Final Report of Theme 5.1 to the National Land & Water Resources Audit
4. Productivity changes
Productivity, measured as total factor productivity, yields and biological productivity generally appears to have increased steadily across the areas of intensive land use over the 1982-83 to 1996-97 timeframe. The rates of increase have varied between regions. Areas with consistently high productivity rates appear to be in the northern grain belt of Western Australia, the irrigated areas along the Murray and Murrumbidgee Rivers, the central grain belt of New South Wales, the coastal areas of Queensland where sugar areas have increased, and in the dairy areas of southern Victoria and north-western Tasmania. Areas of lower productivity growth include most of Queensland except the coastal areas, the high rainfall sheep and beef grazing areas of New South Wales, Victoria, and Tasmania, and the more arid grazing and cropping regions of South Australia.
Total productivity per hectare tends to be high along the southern and eastern coastal regions of Australia and in the inland areas to often directly relate to higher winter rainfall.
Over the last 20 years the land resources of Australia, with additional inputs, have continued to yield useful and desirable products of food and fibre. Although, such effects may well be masked, there is little indication from the evidence at the Statistical Local Area scale that land degradation is affecting productivity.
Productivity in this exercise is most pertinent in relation to land quality or condition, and is particularly relevant because the area or amount of land for agriculture has been essentially static for the last 20 years and will most likely remain so for the next 20 years. Land degradation, the adverse onsite and offsite effects of land uses on the services provided by land and water, is a core focus for the Audit. While soils are at the centre of land values, clearly there are other components such as native biota, water, scale of operation and even climate, that are included in the concept of ‘land’.
Land can be valued for economic uses (such as agriculture, forestry, dwelling, transport and industry) and more social uses (such as nature conservation, recreation and aesthetic), while recently the value as provider of services (such as clean water, air, and greenhouse sinks) has emerged. Consequently productivity can be approached from a number of viewpoints.
There are a range of different measures used to assess the productivity of agriculture through time. Two included here are:
partial productivity measures for a single input that provide insights into the intensification of individual components such as yield when expressed as the volume of production for a given (single) unit of input (eg. a hectare of land, a mm of effective rainfall, a megalitre of irrigation water applied) but may also take into account some climatic factors, such as the Stress Index (STIN) model; and
•total factor productivity (TFP) measures where a diverse range of inputs and/or outputs are involved. TFP measures the productivity of the whole enterprise.
Since the production of any commodity is the result of many inputs, the measurement of yields (which is usually expressed per unit of land) reflects the influence of those other inputs. Consequently, the influence of land condition on productivity may well be masked by changes in other inputs.
Pastures, together with the ‘residual’ area, occupy the largest area of any land use. However, measures of pasture productivity are not collected in ABS agricultural statistics, since they are usually consumed by grazing animals and are complicated by the large ‘residual’ component of agricultural lands for which little is known. Therefore an attempt was made to derive pasture productivity indirectly from ABS statistics. It was based on values from the National Pasture Survey of 1994, area of pasture (actually non-crop area) from ABS data and then calculations for productivity based upon responses of different pasture types to rainfall (see Methodology attachment) but not for temperature. The estimates are only indicative due to the uncertainties in the data used for the three components of the calculations.
As an example the results for two of the case study areas, Boyup Brook and Mingenew in Western Australia, are presented in Figure 4-1. There is considerable variation from year to year most probably due to rainfall patterns. However, the differences that appear between the two shires are less reliable because of the data uncertainties.
Figure 4-1 Estimated pasture productivity in Boyup Brook and Mingenew SLAs in Western Australia during 1982-83 to 1996-97. (source ABS and BoM)
Because of the questions about pasture data (relating pasture types to pasture area), only broad regional patterns should be accepted about productivity from pastures. Figure 4-2 shows calculated pasture productivity as kg DM/ha for the average of 1982-83 to 1996-97 for locations where a consistent data set was possible - because the boundary changes in Victoria and Tasmania were large some parts could not be calculated. Not surprisingly, it indicates that pasture productivity was closely related to mean rainfall distribution, being highest along the coastal regions and decreasing inland. By contrast in the Riverina there are several shires with high productivity due to supplementary irrigation in the MIA.
