1 Ravensdown Fertiliser Co-operative Ltd., P.O.Box 1049, Christchurch, New Zealand
2 New Zealand Centre for Precision Agriculture, Massey University, Private Bag 11-222, Palmerston North, New Zealand
Phone +64 (3) 353 4632, Fax +64 (3) 353 4625
A three year study using three sites in Canterbury New Zealand has shown that there is financial (and environmental) advantage in adopting site specific farming methods. Most of this benefit is derived from managing the crop in a way that is sympathetic to the variation in the paddock, variation such as soil depth and moisture retention. This means varying the timings of nitrogen applications as well as the rate and protecting this investment with appropriate plant protection measures. The work has also highlighted ‘at risk’ areas of the crop as far as disease, invasive weeds and nitrate leaching is concerned. Greater effort needs to be devoted to helping growers derive crop management strategies which go beyond the concept of simply varying rates of fertilisers or other chemical inputs at the time of application.
Interest in precision farming technology has been heightened in recent years in New Zealand by arable farmers now having access to yield monitors and variable rate application equipment. This has highlighted the need to demonstrate to farmers the usefulness of such technologies by understanding the temporal and spatial variability that occurs within a paddock (Cook and Bramley, 1998) and an understanding of crop physiology.
Canterbury is the major cropping area in New Zealand. By world standards paddock size is small, 5-30ha and some may query the justification for applying precision agriculture to small areas. However yields can be high, it is not uncommon to achieve over 10t/ha with feed wheats and 2t/ha with ryegrass seed crops, therefore high nitrogen and plant protection inputs can be justified, especially where irrigation is available. Crop rotations are diverse compared to many overseas. A rotation may include cereals, pulses, specialist small seeds including clover and other legumes, pastoral and amenity grasses, brassica and vegetable seeds and increasingly process crops such as onions, carrots and potatoes. Contract grazing of dairy cows on grass seed and specialist greenfeed crops and increasingly the contract growing of silage crops add to the intensity of the rotation. Although it is now less common to have long pastoral phases, the use of ryegrass/clover pasture and legume seed crops contribute more heavily to nitrogen inputs than anywhere else in the world. This means there is a large divergence in the requirements and economic merits of using fertiliser nitrogen. The challenge becomes how to manipulate inputs and take advantage of paddock variability, especially given that soil types vary from light stony soils to heavy silts and clays and soils are predisposed to summer moisture deficits.
In 1998 a three year study commenced in Mid and South Canterbury involving three paddocks. The three sites were; West and Seadown sites, both irrigated light to medium soils, and the Howey site, on non irrigated downlands of medium soil texture. The three paddocks were grid sampled (60 or 100 points/paddock on a 40 or 50m grid) for intensive monitoring of soil, herbage and grain parameters including soil depth to clay or stones, soil moisture status, nutrient analysis and dry matter determination. Where applicable crops were also scored for height, vigour and disease. More complete details are given in Craighead and Yule, 2001. In the 2000-01 season, duplicated subplots containing extra nitrogen (N) over the grower practice (on all three sites) and less N on the two winter wheat sites (Wests, Seadown sites), were positioned in three or four areas of each paddock to ascertain the degree of response to N in better and poorer performing areas of the paddock.
From this study it has been possible to undertake some economic assessments relative to yield or quality and to calculate breakeven points for differentially applying various practices. The trends in yield maps, current financial returns and the results of previous trial work have been used to make these assessments. Some suggestions for better management practices for specific crops and different seasons are also given.
Four years of yield maps from this site identify the paddock could be split into four potential zones. Zone A is a plateau of low to medium yield; zone B encompasses a slope between the hill plateau and the flats and has less topsoil and low yields; zone C consists of flats of more recent alluvial soils and soil removed by cultivation from the slope, soils with greater depth and higher yields; zone D is another area on the flats where water ponds and yields vary according to the weather conditions, Figure 1.
Figure 1. Possible management zones, Howey site
Using the example of bread milling wheat, if N conditions were uniform, for a given amount of N, higher yielding areas would normally have a lower grain protein (Craighead and Burgess, 2000). However within each zone there are areas with relatively higher and lower yield and higher and lower protein indicating there is scope to improve returns. In this scenario, more N could be applied to zone C, the largest single zone in the paddock, to improve yield and protein, Table 1. Furthermore if the grower had access to modern spreading equipment some of this N could be reallocated from zone B without greatly impacting on the returns. If the grower could adjust planting rates he could further reduce sowing rates in this zone and save costs. Of greatest concern in this paddock is zone D where ponding can occur. This depresses establishment and tillering in the early stages and increases weed invasion. Ideally early N applications should be reduced to minimise leaching and N rates increased later in the season above the paddock average especially if the season turns dry as this area will grow for longer. Long term this area needs to be tile drained to improve yields.
