Rainfall, sowing time, soil type, and cultivar influence optimum plant population for wheat in Western Australia

Document Type


Publication Date


Journal Title

Australian Journal of Agricultural Research


Print: 1836-0947 Electronic: 1836-5795


Agricultural Science | Agriculture | Agronomy and Crop Sciences | Plant Breeding and Genetics


In this paper we analyse existing experimental data (grain yield and yield components) from seed rate experiments on wheat in Western Australia, with the aims of determining which factors most influence the optimum plant population, and advancing some practical guidelines for improving the choice of seed rate under rain-fed conditions. Experiments (32) were conducted in the rain-fed cropping zone of Western Australia between 1996 and 2001, using factorial combinations of wheat cultivars (3–25) and target plant populations (4 or 5). Some of them also contained treatments of nitrogen fertiliser (0 or 40 kg/ha of N) or sowing times (2). Each cultivar × plant population dataset (248) was considered to be a record for the sake of the subsequent analyses. Actual plant numbers were counted in each experiment and the optimum plant population was estimated when the slope of an inverse polynomial curve (choosing the most appropriate of the LDL and QDL models in GENSTAT) fitted to each record was 2.5 kg/ha of grain yield for each extra plant/m2. The optimum populations were initially grouped using a regression tree technique into groups with similar characteristics using pre-sowing rainfall, rainfall in the growing season, sowing date, and soil type. The variables cultivar and nitrogen fertiliser rate were later added to the regression tree analysis. Yield components available for most experiments were used as an aid to interpretation of the results. The optimum plant population varied from 35 to 175 plants/m2 and average grain yields varied from 0.42 to 3.91 t/ha. Rainfall in the growing season (sowing date to harvest date) provided the first split in the regression tree, but pre-sowing rainfall (January to sowing date), sowing date, and soil type further modified the optimum population. The addition of N fertiliser rate as a variable in the regression tree did not induce any different groupings of the optimum population sets, but cultivars were grouped into 4 response types according to pre- and post-sowing rainfall amounts. Where rainfall in the growing season was <205 mm, improved growing conditions due to more pre-sowing rainfall, earlier sowing, and more seasonal rainfall, were associated with higher optimum plant populations. Where rainfall in the growing season exceeded 205 mm, higher pre-sowing rainfall was associated with lower optimum populations. The optimum population was greater on sands than on clay loams. However, on sandy loam soils the optimum was less where rainfall in the growing season was <291 mm, or more for crops sown after 27 May at rainfall >291 mm. Increases in yield components in response to improved growing conditions above about 400 culms/m2, 300 ears/m2, 10 000 kernels/m2, and 600 g/m2 of dry matter at anthesis were not associated with higher optimum plant populations. In general, the optimum plant population increased at about 40 plants/m2 for each tonne of grain yield up to about 3.0 t/ha. The effect of cultivar on the optimum population appeared at yield levels above 2.5 t/ha, but was only detectable when the rainfall in the growing season exceeded 205 mm. Growing conditions and cultivars associated with lower weight per ear (due to fewer kernels and/or lower kernel weight) were associated with higher optimum plant population when the rainfall in the growing season exceeded 205 mm. It is suggested that farmers can make better estimates of the appropriate plant population (and hence can calculate seed rate) on the basis of pre-sowing rainfall (likely stored water), rainfall zone (probability of rainfall in the growing season), sowing date, soil type, and characteristics of individual cultivars where known.