Further calculations based on animal numbers in each shire for each year allow an estimate of productivity in terms of dry sheep equivalents (DSE). However, there are conerns about the accuracy of this estimate due to assumptions about the area of pasture grazed, assumptions underlying conversions of animal numbers to DSE, and the location of animals in feedlots and agistment. The results shown in Figure 4-3 generally follows the pattern for pasture productivity, with the highest pasture DM/ha also having the highest DSE per hectare. However, it appears that animal production in the Central Wheat Belt of Western Australia, western Riverina in New South Wales, and the Burdekin and Burnett of Queensland were not as high as might be expected from the calculated pasture productivity. Also the North Coast and Northern Tablelands of New South Wales and the Darling Downs had higher animal productivity than might be expected from calculated pasture production. Some of this could well be a consequence of cattle feedlots in these locations.
Figure 4-2 Estimated mean dry matter production by pastures in the areas of intensive land use of Australia over 1982-83 to 1996-97. (sources ABS, BoM)
Figure 4-3 Mean productivity from pasture calculated as dry sheep equivalents during 1982-83 to 1996-97. (source ABS).
Trends in wheat yields over the 1982-83 to 1996-97 time frame were calculated using the Stress Index (STIN) model to remove the major effects of climate (details are given in the Appendix 2 - Methods). The amount and distribution of growing season rainfall has large effects on yield of wheat crops that can overshadow trends in productivity. However, models that predict grain yield can be used to allow for the effect of rainfall. Then by relating, using a multiple regression, the calculated wheat yield with the average wheat yield achieved in a Statistical Local Area an estimate of technological yield gain can be derived. The results in Figure 4-4 generally shows the highest wheat yield trends or more than 60 kg/ha/year occurred in southeastern and northeastern NSW, northwestern and southwestern WA, wetter districts in SA and the southeastern edge of the Darling Downs (Qld).
In the wheat belt of Western Australia the yield trend weakened towards the inland and east. The area in the northern part of the wheat belt around Geraldton showed large increases in yield and also showed high broadacre crop diversity (see Figure 3-27). It is likely that the extra nitrogen contributed by the lupin crop and better disease and weed control from the crop rotations are having a beneficial effect on wheat yields. The generally high wheat yield trends across Western Australia probably is a consequence of the widespread adoption of whole new agronomy packages during the timeframe. The packages included: a change from taller, short-season crops to higher yielding, long-season, semi-dwarf varieties (an increase from15% to 90% of sown wheat); a move to sowing 2-3 weeks earlier; a switch to minimum/zero tillage; an increase in crop diversity by using grain legumes (especially lupins) in rotation with wheat; and substantially higher application of nitrogen fertiliser. Low yield variability and generally reliable growing season rainfall (Figure 2-1) has encouraged a number of these changes and a move to a high input agricultural system in general.
In South Australia wheat yield trends are closely related to rainfall, with the wetter districts consistently having yield increases greater than 40 kg/ha/year and the drier districts consistently lower than 40 kg/ha/year. In the drier districts there has been slower adoption of break crops, greater root diseases, higher yield variability and therefore an increased risk for applying nitrogen fertiliser. Also, in the northern Eyre Peninsula highly calcareous soils with sodic subsoils have been unsuitable for lupins, and oilseed crops often given uneconomic yields.
In Victoria the yield increases were most marked towards the south. However, the Wimmera of western Victoria experienced negative/negligible wheat yield trends, although there was a large increase in crop diversity (see Figure 3-27). Here farmers have concentrated on high yielding cash crops of chickpeas, lentils and faba beans which bring higher returns than wheat. Wheat yields have suffered as long fallows (and accumulated soil moisture) have been reduced, better land has been planted to pulses and the pulses have contributed less nitrogen to the soil due to high pulse grain yield and low biomass incorporation.
New South Wales showed strongest gains in wheat yields in pockets along the eastern border and in irrigation areas of the MIA. The very low yield trends found by Hamblin and Kyneur (1993) in northern NSW have been reversed as farmers have: 1) substantially increased the application of nitrogen fertilisers, 2) improved fallowing and weed control management and 3) increased summer cropping (sorghum) which forms a break crop reducing root disease. Successful summer crops (sorghum/cotton) have allowed farmers to invest more on inputs for winter crops and increased skills in crop management. Also in this region, nitrogen applied in one season can carryover from a low yielding crop to the next season, and reduces the risks of high input costs and losses in bad seasons.
There were only a few, small areas showing strong gains in Queensland, mostly around the inner Darling Downs. One analysis suggested that the low rates of increase in wheat yields in this northern part of the grain cropping belt was due to low use of nitrogen fertilisers, limited phosphorus and the adverse effects of pathogenic fungi (Cornish et al 1998).
