We have many major achievements to report for 2002. The first major success story is the fact that this project commenced on time in January 2002 due to significant effort and goodwill, and some risk taking, from officers of the parties involved CSIRO, ACIAR, ICAR, PAU, NSW Agriculture and/or their staff. We have come a very long way in the past 18 months, for a group of people who had never met before, separated by thousands of kilometres.

The opportunities provided by the ACIAR project have been enhanced significantly by additional funding from the International Atomic Energy Agency to assist the N dynamics studies at PAU, and a small training award from The Crawford Fund for one of the Indian scientists. Substantial funding was also secured to enable the Australian field experiments to be undertaken as proposed, with support from: the CRC for Sustainable Rice Production, the Rural Industries Research and Development Corporation/Rice Research and Development Committee, and the Grains Research and Development Corporation, and in-kind from Coleambally Irrigation Cooperative Ltd and Murray Irrigation Ltd.

A highlight of the year was the successful development of technology for direct drilling into rice stubble - the Happy Seeder was successfully built and tested at PAU with CSIRO. The Happy Seeder requires further refinement and evaluation, and will potentially remove the need for burning 10 million tonnes of rice straw in Punjab alone, with substantial impacts on air and soil quality.

A major key achievement was the establishment of excellent field experiments for comparing conventional and permanent bed rice wheat or alternative cropping systems on 3 soil types in Punjab and one soil in Australia.

Purchase and delivery of equipment for the PAU team is almost complete, and this is a significant achievement in the face of the large logistical and bureaucratic hurdles involved.

Links were established or strengthened between this project and other major modelling groups (DSSAT, APSRU, Wageningen) through a small international workshop held at CSIRO Griffith. Excellent links have been established with the key developers of DSSAT, and the project has a good base from which to progress the capability of modelling rice wheat systems under changing soil and water management.

