Previous PageTable Of ContentsNext Page

Australia’s declining crop yield trends I: Donald revisited.

Andrew Lake

Pristine Forage Technologies, Email


CM Donald’s seminal paper on Australian wheat yield trends is now nearly half a century old. An updated analysis, based on data of average wheat yields per hectare over the period 1970 to 2010, obtained from ABARES is presented. Yields averaged as 5 year moving means gave a reasonably smooth trend curve, and these show a consistent, strong upward trend from the late 1970s through to the early-mid 1990s. After this, and in parallel with rotation changes such as reduced use of legume leys, there is a clear plateau and decline, which is still evident after the confounding effect of drought is removed. This deterioration is corroborated by comparing Australian and global wheat yields per hectare; while our average reached a very creditable 75% of global average in the early 1990s, it has since declined to less than 60%. Given the amount of investment both on and off farm in R&D, higher yielding varieties, production technologies and other cutting edge inputs, if this trend is real as it appears, then it is a cause for considerable concern.

Key Words

Australian wheat yield trends, cropping, ley farming, rotations, productivity, sustainability.


Statistics of Australian average wheat yields per hectare have been calculated and collected for more than a century. Trends in these average yields over time have been used to demonstrate the state and sustainability of Australian agricultural practices and techniques through various phases and periods. Donald’s seminal analysis of these trends up to the 1960s (Donald 1967) showed three distinct phases of Australian agriculture. These phases can be broadly summarized thus;

  • an initial phase (up to 1900) during which exploitative crop production led to gradually declining productivity (from about 0.85 t/ha to a low of about 0.5 t/ha) caused by nutrient mining and depletion;
  • a long fallow-superphosphate phase, (1900 to 1950) which while replacing phosphate still relied on fallows to further mine other soil nutrients, particularly N, raising crop yields back to about 0.85 t/ha; and,
  • a ley farming phase, (1950 to 1960) where fallow years were replaced by grazed legume pastures to fix N and replenish soil N banks, raising average yields to about 1.15t/ha by 1960.

Donald forecast a continuation of the trend productivity growth for at least a further two decades through ley farming, and through use of new (dwarf) wheat varieties and nitrogenous fertilizers.

Given that these predictions are now almost half a century old, and that there have been very significant changes to rotations, with a greater focus on cropping and a steep decline in use of legume leys since the late 1980s, this paper examines more recent trends in wheat yields.


Data sources and first level (decadal average) analysis

Data on average Australian wheat yields per hectare since the late 1960s were obtained from the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES). These data were initially analyzed via straight decadal averaging. To create a baseline for this analysis, Donald’s data (Donald 1967) were used for the 1950-1960 decadal average, while the 1960s decadal average was estimated from combining Donald and ABARES data. These decadal averages were then graphed accordingly.

Second level data analysis; using 5 year moving means.

Five year moving means were calculated for individual years since 1980-81 by averaging annual yields from the year, the two previous years and the two following years. These were then graphed against years.

Third level data analysis; accounting for recent droughts.

While there was one serious drought in the 1990s (1993-94), there were three in the decade since 2000 (2002-03, 2006-07 and 2007-08). In order to remove these drought effects, data for those respective years were replaced with an average of the yields for the closest four (non-drought) years. Thus for example, data for both 2006-07 and 2007-08 were replaced with an average yield for the years 2004-05, 2005-06, 2008-09 and 2009-10. The resultant data were then also graphed and analysed as a five year moving mean.

Fourth level analysis; comparing Australian and global yield trends.

While the previous analyses revealed significant trends in Australian wheat yields per hectare, it was not clear if these trends were similar to trends elsewhere in the world or if they were unique to Australia. Hence in this fourth level analysis, data for global average annual wheat yields per hectare obtained from ABARES for the period since 1980-81 were analysed to identify the underlying trend.


First level (decadal average) analysis

From the graph (Figure 1), yield growth was particularly pronounced in the decades of the 1980s and 1990s, but was followed by a dramatic reversal in the average yield for the most recent decade.

Figure 1. Australian wheat yield per hectare; decadal averages from 1951-2011.

Second level data analysis; using 5 year moving means.

The graph of the five year moving averages (Figure 2, green line) does indicate the existence of a fairly distinct peak of wheat yields. However, there is some evidence of a plateau rather than a sudden effect, with the moving mean for the peak three years being clear of both the 1994 and 2002 droughts. The droughts of the last decade are also likely contributors to at least part of the subsequent decline in this trend line.

Third level data analysis; accounting for recent droughts.

The five year moving mean trend line for data where drought years are replaced with averaged data for the next nearest non drought years is shown in Figure 2 in blue. This non drought trend line analysis shows a steady yield increase in the 1980s and early 1990s followed by a yield plateau of about 1.95t/ha that lasted for almost a decade. That is then followed by a yield decline which, while significantly less dramatic than indicated by the raw data trend line, is still clearly in evidence, and shows little sign of abating. That trend line suggests that current (non drought) Australian average wheat yields are about 1.7t/ha; well below the 1.95t/ha 1990 peak. The current rate of yield decline indicated by this trend line is around 15% per decade.

Fourth level analysis; comparing Australian and global yield trends.

The graph of data and trend lines for this analysis comparing Australian and global wheat yield trends since 1980, is shown in Figure 2. Over the entire 30 year period, global average wheat yields per hectare (light blue) have increased at a remarkably steady rate. In consequence, the straight trend line of these yields (magenta) provides a very good indicator of global average yields. From these trend lines, there is no sign of any plateau in global wheat yields, far less any decline as is seen in the Australian trend lines.