Table 1. Economics of applying N differentially for a milling wheat crop, Howey site
A - ‘normal N’
B - reduce inputs (if practical)
C - increase inputs for most crops unless dry
D - adjust timings/rates to suit season (moisture) - drainage
Parameters – assuming normal application of 120kgN/ha 1
6.8t @ 11.8% protein ($271.59/t) 2
7.0t @ 12.8%protein ($285.03/t)
Less 40kgN (urea $442/t) + spread ($8/ha)
4.7t @ 11.8% protein ($271.59/t)
4.6t @ 11.6% protein ($268.90/t)
Loss of return
Saving 20kgN (urea $442/t), spread saving
- $ 11.96 3
1 yield and quality data based on data from Craighead and Burgess, 2000
2 Protein – Champion Mills 2000/01 contract (for cultivars Domino, Monad, Otane)
3 Management changes to complement lower N inputs could easily reverse these losses (see text)
In 2000/01 subplots of N were set up in spring barley to which 37kgN of planter N had already been applied. These showed yield and responsiveness to N was greatest in zone C > zone D > zone A. It was only economical to apply 50kgN as a sidedressing to zone C (a 2:1 payout for N where barley contracted @$200/t). However given the resulting summer/autumn drought and the resultant demand for feed barley, if the barley had not been on contract it would have been economical to apply N to more of the flats.
In practical terms the grower would make the major economic gains from treating zone C differently to the rest of the paddock, preferably with some adjustment in timing of N for zone D or remedial drainage action.
On this site it has not been possible to use the first years yield map as late uncontrollable disease decimated the crop. However the second and third year maps and the subplot data for N indicate that the central and southern end of the paddock may in general yield higher than the northern end, Figure 2. This soil is extremely stony and although depth to stones as assessed by grid sampling is not a good indicator of yield response there is some relationship with soil organic matter. In 1999/00 when field peas were grown, a protracted wet spell through the mid to late vegetative stages caused a secondary weed infestation and competition for assimilates at pod fill. Consequently it was the intermediate vigour areas that yielded better. It would have cost $36-79/ha (depending on the weed spectrum) to apply a second weed spray differentially to the more infested areas in the bottom half of the paddock. The breakeven point would be 0.1-0.3t/ha or a 3-8% yield increase above the paddock average yield (3.5t/ha), an easily achievable goal.
Figure 2. Organic matter and yield maps, Seadown site (darker areas represent a higher value)
In 2000/01 the paddock was in feed wheat worth $210/t, (this season cereals contracts are worth 15-20% more) and received on average 240kgN/ha. Subplots indicated a yield response to 50 or 100kgN above 190kgN with an average payout (see table 1 for N costs) to the extra N of 3:1. However the responses varied within the paddock. It was slightly economic to apply more than 190kgN in the northern end of the paddock, but only where the paddock was not very stony. Conversely there was a 20% yield response (almost 2t/ha) to applying 240 or 290kgN/ha to the southern end of the paddock, a payout to the extra N of approx 5:1 or >$300/ha extra profit. It is also worth noting the paddock edge effects in the 2000/01 yield map. These reflect the yield and financial loss associated with replanting 7 weeks later due to overwet conditions at sowing, when the crop was a true winter wheat that required vernalisation.
In practical terms in this paddock it would be appropriate to increase inputs in the southern part of the paddock in accordance with the crop grown and the conditions experienced. Additional measures include reducing the risk of wet establishment conditions in some northern parts of the paddock by removing some shelter belts and fencelines.
This site has a gradual change in soil type decreasing in soil depth from a silt loam to a stony silt loam as you move from south to north. The yield maps reflect this change in soil type with grass seed yields better on the lighter soil 1998/99 (see next section) and rape seed 1999/00 and feed wheat 2000/01 yielding better on the heavier soil, Figure 3. In the rape seed crop yield was strongly related to density/vigour r = 0.56. With the cost of an extra 50kgN equating to an extra 45kg seed @ $1.25/kg and with an average yield of 1000-1200kg seed/ha this equates to an extra 4-6% yield increase. Under present management the best option is probably to apply extra N to those areas with average vigour as on the lightest soil moisture is likely to be the most limiting factor. However if irrigation frequency could be increased on the lighter half of the paddock there would be scope to increase N inputs on this part of the paddock (nutrients were not seen as a limiting factor on this crop or paddock).