Major experiments to compare permanent bed and conventional rice-wheat cropping systems in Punjab, India and rice-based systems in NSW, Australia, were established in 2002, and are now into their third crop (wheat in India, rice and soybeans in Australia).
The weather in Punjab 2002-03 was unusually foggy and cold mid-season, and rainfall was unusually high prior to heading. Yields of wheat on beds were comparable to yields on the flat at two sites with loam soils and lower on two sandy loam sites on the PAU farm at Ludhiana. The reasons for the poorer performance on beds relative to flats on the sandy loam at Ludhiana are unclear, and this result is in contrast with many reports of similar or better yields on beds. As the season was unusually wet, there was no effect of irrigation treatment on wheat growth or yield. Yields ranged from 3.5-3.6 t/ha on a reclaimed sodic soil where the crop suffered from lodging prior to heading due to storm damage, to 4.1-4.9 t/ha at the other sites.
Results from the 2003 rice season in Punjab showed no irrigation water savings on beds compared with puddled transplanted flats, and yield losses of 10-20 per cent for transplanted rice on beds compared with the recommended practice of puddled transplanted rice irrigated two days after the floodwater disappeared (PTR-2d). Yield loss was larger (around 30 per cent) with dry seeded rice on beds irrigated two days after the furrows had drained (DSRB-2d). Iron deficiency and nematodes contributed to yield loss on the beds, and hand weedings were required as chemicals did not provide adequate control. Irrigation water use was very high and ranged from 1,680 mm on PTR-2d on the loam to 4,950 mm for dry seeded rice on beds irrigated daily. Irrigation water use was roughly halved by irrigating two days after the floodwater had disappeared compared with daily irrigations to try and maintain continuous flooding in both PTR and beds.
Yields of 2002-03 rice on beds in Australia with continuous ponding between panicle initiation and maturity (10.1 t/ha) were significantly lower than yields on the flat (12.7 t/ha) due to the large gap (0.6m) between outside rows on adjacent beds. Input (irrigation plus rain) water use in these bed and flat treatments was similar (1,830 mm), and was reduced to 1,720 mm by maintaining water in the furrows only from PI to maturity. This resulted in a decline in yield to 9.4 t/ha and a reduction in water productivity from 0.68 to 0.55 g/kg. Land use after rice was wheat or fallow on flats and beds, and barley on beds only. Rice was sown on the fallowed beds in October 2003. The gap between outside rows on adjacent bets was narrowed to 0.3 m for the 2003-04 rice on beds by widening row spacing from to 0.22 m and sowing on the shoulders of the beds. Soybeans were sown on beds in December 2003 after barley harvest.
Soybean-wheat and maize-wheat field experiments were established in June 2003 in Punjab comparing bed and flat layouts and wheat straw mulching (6 t/ha) in the direct drilled treatments on beds and flats. There was no effect of layout or mulching on soybean (1.45 t/ha) or maize (5.5 t/ha) yields in 2003, a season with well-distributed rainfall, and as a consequence the soybeans only received one irrigation, while the maize was not irrigated.
A separate experiment evaluating the effect of straw mulching and N rate for rice-wheat on beds and flats in Punjab commenced in June 2002, and is now into its fourth crop (wheat). The soil is a sandy loam. In 2002-03, yields of rice and wheat on beds were lower than on flats, with large yield declines in the case of rice on beds, partly due to increased disease (brown leaf spot). These lower yields on beds on the sandy loam are consistent with the findings of sub-project 1. There was no effect of wheat mulch (3 t/ha) on rice yield, however yields of wheat mulched with rice straw (6 t/ha) were depressed at all but the highest N rate (160 kg N/ha). This experiment is being undertaken in collaboration with the IAEA Coordinated Research Project on integrated nutrient, soil and water management for rice-wheat.
A review of the performance of the CERES Rice and Wheat models showed that CERES Rice has generally performed well in its ability to predict anthesis and maturity dates in Asia. However, there are few reports of its ability to predict a wider range of parameters, and the few evaluations under water and N limiting conditions suggest that the model does not perform well under high stress conditions. Evaluation of CERES Rice in Australia has been limited and demonstrates the need for a chilling injury routine. Such a routine has been developed in the past but needs further revision and testing. The few evaluations of CERES wheat in the rice-wheat areas of Asia showed reasonable ability to predict anthesis and maturity dates and yield, but prediction of biomass is less reliable. There has been only one study evaluating the performance of the CERES rice-wheat sequence model under the DSSAT framework. In that study the sequential model performed fairly well in terms of yield prediction, but simulation of soil C and N was not realistic. Interaction with the developers of DSSAT and the CERES Wheat, Maize, Rice and SOYGRO models in the USA has increased to the level that hands on collaboration in model validation and applications has commenced.
The development of the Happy Seeder for direct drilling into stubble has continued to progress through additional funding from ACIAR and collaboration with Dasmesh Mechanical Works, Punjab, India. Trials established in 2002 at PAU Ludhiana showed good establishment and growth of wheat on flats and beds with up to 8 t/ha of rice straw mulch, although results in 2003-04 suggest that establishment is impaired with more than about 6 t/ha of mulch in some situations, but that sowing depth is also an important factor. Inspired by this work, a group in Pakistan has also commenced to make a Happy Seeder based on photos of the machine built in India.

Over the past twelve months there have been many highlights arising from this large project, in terms of new understanding and new science, new questions, machinery development, collaboration, communication, impacts, and outputs. The project teams in India and Australia have worked extremely hard and effectively to achieve this.

