In rainfed lowland rice, nutrient status of soils is often poor, and respon
se to applied nutrients is often unreliable. This paper seeks an improved u
nderstanding of these patterns of nutrient response, then considers likely
approaches for increasing and stabilising yields of rainfed lowland rice in
south and southeast Asia. At each of 78 locations across India, Bangladesh
, Thailand, Indonesia and the Philippines during 1995-1997, three replicate
s of each of six treatments were established: (no fertiliser control (NIL),
farmyard manure (FYM), phosphorus and potassium only (PK), application of
nitrogen as well as P and K (NPK), a controlled-release formulation of N to
gether with P and K (CR-NPK), and an all-nutrient dressing (ALL)). Combined
analysis of variance and cluster analysis were used to examine the pattern
s of nutrient response. Yields obtained without applied fertiliser were not
closely related to soil test values. The greatest nutrient response was to
nitrogen, with NPK increasing yields from 2.25 to 4.00 t/ha on average. Th
e effect of adding micronutrients was small, and PK was of little benefit u
nless N was added. But the magnitude of the N response varied substantially
with water regime. We conclude that substantial yield gains are possible i
n rainfed systems with application of appropriate nutrients, especially if
used in conjunction with cultivars suitably adapted to the target environme
nts. (C) 1999 Elsevier Science B.V. All rights reserved.