Yield variability is a major constraint on farmers ability to increase crop productivity as increased variability is particularly unfavourable for grain legumes and poses a risk to getting returns on investment in nitrogen fertiliser. Over the timeframe, climate variability (see Figure 2-1 and Appendix 1, Figure 1) was also very high in regions of low trends in Queensland, northern Eyre Peninsula (SA) and in central NSW where application rates of N were generally very low.
The map of variability in wheat yields (Figure 4-5) over the time frame is interesting in comparison with that of rainfall reliability (Figure 2-1). It is apparent that those areas with a rainfall reliability of 70 per cent or better have low (less than 0.25 coefficient of variability) variability in wheat yields. In some cases variability may be slightly higher, likely because of disease or water logging constraints. Areas with high yield variability (cv greater than 0.38) all appear to occur in areas where rainfall reliability is lower than 30 percent during the growing season for wheat. It again emphasises the dominant role of rainfall in defining Australian broadacre agriculture.
Figure 4-4 Trends in wheat yields (kg/ha/year) for different Statistical Local Areas (SLAs) of Australia. (source ABS and AgWA)
Figure 4-5 Variability in wheat yields (expressed as coefficient of variation) in the Intensive Landuse Zone during 1982 to 1997.
Barley yield trends were calculated in a similar manner to wheat. Barley yield trends show the same regional pattern of trends as wheat, but at a generally lower level. This is due to a higher price for malting barley which requires a lower protein/nitrogen content in the grain and therefore limits the amount of N fertiliser applied. Barley yield trends are better than wheat in southern Victoria and southern WA where barley is better able to handle waterlogging and frosts. The average yields over the last 7 years of the timeframe in New South Wales were 1.92 t/ha for wheat and 1.71 for barley, in Victoria 2.03 and 1.84, in South Australia 1.60 and 1.78 and in Western Australia 1.60 and 1.68 t/ha respectively. This could indicate a comparative advantage for barley in some parts of Western Australia and South Australia which are the largest growers of barley.
Trends in average barley yields are shown in Figure 4-6. Notable features include small average increases (less than 40 kg/ha/year) in most of Queensland, in the Eyre Peninsula and Murray Mallee of South Australia, the Mallee and much of the Wimmera of Victoria, significant portions of the central grainbelt of Western Australia, and western margins of the cropping belt of NSW. There were high increases (greater than 60 kg/ha/year) in a belt between Tamworth and Dubbo and west of Canberra in New South Wales, in some irrigated areas of the Riverina, and southern Wimmera, the south-west of South Australia, and in the Great Southern region of Western Australia. To a limited extent it would appear that the highest increases occurred in regions with the highest reliability of winter rainfall (Figure 2-1).
Similar responses to those of wheat, ie within one group of yield change, occur in most of the grain belt. Striking differences to the patterns for wheat occur in the northern and central wheat belt of Western Australia. Isolated SLAs with larger differences occur in the Darling Downs (Jondaryan), in the Moree plains (Coonamble), the Riverina (Cootamundra, Culcairn), the MIA (Griffith, Murrumbidgee), northern Victoria (Molra East), the Wimmera (Horsham Balance), the Murray Mallee (Peake, Meninge) in the middle of the Yorke Peninsula.
Oats and other cereals
Oats and other cereals (rye and triticale) are also widely grown in the cropping belt (Figure 4-7). Oats do better in wetter/cooler environments, consequently yield trends are higher in southern Australia and negligible in northern areas where they are grown in very small amounts. The pattern of yield increases is more similar to that of barley than that of wheat. Generally yield increases are lower than for wheat and barley, particularly in northern New South Wales. Rye and triticale would have a competitive advantage over wheat and barley in soils that are acidic, which commonly occurs in southern New South Wales, Victoria and south-west Western Australia (SCARM 1998, p41). Also during the 1990s, special contracts for milling oats were offered in the highest yielding/ most reliable areas where the highest yield trends are found (southwest WA, central mid-north SA, southeast SA, southeast NSW). In these areas farmers improved the management practices of oats and started applying more nitrogen as oats became more profitable.
Figure 4-6 Trends in barley yields (kg/ha/year) for different Statistical Local Areas (SLAs) of Australia. (source ABS and AgWA)
Figure 4-7 Trends in oats and other winter cereal yields (kg/ha/year) for different Statistical Local Areas (SLAs) of Australia. (source ABS and AgWA
Grain sorghum is the most widely grown crop in summer ranging from a maximum 818,000 hectares in 1986-87 to a minimum of 376,000 hectares in 1990-91. Grain sorghum had a value of production of $257 million in 1996-97. The area sown is affected by summer rains and the price of product, which has been less regulated than wheat and barley.