New knowledge, new science and new questions

In Punjab:
In 2002/3, the performance of wheat on beds on the sandy loam in India was inferior to that on the flats, contrary to many reports of higher or similar yields. Further analysis of the data suggested that the beds dried down more rapidly than the flats, and that water deficit stress probably inhibited tillering so that the crop on beds, sown at a lower rate, was unable to compensate for the lower sowing rate/wider row spacing. The practical implications of this are that the first irrigation after sowing wheat on beds on sandy loam soils should be brought forward, and/or sowing rate should be increased to minimise the problem in the first case. It highlights the importance of site specific recommendations for new technologies such as permanent beds, rather than blanket recommendations.
Further analysis of the data from the 2003 rice in India indicated that edge effects in small plots (drainage losses under bunds, lateral seepage, advection) can be huge. A comprehensive review revealed that reports in the literature of determinations of components of the water balance for rice as affected by soil and water management in the north west IGP almost never provide enough detail to indicate whether the results might be confounded by edge effects, and we suspect that many reports of rice plot water use and percolation losses (calculated from decrease in plot water depth) may exaggerate the true situation because of large edge effects in small plots. We were unable to completely eliminate edge effects in our replicated small plot experiments in 2004, even with well managed buffers, and have therefore commenced endeavours to monitor components of the water balance in large fields where edge effects will be more realistic.
The performance of rice on beds on sandy loam and loam soils in Punjab is consistently poorer compared with conventional practice, especially for direct seeded rice. We have not been able to overcome problems of iron deficiency in direct seeded rice, despite early applications of iron sulphate sprays. The reasons for poorer yields of transplanted rice on beds compared with puddled transplanted rice remain a mystery.
Wheat sown on fresh beds after rice performed much better than wheat sown on permanent beds which previously grew transplanted rice, which was in turn much better than wheat on permanent beds which previously grew direct seeded rice. This was a consistent finding on both the sandy loam and loam soils in sub-project 2. We suspect that this at least partly due to differences in soil structure, and analysis of some soil physical determinations is in progress. However in sub-project 4 on an adjacent sandy loam site, the performance of wheat on permanent beds after transplanted or direct seeded rice was similar. In this case there was more soil disturbance with direct seeding as the crop was sown by hand seed drill, and fertilizer was drilled into the centres of the beds. Appropriate soil management for permanent beds, as affected by soil type and crop sequence, is a big unknown.
There was a consistent trend across sites for lower grain yield of wheat sown on the flat compared with fresh beds or permanent beds that had previously had transplanted rice, with yield on fresh beds significantly higher than yield of direct drilled wheat on the flat at both sites .
After 2 rice and 2 wheat crops, we are not seeing any negative effects of continuously mulching wheat and rice straw on beds on the performance of the wheat on a sandy loam in Punjab where recommended fertilizer N rates are applied. There is evidence of an increase in organic C in the topsoil (from 4.1 to 4.5 g/kg).
Maize-wheat and soybean-wheat on a loamy sand - yields were similar on flats, fresh beds and permanent beds for the first two crops of maize, soybeans and the first wheat crop; whether this would be the case for the summer crops in a more typical, wetter monsoon season is yet to be seen.

In Australia:
Yields of soybeans on beds with furrow irrigation were higher than with subsurface drip irrigation (3.7 vs 3.1 t/ha), while irrigation water use was much less with subsurface drip (530 vs 920 mm). This resulted in higher irrigation water productivity for the drip (0.58 vs 0.40 g/kg or t/ML). The soybeans on drip probably suffered from water deficit, and there is a big knowledge gap in irrigation management for soybean (and other crops) grown using drip.
In both 2003 and 2004, yields of wheat on beds in Australia were similar to yields on the flat (~5.0 and 7.5 t/ha in 2003 and 2004, respectively).The beds were disadvantaged by the inability to control sowing depth in the furrows and poor establishment in 2003, but not in 2004.AStubble King seed drill with parallelogram assemblies was purchased through GRDC funding to solve the problem of sowing in both beds and furrows. In both 2003 and 2004 rainfall was relatively low, and we need to evaluate the beds in more normal or relatively wet winter conditions.
Yields of barley on beds with furrow or subsurface drip irrigation were similar (~5.3 and 5.7 t/ha in 2003 and 2004, respectively). Irrigation water use was lower with subsurface drip managed replace crop water use requirement on a daily basis, leading to higher applied (irrigation + rain) water productivity with the drip (1.47 g/kg) than the furrow irrigation (1.14 g/kg).
We are beginning to calculate various measures of water productivity, and need to put more effort into defining clearly which measures we are reporting, and in determining various measures of water productivity including irrigation water productivity (g grain/kg irrigation water or t/ML), applied (irrigation + rain) water productivity, and total (irrigation + rain + soil water depletion), so that we can readily make comparisons across sites, seasons and crops. In Punjab, wheat is extracting water at least to the sampling depth of 180 cm, and we are yet to determine if it is extracting water beyond this depth, or if we are seeing deep drainage beyond this depth, in any of the crops. There are some suggestions of deep drainage beyond 2.4 m in the rice. Rigorous determination of all components of the water balance remains a major challenge for our project, and we have a backlog of soil water data to analyse.
Based on the limited data available CERES Rice seems to perform reasonably well for rice in north west India, but not so well in Australia in cold years; the lack of comprehensive, good quality data sets for model calibration, validation, refinement and application is a major limitation. Our review on the performance of CERES Rice and Wheat strongly reinforces the need for rigorous calibration and validation before application for the environment and problem of interest.
The development of a chilling injury routine with Dr Upendra Singh of IFDC for CSM-CERES Rice ver. 4.0 is a nice example of new science developed through the ACIAR project and collaboration, although the model requires further validation; this collaboration was possible due to funding from the Rice CRC to enable Dr Singh to spend 6 weeks at CSIRO Griffith.
The CSM-CROPGRO-soybean model simulates growth and development of irrigated soybeans in southern NSW extremely well for crops sown around the optimum time. However the model overpredicts yield for early sowings, due to inability to take into account the 'aging' effect associated with the longer duration of earlier sown crops. It also under predicts yield for late sown crops due to lower temperatures during the reproductive stages.

