Yield and water-use efficiency of water- and nitrogen-stressed wheat crops increase with degree of co-limitation

Availability of water and nitrogen are key constraints to primary productivity in arid and semiarid ecosystems. Theoretically, plant growth is maximised when all resources are equally limiting. This paper tested the hypothesis that for a given amount of available water, the gap between actual and attainable yield of dryland crops in semiarid southern Australia is inversely proportional to the degree of nitrogen and water co-limitation.

Field and simulation experiments were combined in an analysis involving three steps. Step 1 assessed the capacity of a crop simulation model to estimate yield and its responses to water and nitrogen inputs in the semiarid Mallee region. Step 2 derived a boundary function relating grain yield and water availability using simulations with long-term weather records. Step 3 explored the link between degree of co-limitation and deviations between actual yield and the boundary function. Degree of co-limitation (C_WN) was calculated as a function of model-derived nitrogen (NSI) and water stress indices (WSI), i.e. C_WN = 1 ? |NSI ? WSI|. Stress indices range from zero (no stress) to 1 (maximum stress), and C_WN tends to 1 when both resources impose constraints of similar magnitude to crop growth.

The field experiment combining locations, seasons and management practices generated a range of grain yield from 0.6 to 3.8 t ha^?1. Water availability, i.e. seasonal rainfall plus change in soil water content from sowing to harvest, ranged from 127 to 370 mm. Nitrogen fertiliser varied from nil to 36 kg N ha^?1 and inorganic nitrogen in the soil profile at sowing ranged from 29 to 497 kg ha^?1. For these ranges of conditions, the relationship between simulated and measured yield was statistically undistinguishable from the y = x function.

A factorial modelling experiment combining sites, seasons, initial soil water content and dose of nitrogen fertiliser was used to derive a boundary function which provided an objective and independent upper limit for the field data. Actual yield was below the boundary function in most cases. The difference between actual and attainable yield was inversely proportional to C_WN. This study thus supported the hypothesis that yield and water-use efficiency of water- and nitrogen-stressed crops increase with increasing degree of co-limitation.