Yields have generally increased in all regions of production (Figure 4-8), however the rate of increase varies enormously. Particularly noticeable is the higher rates of yield increase in the southern parts of the sorghum growing region (in northern NSW). In these areas, sorghum is often grown in rotation with winter cereals, fertiliser nitrogen is applied, weeds are controlled well, the crop is sown in late spring and conservation tillage has shown yield advantages in sorghum. Lower rates of increase occur in the northern parts of the sorghum region (in Queensland). Serious droughts in the early 1990s severely impacted on the economics of higher inputs and yield trends were low or negative, particularly around the Darling Downs where the drought was the most severe, and growing sorghum competes with many other crops, such as cotton, maize and millets. In the Central Highlands of Queensland sorghum is sown on large areas, with minimal fertiliser inputs, often in mid-summer when soil temperatures are high, rains come in storms, and there is often little rotation with other crops.
Water use efficiency in grain crops
Water use efficiency provides an index of how much water, both stored in soil and as rainfall, is used by the wheat crop. The remainder may run off, evaporate from the soil surface or drain into the subsoil groundwater. It is the drainage of excess water into subsoil layers and its subsequent lateral movement that is one of the contributing factors to dryland salinity. This excess water also represents an unused resource for achieving higher productivity. Water use efficiency in this exercise was calculated as the ratio of actual yield to the potential yield (as estimated by the STIN model).
For wheat there was a pattern (Figure 4-9) that areas receiving summer rainfall ie northern New South Wales and Queensland have a low water use efficiency (less than 50% of potential). However, in many cases these areas have the option of growing summer crops to use rainfall more efficiently. Higher water use efficiencies (greater than 70% of potential) occur within the MIA and along the Murray River, in the Wimmera, Yorke Peninsula and southern Eyre Peninsula, and in shires in the drier eastern part of the Western Australian wheatbelt.
Barley showed lower water use efficiencies in general (Figure not included here ). The overall pattern of lower water use efficiencies in summer rainfall and higher winter rainfall areas also applied to barley. For oats and other cereals (Figure not included here) even lower efficiencies were generally noted, with only irrigated areas of the MIA showing high efficiencies.
Figure 4-8 Trends in grain sorghum yields (kg/ha/year) for different Statistical Local Areas (SLAs) of Australia. (source ABS and AgWA)
Figure 4-9 Trends in the water use efficiency of wheat (% of total annual available water used by the crop) for different Statistical Local Areas (SLAs) of Australia. (source ABS and AgWA)
The water use efficiency of sorghum (Figure 4-10) also shows considerable variation from 20% to 80% of growing season rainfall over the last 5 years of the time frame. There is no clear relationship with yield trends, since some areas show low efficiency and low yield gains (eg in the northern parts), some show high efficiency and high yield gains (eg around Dubbo), some show low efficiency and high yield gains (eg to the west of Tamworth) and others show high efficiency and low yield gains (eg to the west of Toowoomba). During the 1990s the sorghum growing areas have experienced a range of summer rainfall conditions from droughts to floods that may well affect the water use efficiency calculations.
Nevertheless, the information would indicate that water use efficiency by these grain crops could be a major direction for improving not only productivity but also reducing contributions to dryland salinity in some areas. However, finding these improvements may sometimes be achieved at the cost of higher risks to profitability.
Figure 4-10 Trends in the water use efficiency of sorghum (% of total annual available water used by the crop) for different Statistical Local Areas (SLAs) in New South Wales and Queensland. (source ABS and AgWA)
Figure 4-11 shows yields of cane sugar in three different cane growing districts calculated from the AgStats data to illustrate some of the variation within the sugar regions. It shows highest yields obtained in Ayr, and little indication of any trend for yields to increase.
Figure 4-11 Yield in cane sugar for three SLAs of Ayr, Mackay and Burdekin from 1982-83 to 1996-97. (source ABS)
Potential sugar yields were calculated using a model designed by Dr Russel Muchow of CSIRO using local climate and soils information for similar years. The results in Figure 4-12 show that Ayr achieved a consistently higher percentage of its potential yield than the other shires. Even allowing for the potential yields being in error by 25%, the comparison would indicate that there are some unexplained constraints to achieving higher field yields in the Mackay area. It was also notable that actual yields in Mackay were more variable than Ayr and Bundaberg (Figure 4-11). This lack of yield increase, which has been noticed for some years, is a major concern to the industry. The constraints are not well understood and are a major topic under investigation by research bodies funded by the industry.