Machinery development
2003/4 results with the Happy Seeder in India for wheat sown into rice stubble were promising, but also revealed problems of patchy establishment associated with high straw load and/or uneven soil. Achieving good establishment under the range of likely conditions (e.g. soil type, soil moisture, amount stubble, time of sowing etc.) is the next challenge that the project is now addressing.
5 Happy Seeders for direct drilling into stubble have now been built by Dasmesh Mechanical Works in Punjab, India, in collaboration with our project team,a major advance being the 'Combo Happy Seeder'. Thisis a compact lightweight single unit with the capability of altering the cutting height and cutting strips in front of each sowing tyne. Combo Happy Seeders with inverted T ('zero till') and strip tillage have now been constructed. The Combo seeder with ZT drill was used to successfully establish mungbean in wheat stubble. In two wheat crops sown into rice stubble in the 2004/5 wheat season, establishment was much better using strip tillage.
Funding was provided by Twynam Pastoral Company Pty Ltd to build an Australian version of the Happy Seeder, with a 4 m wide seed drill with a parallelogram sowing assembly providing the capability of sowing on beds/furrows and flats. The machine was recently used to establish soybeans in wheat stubble.

Collaboration, communication and impact
By it's very nature, the project is facilitating good collaboration between PAU, CSIRO and NSW DPI. However, the project has also resulted in several other very strong collaborative relationships - in particular those with Dasmesh Mechanical Works in Punjab, with Twynam Pastoral Co in Australia, and with leading DSSAT modelers at various institutions in the USA.
Through the ACIAR project, Drs Sidhu and Dhillon undertook a highly successful visit to Pakistan to share experiences with stubble management and bed planting technologies, making many presentations and influencing senior government officials; they were hosted by the Dr Shabbir Kalwar of the Farm Mechanisation Institute, Pakistan Agricultural Research Council. In 2003 Shabbir completed construction of the FMI seeder, which is a similar concept to the Combo Happy Seeder.
We have made many presentations on our work at field days, farmer meetings and national and international meetings and conferences, and to numerous national and international visitors to our field sites in India and Australia.
Two scientists from PAU and Tribhuvan University (Nepal) received training on the DSSAT/CSM-CERES Rice and Wheat models at CSIRO Griffth - the training of the Nepalese scientist was made possible through financial support from The Crawford Fund.
the Rice Wheat Consortium is about to purchase 10 Happy Seeders and provide them to Dr Sidhu, together with 10 double disk drills and a punch planter, for further evaluation and demonstration of technologies for direct drilling into rice stubble.

Sub-project 2. Field evaluation of permanent bed and conventional RW/rice-based cropping systems in Punjab/NSW

India
Field experiments on a sandy loam and a loam soil were continued with the 3rd wheat crop (5thcrop) in 2004/5 followed by the 3rdrice crop in 2005. As in previous years, grain yields of wheat on both fresh and permanent beds were around 20% and 10% lower than yields with conventional tillage on the sandy loam and loam soils, respectively. Yields on fresh and permanent beds have been comparable to date, suggesting no significant effects of "aging" of the beds to date, and there is no trend for decline in relative wheat yield with time on the permanent beds. We need to look at methods of increasing leaf area faster to identify whether this is a causal factor, using methods such as earlier sowing (not possible if we continue with direct seeded rice in the rotation) or increasing the number of rows on the beds.