Figure 3. Yield maps and depth to stones, West site (darker areas represent a higher value)
The feed wheat crop received 300kgN/ha, significantly more than is usually applied, a reflection of early spring N leaching losses and a new high yielding cultivar. Subplots with variable N rates showed an average 5% yield increase (average yield 11t/ha) to the last 50kgN applied (equating to >2:1 payout for this N), however the response varied from 0-10% and it was independent of the soil type. By contrast the heavier soil averaged 10% higher yield than the lighter soil again showing the potential on this paddock to irrigate the lighter areas more frequently to improve returns.
Grass seed 1998/99
In the 1998/99 season grass seed was grown on the West and Howey sites. Some relationship was evident between yield and dry matter after closing but there was a stronger negative relationship with herbage N content at this time (Figure 4.), an accepted way of monitoring the responsiveness of grass seed crops to N (Rowarth et al., 1998, 1998a). Previous trial work (Craighead, unpublished data) indicates maximum seed yields are generated by N treatments which maintain intermediate growth usually because excessive herbage reduces threshing efficiency and because late N generates secondary vegetative tillers rather than reproductive tillers. Therefore control of herbage by reducing N inputs on the heavier parts of the paddock may be one way of improving seed yield. However an alternative may exist, use of the plant growth regulator ‘Moddus’ at 650mls/ha ($95/ha applied) appears to dramatically improve yields through controlling herbage mass. It only takes 60-70kg/ha of extra seed (or a 3-5% yield increase) to pay for the ‘Moddus’, suggesting that rather than putting less N on heavier soils, more N could be used on the lighter soils with adequate water, to further improve yield. This has still to be verified scientifically.
Figure 4. Grass seed yield in relation to dry matter and herbage M content. (darker areas represent a higher value)
Seasonality of responses
In the 2000/01 season yield maps showed less variability than in previous seasons, when yields in general were good, this despite drought conditions being prevalent from December onwards. Using meterological data from the Winchmore Research Station in Mid Canterbury to compare August to December data for the past three seasons, the growth pattern differences become evident. August and September 2000 were warmer and wetter than the previous two seasons with less frosts than usual in September. October and November were marginally cooler than in the previous two years but evapotranspiration was higher indicating that sunshine was not a limiting factor. It wasn’t until December that conditions became warmer and drier than the previous two years. This meant that spring growth conditions in 2000 were ideal for temperate crops as there was a less defined change from cold/wet winter conditions to warm/dry summer conditions. It is likely that in seasons where growth conditions are not ideal that differences in soil and moisture status will manifest themselves more clearly. In these seasons it is likely there is more financial gain to be made by differentially applying management to the farm.
Precision agriculture technology offers some opportunities to improve returns to growers. However interpretation may not be easy, inputs for different management zones may need to be adjusted according to the crop type grown, the season and the level of crop management applied. This requires a data base of yield maps, soil and climatic information and some knowledge of crop agronomy in addition to the equipment. This study also shows that in general financial returns are likely to be greater by boosting inputs in the better zones of a paddock rather than necessarily trying to lift performance of the poorer performing zones.
To David and Roger West, David and Jenny Howey, the Canterbury Precision Farming Group, Tim Phillips, Andy Howie and Ravensdown fertiliser Co-op., Canterbury field staff. This project was jointly funded by Agmardt, Ravensdown Fertiliser Co-op Ltd and FAR.
Cook, S.E. and Bramley, R.G.V. (1998). Precision agriculture – opportunities, benefits and pitfalls of site-specific crop management in Australia. Australian Journal of Experimental Agriculture, 38, 753-763.
Craighead, M.D. and Yule, I.J. (2001). Agronomic Interpretation of Precision Farming Data Maps – some results from a Canterbury arable farming study. In: Proceedings of the Workshop ‘Precision Tools for Improving Land Management’ 14-15th February, L.D. Currie and P. Loganathan ed. Occasional Report No. 14 Fertiliser and Lime Research Centre, Massey University, Palmerston North, New Zealand In press.
Craighead, M.D. and Burgess, W.B. (2000). The influence of late nitrogen applications on autumn sown milling wheat. Agronomy New Zealand, 30, In press.
Rowarth, J., Cookson, R. and Cameron, K. (1998). Measuring nitrogen in ryegrass – relationship to seed yield. Herbage Arable Update No.13, August, Foundation for Arable Research, Lincoln, New Zealand.
Rowarth, J., Cookson, R. and Cameron, K. (1998a). Optimising N application in ryegrass. Herbage Arable Update No. 14, August, foundation for Arable Research, Lincoln, New Zealand.