Figure 2. Trends in Australian and global wheat yields per hectare since 1980. The Australian trend lines are from 5 year moving means with and without drought. The light blue line shows global annual average yields per hectare, while the magenta line shows the calculated (straight) line trend for those yields.

General discussion; the implications of the observed trends.

By removing the perturbations of the drought effects, it is evident that while a considerable part of the apparent decline in Australian wheat yields per hectare over the last decade is due to those droughts, there is nevertheless still a significant underlying decline in average yields. This conclusion is further confirmed by the comparisons of Australian and global average wheat yields.

The shape of the non drought trend line in particular shows a gradual and progressive change in underlying yields per hectare over time. This indicates that the cause(s) are unlikely to be due to random events that will self-correct over time or to a sudden and dramatic change in the production environment, but are themselves both gradual in onset and widespread across the wheat belt.

Nevertheless, in terms of historical productivity change, this decline is not gradual. As Donald’s graph demonstrates (Donald 1967) even during the first stage of exploitative cropping, yield decline did not reach the current rate estimated from the above trend data of 15% per decade. Further, advances in production technologies, varieties and farmer inputs, such as are resulting in steadily growing global wheat yields per hectare, show that even the yield plateau evident in the 1990s is an unexpected outcome, and the subsequent decline disturbing. In effect, despite the additional effort, expense and intellect invested in crop production improvement since the early-mid 1990s, actual on-farm yields are now less than they were then. From a farmer perspective, prima facie it appears that effective return on his additional investment is negative.

However, as extensive trials of the vast majority of these innovations demonstrate and verify, positive yield responses (for example in industry sponsored National Variety Trials of crop varieties) this is very unlikely to be the case. This suggests the presence of other negative trends that are probably due to a very significant deterioration in the productive base or capacity of our major cropping regions. This may be attributable to a number of types of factors, such as to a deterioration in climate (eg increased drought) or to biotic factors (eg an increase in diseases) or to abiotic factors (eg declining nutrient availability) or even to their interaction (eg nutrient deficiency leading to poor crop nutrition, declining “good” soil biota and increased disease).

Farmers and industry tend to blame the first of these factors (climate and weather) but this is the least likely cause; after drought effects are removed, there is still an underlying productivity plateau and decline. Further, climate change did not start in 1995, should impact on global yields as well, and is not generally steady. In fact, as shown by Donald, progressive changes of this nature indicate changes induced by the production system itself; it is caused by the inherent nature of that production system. This also fits with the observation that production systems and rotations have changed significantly since 1990. Increased crop pressure and frequency and reduced use of good legume leys to spell crop land are central features of those changes. Accordingly, it would seem to be sensible to look at the effect of these factors on crop productivity as a starting point to determining the primary cause or causes of this productivity decline.

In any case, this decline raises immediate and serious questions for future Australian wheat production in particular, and crop production in general. Barring major real increases in returns for that crop product, farmers will be increasingly faced with the prospect that their returns are insufficient to justify their cost of inputs. That can be a dangerous situation for the industry, as the first tendency is to cut inputs such as fertilizer, and thereby potentially accelerate productivity decline.

However, if the reasons for this productivity decline can be elucidated and (economically) reversed, then the pent up increases in yield that are inherent in the innovations adopted over the last two decades will be released. Under those circumstances, current yields should be on a par with where they were relative to global average yield in the early 1990s. That being the case, current average Australian wheat yield should be about 2.3 t/ha; or almost 40% above where the trend lines indicate they currently are, adding 10 million tonnes to production and increasing farm gate returns by more than 2 billion dollars p.a.

That is sufficient to justify a considerable R&D effort to determine the reasons and causes of this decline and the means for its reversal. At least for the interim, it would also seem sensible for industry to encourage adoption of more conservative practices such as increasing use of restorative legume leys in rotations.


In updating Donald’s review of Australian wheat yield trends and their usefulness in analysing the impacts of common farming systems and practices on productivity (Donald 1967) some mixed messages on our productivity performance have emerged. While it is clear that good gains in on-farm productivity were made beyond 1960 as predicted by Donald, and that these continued up until the 1990s, there is solid evidence that at the very least these gains have since stalled and have more recently gone into reverse.

Common belief has it that this deterioration in productivity is primarily due to the relatively higher frequency of drought in the last decade, but removal of those drought effects shows that this is not the case; there is still an underlying decline in productivity that began years earlier following a production plateau reached in the 1990s. This is a situation that is typical of the effect noted by Donald where the production system itself is responsible for progressive changes in productivity. As such, the production system changes since 1990 (increased cropping intensity and reduced use of legume leys) are prime suspects.

In historical terms, this productivity decline is more rapid than in any previous period in the history of Australian wheat growing. In these circumstances, an increasing number of farmers are caught between rising costs and falling returns, and will very likely make cuts to inputs that could well exacerbate the rate of decline and potentially render it economically irreversible. Thus pending further R&D, farmers should be strongly encouraged to adopt more conservative approaches to farming by reducing cropping intensity and using legume leys to spell crop land and rebuild soil organic nitrogen, carbon and biotic life.


Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES). The Agricultural Commodities Statistics Report, 2008 et seq. Data; Rural commodities – wheat.

Donald CM (1967). Innovation in agriculture. In Agriculture in the Australian Economy. Editor DB Williams. pp. 57-86, Sydney University Press, Sydney.

Previous PageTop Of PageNext Page