Yields of both transplanted and direct seeded rice on beds have continued to decline as the beds age relative to continuously flooded puddled transplanted rice (PTR-CF) - to very low values of about 40% and 20% of PTR-CF in 2005. The problem of iron deficiency in direct seeded rice and beds is getting worse with time, despite daily irrigation for the first 6 weeks and application of 15 t/ha of farmyard manure prior to sowing to increase soil organic C and try create more reducing conditions. Causes of the continuing decline with transplanted rice on beds are unknown.

Monitoring of the dry down of the soil after the last irrigation of rice in 2005 showed that drainage continued out of the profile (0-1.8 m) for about 4 weeks. By the time of wheat sowing, the profile was only about half full on both soil types. The results show the free draining nature of these soils, and have implications for accurate determination of the water balance of individual crops and the total cropping system.

In farmer-size fields (~0.5 acre), yields of transplanted rice on fresh beds were about 10% lower than PTR-CF, with less than half the amount of irrigation water.

Australia
In Australia, establishment of rice on beds and bed shoulders was excellent using the Stubble King seeder with
parallelogram disc assemblies, however, extremely cold conditions during early pollen microspore resulted in 41-65% floret sterility, bringing yields down to 6.1 t/ha on the beds which were ponded between panicle initiation and anthesis, and 3.5 t/ha on the non-ponded beds (compared with the Coleambally district yield of 5.7 t/ha). Yield and input (I+R) water productivity of soybeans on beds were slightly higher with sub-surface drip irrigation (3.3 t/ha, 0.60 t/ML) compared with furrow irrigation (3.1 t/ha, 0.54 t/ML). Establishment and growth of wheat sown in May 2005 were similar on beds and flats up to anthesis.

In October 2005, all treatments except those with wheat in 2005 were sown to rice, with 4 N rate sub-plots, to compare performance for the rice-based system intercropped with barley and soybeans versus a rice-fallow system on beds, and the rice-fallow-wheat-fallow-rice on flats and beds.

Sub-project 3. Maize-wheat and soybean-wheat systems in Punjab

Field experiments were continued with the 3rd wheat crop (6thcrop) in 2004/5, followed by the 4thmaize and
soybean crops in 2005. As in previous years, growth and yield were similar for all treatments - fresh and
permanent beds, conventional tillage and direct drilling on the flat. Irrigation applications were always lower on the beds (by design), resulting in higher irrigation water productivities on beds, presumably due to greater deep drainage losses on the flats. Preliminary results indicate deep drainage below 2 m in both beds and flats associated with irrigation. Evaluation in farmer-size fields is needed for realistic determination of components of the water balance because of the short irrigation time (a few minutes) and reduced opportunity for deep drainage losses in small plots compared with large fields (a few hours).
The comparable performance of wheat on beds and flats in this sub-project continues to be in contrast with the findings in the rice-wheat systems in both sub-projects 2 and 4. This suggests that the difference may be
associated with the long history of rice in the rotation (>20 years) in the rice-wheat systems, in comparison with the loamy sand which has never grown rice. This needs further investigation.

Sub-project 4. Nitrogen use efficiency in RW cropping systems in Punjab (with IAEA)

Field experiments continued with the 3rd wheat crop followed b the 4th rice crop. As in sub-project 2, rice yields continued to decline on the permanent beds with both transplanted and direct seeded rice, relative to the puddled transplanted control with the same irrigation scheduling. Mulching had no effect on transplanted rice yields on beds, but reduced growth and yield of direct seeded rice on beds as in previous years. Iron deficiency and nematodes were again serious problems for direct seeded rice.

As in sub-project 2, yields of wheat on permanent beds have been consistently lower than yields of conventionally tilled wheat on the flat, but with no declining trend as the beds age.

After 7 crops with mulching, soil organic C at 0-15 cm has increased significantly by 0.6-0.7 g/kg or 12-13%.

