- Top of page
A method is presented to quantify the net effect of disease management on greenhouse gas (GHG) emissions per hectare of crop and per tonne of crop produce (grain, animal feed, flour or bioethanol). Calculations were based on experimental and survey data representative of UK wheat production during the period 2004–06. Elite wheat cultivars, with contrasting yields and levels of disease resistance, were compared. Across cultivars, fungicides increased yields by an average of 1·78 t ha−1 and GHG emissions were reduced from 386 to 327 kg CO2 eq. t−1 grain. The amount by which fungicides increased yield – and hence reduced emissions per tonne – was negatively correlated with cultivar resistance to septoria leaf blotch (Mycosphaerella graminicola, anamorph Septoria tritici). GHG emissions of treated cultivars were always less than those of untreated cultivars. Without fungicide use, an additional 0·93 Mt CO2 eq. would be emitted to maintain annual UK grain production at 15 Mt, if the additional land required for wheat production displaced other UK arable crops/set aside. The GHG cost would be much greater if grassland or natural vegetation were displaced. These additional emissions would be reduced substantially if cultivars had more effective septoria leaf blotch resistance. The GHGs associated with UK fungicide use were calculated to be 0·06 Mt CO2 eq. per annum. It was estimated that if it were possible to eliminate diseases completely by increasing disease resistance without any yield penalty and/or developing better fungicides, emissions could theoretically be reduced further to 313 kg CO2 eq. t−1 grain.
- Top of page
Work to assess how climate change might affect plant diseases was reviewed recently by Garrett et al. (2006) and is an active area of research (e.g. Evans et al., 2008). This paper concerns the reverse: how diseases and their control might affect climate. The mechanisms by which fungal pathogens of wheat affect net greenhouse gas (GHG) emissions are used as an example.
Most of the economic and GHG costs of wheat production are associated with the establishment and growth of a green canopy, which needs to be large enough to intercept most of the incident photosynthetically active radiation (PAR) by the start of the yield-forming period. This requires sufficient available nitrogen (N) to grow a canopy with a green area index of at least 6, and 3 g N m−2 green tissue (Sylvester-Bradley et al., 1997). Fertilizer N accounts for approximately 70% of the total GHG emissions associated with wheat production (Mortimer et al., 2004), when calculated in terms of the equivalent global warming potential of CO2. The remaining emissions arise from the energy used for soil cultivations, sowing, the application of agrochemicals, harvesting and drying the crop.
Hence, the majority of GHG and economic costs are incurred well before the start of yield formation, in the expectation that the investment in photosynthetic machinery will be repaid through dry matter assimilation during grain filling. Unfortunately, epidemics of foliar diseases usually have their greatest effect during the yield-forming period, after production of new green tissue has ceased. The effects of pathogens on plants were categorized by Johnson (1987) as type-I systems, which reduce radiation use efficiency (g dry matter fixed per MJ of PAR intercepted by the crop canopy), and type-II systems, which reduce radiation interception, usually by causing premature loss of green area. In both cases, disease reduces the weight of grain obtained in return for a given level of GHG emissions.
It is questionable whether carbon fixed by annual crops has a substantial net effect on atmospheric CO2, because the carbon removed from the atmosphere during growth is re-emitted when the plant product is consumed, burned or decomposes. Hence, the largest effect of disease on climate is likely to be through increasing the area of cropped land that is required to satisfy demand for grain, thus increasing the GHG emissions per tonne of grain.
Diseases may also affect GHG emissions by altering the processability of plant products. For example, diseases have been shown to affect the protein concentration of wheat grain (Dimmock & Gooding, 2002) and grain with more protein gives a lower yield of bioethanol (Smith et al., 2006; Kindred et al., 2007). In addition, there are GHG costs associated with disease control, for example through the manufacture and application of fungicides (Lal, 2004).
The important foliar and stem-base diseases of wheat in the UK are caused by fungi, principally Mycosphaerella graminicola (anamorph Septoria tritici) (septoria leaf blotch), Puccinia triticina (brown rust), Puccinia striiformis (yellow or stripe rust), Blumeria graminis (powdery mildew), Oculimacula acuformis and O. yallundae (eyespot). Of these pathogens, M. graminicola generally causes the greatest reduction in UK wheat yields. All of the diseases listed above are reasonably well controlled in commercial crops in the UK by a combination of host resistance and fungicides (Hardwick et al., 2001). Current fungicide inputs, and the amount of plant breeding effort devoted to selecting for disease resistance, are both driven by market forces to minimize the economic unit costs of production. Different levels of inputs and breeding effort may be appropriate to minimize the GHG unit costs of production.
