg fertilizer N per hectare for each ton in additional grain yield. The optimal amount of fertilizer N required to attain the yield targets of 5.5 t ha^-1 in the dry season and 6.5 t ha^-1 in the wet season at this site was consequently near 100 kg N ha^-1. Thi s amount of fertilizer N can be split into three doses with an early N application of about 20 to 30 % of the total requirement. The remaining 70 to 80% is split into two doses with the timing of N application based on the need of the rice crop, as determined from leaf color using the leaf color chart.

Farmers in West Java often believe the supply of K in the soil is sufficient for high rice yields and hence there is no need to apply fertilizer K to their rice fields. However, results of the LTFE show that the total loss in yield without application of fertilizer K was 10 t ha^-1 across the 21 seasons (Fig. 1). For the first eleven seasons, there was a slow decline in yield (0.3 t ha^-1 per crop) without K fertilization. But then from the 12th season onwards the yield loss increased to 0.6 t ha^-1 per crop. The results indicate that with continuous cropping, K becomes depleted in the soil, and thus it should be replenished for the attainment of high yields. With fertilizer K application, there was an average increase in grain yield of 0.5 t ha^-1 per season.

Based on the SSNM approach, the rice crop requires an estimated 40 kg K₂O ha^-1 to achieve a yield target of 6.5 t ha^-1 in the wet season and 25 kg K₂O ha^-1 to attain the yield targets of 5.5 t ha^-1 in the dry season (Fairhurst and Witt, 2002). This is based on an average yield response of 0.5 t ha^-1 in each wet- and dry season.

The application of sufficient fertilizer K to overcome deficiency of K can increase the efficiency of fertilizer N use. Fig. 2 illustrates a situation where the rate of fertilizer N application (120 kg N ha^-1 was sufficient with adequate application of fertilizer P and K to achieve a rice yield of 5.7 t ha^-1 (SSNM treatment). With insufficient fertilizer K the yield was 5.2- 5.4 t ha^-1 (FFP before and during SSNM). The application of additional fertilizer K, through an increase in yield with no additional use of fertilizer N, increased the recovery efficiency of fertilizer N by the rice crop (REN) to 37% of the applied fertilizer N (Dobermann, 1999; Fig. 2).

The actual amount of fertilizer K required in a specific field depends upon the quantity of crop residues returned to the soil, the quantity of K in irrigation water, and the soil properties. A simple tool to assess whether a rice field is deficient in K and to determine the quantity of required fertilizer K is the 'K addition plot' (as all the field is a large K omission plot). With this technique, a small plot (5m x 5m) is placed in a farmer's field and fertilizer K is applied to the rice plants in the plot. The increase in grain yield in the K addition plot compared to the rest of the field indicates the benefit and need for fertilizer K application. Addition plots are a useful tool to decide whether the current rate of fertilizer K used by farmers should be increased. It should be noted that the greatest benefit from K application is often only achieved with optimal management of other nutrients and best management practice.

Acknowledgment: The long-term fertility experiment was conducted by Dr. Sarlan Abdulrachman and other scientists at the Indonesian Center for Rice Research.

References
Dobermann, A. (1999): Reversing diminishing growth of rice yields in Asia. 6th IFA Annual Conference, 17-20 May 1999, Manila, The Philippines.

Fairhurst, T.H. and C. Witt (eds.). 2002. Rice: A practical guide to nutrient management. Singapore and Makati City: Potash & Phosphate Institute/Potash & Phosphate Institute of Canada (PPI/PPIC) and International Rice Research Institute (IRRI). p 1-89.

International Rice Research Institute (IRRI). 2006. Site-specific nutrient management. http://www.irri.org/irrc/ssnm/. Accessed 23 Oct 2006.