Sub-project 5. Modelling to identify management options for maximizing the productivity of RW cropping systems

CSM-CERES-Rice predicted maturity and grain yield of puddled transplanted rice on the flat in our ACIAR
experiments (India) reasonably well. Agreement between predicted and observed values on the beds was much poorer, which is not surprising given the many non-modelled factors that appear to be reducing rice yields on the beds. CSM-CERES-Rice currently does not have the capability to automatically schedule irrigations according to the duration of the non-ponded period after the floodwater has gone - the recommended practise in NW India is to irrigate 2-3 days after flooding has ceased. We are currently programming this capability into the model, to allow realistic scenario analyses using historic (30 years) weather data.

CSM-CERES-Wheat predicted yield of wheat on both flats and beds reasonably well for our ACIAR experiments in India. However, it does not have the capability of automatic irrigation based on cumulative net ET (ETrain) or growth stage. Irrigation scheduling guidelines for wheat in NW India are based on cumulative net ET. We are currently programming both these capabilities into the model to enable realistic simulations using historic weather data.

Model simulations for wheat in Australia showed that both yield and water productivity vary greatly with seasonal conditions. Irrigation to avoid water deficit stress greatly increased yield and water productivity, and sprinkler irrigation had higher irrigation water productivity due to slightly higher yield, lower irrigation amount and lower deep drainage losses. Timely sowing also improved both yield and water productivity.

Crop modelling capacity in Nepal has been greatly enhanced as a result of support from The Crawford Fund,
enabling two Nepalese visiting scientists to receive on-the-job training with the ACIAR team at CSIRO Griffith in 2004 and 2005. This was followed by a week-long training course for 10 participants at Tribhuvan University, Nepal. As a result, Tribhuvan University has established a multidisciplinary crop modelling group, and two students have commenced Masters degrees in crop modelling.

Sub-project 6. Economic impact assessment of permanent bed systems

The economic analysis of permanent bed systems for rice-wheat, soy-wheat and maize-wheat in Punjab was
revisited by the Australian economist (Dr Rajinder Pal Singh) in collaboration with the Indian team. The approach was expanded to include financial analysis, economic analysis (includes corrections for distortions in financial values due to government intervention e.g. power subsidy) and crop sequence analysis for a 20 year duration (includes effects such as decline in yield of transplanted rice on permanent beds during first 4 years, yield increase of wheat in soy-wheat rotation after the third soybean crop). The findings will be reported once the inputs, assumptions and scenarios have been finalised.

A preliminary analysis was done of the impact of switching to permanent raised beds from current layouts used on Australian rice farms. A typical crop rotation was identified for each irrigation layout, and crop sequence gross margins were prepared and compared with the gross margin of a crop rotation on permanent lateral raised beds. There was an increase in the present value of the aggregated crop sequence gross margin in the long term for
progressively more developed irrigation layouts. While these preliminary results are very encouraging, they didn't take into account the development costs of the different layouts. The next steps are to undertake a benefit cost analysis of such changes in irrigation layout including development and operating costs, and to identify and compare more rotations with varying lengths for each field irrigation design.

Sub-project 7. Development of stubble management machinery for direct drilled flat and bed farming
systems

Dasmesh Mechanical Works and PAU have developed the Turbo Happy Seeder, a very clever approach which eliminates the chute, greatly reduces the amount of dust, and the sowing lines are now much more exposed (with benefits of both better establishment and more accurate sowing). Dasmesh have established trials in ~100 acres of farmers' fields with good results to date (establishment and early growth), watched by over 1,000 visitors. Results with the earlier Combo+ Happy Seeder (with strip tillage and a chute) have also been very good, and procedures are underway for the Happy Seeder package to be recommended by Punjab Agricultural University and included in next winter's package of practices for farmers in Punjab.

The Rice-Wheat Consortium purchased 6 Combo+ Happy Seeders which were sent to 5 locations in India and to the ADB project in Punjab, Pakistan, where it was demonstrated at 7 locations in farmers' fields.

Experiments have been initiated to guide development of N management and irrigation strategies for mulched wheat sown with the Happy Seeder. Preliminary results indicate a cumulative saving of 30 mm prior to the first three irrigations due to reduced soil evaporation on a loamy sand in 2004/5, but little saving on a sandy loam where soil water content was very low prior to the first irrigation.

see final report