Figure 4-12 Comparison of actual cane sugar yields with estimates of climate limited potential yield for three SLAs of Ayr, Mackay and Bundaberg. (sources ABS and CSIRO Sustainable Ecosystems)
Total factor productivity (TFP) measures the change in quantity of outputs produced by a given quantity of inputs for whole enterprises and is based on detailed information on farm operations. Total inputs included fertiliser, fuel, seed, fodder and chemicals as well as labour and capital. Total outputs included all products exported from the farm such as grain, wool, hides, and animal sales. Therefore, TFP does not only measure changes in the productivity of the resource base. Rather it measures the net effect of factors such as technical efficiency, better production and animal husbandry methods, and the underlying quality of the resource base. After detailed investigation of the dairy dataset and consideration of the geographical spread of dairying regions, it was concluded it would not be possible to produce meaningful maps of TFP for the dairy sector.
Growth in total factor productivity (TFP) was calculated over the twenty years 1978-79 to 1997-98 using ABARE farm survey data (Ha and Chapman 2000). Regional variation in the attributes and economic performance of farms, a long recognised characteristic of the Australian farming sector, can be shown using local averages obtained through smoothing of sampled data, while not breaching the confidentiality of individual farms.
The technique uses smoothing parameters to determine to what extent nearby or distant farms contribute to a ‘local’ average. Using small smoothing parameters will result in more detailed maps, which give a reliable picture of farm diversity in regions where the sample density is high. However, it should be recognised that the smaller the smoothing parameter, the fewer farms contribute to the local average at each point, which can be a problem in areas where the sample is sparse. The smoothing parameter used depends upon the survey weight of the farm (a multiplier to align individual farm values with known regional totals) and the area of the farm. For more details see section in Appendix 2-Methods.
The map in Figure 4-13 indicates that significant variation in productivity growth has occurred across Australia. In particular, it appears that the distribution of broadacre cropping industries is a key factor contributing to the observed patterns. The largest productivity gains have occurred in the wheat-sheep zone where cropping activities are concentrated. Lower productivity gains appear to have occurred in the regions where livestock activities dominate the broadacre production mix. The areas of lowest growth are concentrated in the high rainfall zones where the combination of livestock focussed activities and small farm size may have contributed to the relatively lower productivity gains.
Figure 4-13 Trends (% change) in Total Factor Productivity over Australia during 1978-79 to 1997-98. (Source ABARE farm surveys)
With changes occurring between land uses, it is possible that effects of land degradation are not shown up. Estimating total agricultural (biological) productivity values across the different industries would provide a summary to show any immediate effects of land degradation. There are known conversions from cattle to sheep, and various conversion factors were calculated for the major broadacre crops based on their energy availability for livestock (metabolisable energy - ME). Dry sheep equivalent (DSE) was used as the constant unit, because it is are a common means of converting forage to a product. Pig and poultry production was ignored because they are largely fed from grain, and hay and silage production were assumed to be eaten by sheep and cattle. Horticultural production was too complex for these conversion (a very wide range of products with highly variable moisture contents and often unknown energy contents) and were left out of the calculation, but in terms of biological energy production they would usually be dwarfed by the other factors. More details are given in the section in Appendix 2-Methods.
For 1996-97, total agricultural productivity in the areas of intensive agriculture (Figure 4-14) was reasonably closely aligned to the distribution of rainfall presented in Figure 2-1. However, productivity in the higher rainfall zones of the tablelands of NSW, eastern Victoria and the Midlands of Tasmania were lower than might be expected from rainfall patterns. It is likely that cold winter temperatures limit potential plant growth in these areas although some of them are associated with lower rainfall reliability in winter (Figure 2-1). Particularly high levels of total productivity occur in coastal Queensland where sugar cane is grown, in the Darling Downs of Queensland where multiple cropping can occur, in the irrigation areas of southern NSW and northern Victoria and in the dairying areas of southern Victoria, northeast Tasmania and near Adelaide in South Australia.
Changes between the early 1980s (represented by the average of 1983, 1984 and 1985) and the mid 1990s (represented by the average of 1995, 1996 and 1997) are given in Figure 4-15. Overall most SLAs showed an increase in agricultural productivity between these times. The increases appeared greatest for the cane-growing districts of coastal Queensland, for the dairy districts of Victoria and Tasmania. Some areas appeared to show a decrease in productivity, particularly in the larger Darling Downs up to the Burnett in Queensland. The regional basis of this decline suggests that drought impacts during 1994 to 1996 are showing out. Other pockets occurred in northern NSW, and in two small parts of Victoria, which may be errors in the concordance of AgStats.
Overall there is little evidence to indicate that resource condition has adversely affected productivity over the past 20 years.