This paper presents a simple analytical method to quantify the potential net effect of disease and disease control on GHG emissions per hectare of wheat and per tonne of crop produce (grain). The impact on processed products of wheat grain (animal feed, flour and bioethanol) was also considered. Only a small proportion of wheat is made into bioethanol, but this product is of particular relevance for this study because there is already a requirement to report GHG emissions associated with the manufacture of bioethanol under the UK Renewable Transport Fuels Obligation (RTFO).
Fungicide-treated and untreated crops of elite cultivars were compared, to calculate the contribution made by fungicides to minimizing GHG emissions. The relative GHG efficiency of cultivars with contrasting levels of disease resistance were then calculated. Lastly, the maximum potential improvement which might be achieved if diseases could be controlled completely was calculated.
- Top of page
The production and harvest of 1 ha wheat was estimated by the model to be associated with emissions of 3187 kg CO2 eq., based on the default values described (Table 1). The production and drying of 1 t grain was associated with 408 kg CO2 eq., assuming grain and straw yields of 8 and 5 t ha−1, respectively. The production of 1 t animal feed, wholemeal flour or bioethanol was associated with the emission of 478, 532 and 1477 kg CO2 eq., respectively (Fig. 1). About 75% of the GHG emissions associated with producing, harvesting and drying wheat were calculated to result from the manufacture and application of nitrogen fertilizer. Diesel used during farm operations accounted for 12% of GHG emissions and pesticides (including fungicides) accounted for less than 1% of emissions. Processing the grain accounted for 13, 21 and 18% of the GHG emissions associated with producing animal feed, wholemeal flour and bioethanol, respectively. As a result the percentage of GHG emissions associated with the manufacture and application of nitrogen was between 56 and 62% for these products.
Figure 1. Calculated relationship between grain yield per hectare and greenhouse gas emissions associated with the production of 1 t grain (a), animal feed (b), wholemeal flour (c) or bioethanol (d). All crop inputs apart from grain yield were kept the same for these calculations.
Download figure to PowerPoint
Changes in grain yield, whilst keeping all other model parameters the same, were estimated to have a substantial effect on the GHG emissions per tonne of grain, animal feed, wholemeal flour and bioethanol (Fig. 1). GHG emissions per tonne were inversely related to grain yield. GHGs per hectare were allocated between the grain and straw in direct proportion to the economic values of grain yield and straw yield per hectare. If grain price was low (£60 t−1) and baled straw price was high (£30 t−1), then the GHG emissions for a crop yielding 8 t grain ha−1 with 5 t ha−1 straw were calculated to be 357 kg CO2 eq. t−1 grain and 145 kg CO2 eq. t−1 straw. If grain price was high (£140 t−1) and the baled straw price was low (£10 t−1) then the GHG emissions were calculated to be 426 kg CO2 eq. t−1 grain and 131 kg CO2 eq. t−1 straw.
Experiments carried out in the UK to compare fungicide-treated yields with untreated yields for 842 cultivar/site/season combinations showed that, on average, disease reduced yield from 10·20 to 8·42 t ha−1 (Table 2). This level of disease-induced yield loss, after allowing for the reduction in GHGs associated with the manufacture and application of fungicides, resulted in a net increase in GHG emissions from 327 to 386 kg CO2 eq. t−1 grain (Table 2). GHG emissions per tonne of bioethanol were estimated to increase from 1235 to 1421 kg CO2 eq. t−1 for crops grown without fungicide.
Table 2. Measured fungicide-treated and untreated grain yields from wheat cultivar evaluation experiments in the UK, with calculated greenhouse gas emissions
|Year||Cultivar||Resistance rating against Mycosphaerella graminicola, (9 – high, 1 – low)||No. of treated/untreated cv. comparisons||Grain yield (t ha−1) 85% DM||kg CO2 eq. t−1 grain|
|2004–06||57 cultivars|| ||842||10·20||8·42||1·78||327||386|
|2004||34 cultivars|| ||306||10·27||8·52||1·75||324||382|
|2005||35 cultivars|| ||280||10·52||8·53||1·99||317||381|
|2006||32 cultivars|| ||256|| 9·76||8·18||1·58||339||396|
|2004–06||Claire||6|| 25|| 9·93||8·15||1·78||332||385|
|2004–06||Consort||4|| 25|| 9·71||7·24||2·47||333||427|
|2004–06||Hereward||6|| 25|| 9·18||7·91||1·27||350||395|
|2004–06||Malacca||5|| 25|| 9·52||7·66||1·86||339||406|
|2004–06||Nijinsky||5|| 25|| 9·94||8·16||1·79||326||384|
|2004–06||Soissons||5|| 25|| 9·11||7·69||1·30||352||405|
There were large differences between cultivars in the amount of yield lost to disease (Table 2; P < 0·001). Seventeen cultivars were common across the 3 years of data, of which cv. Consort experienced the greatest disease-induced yield loss of 2·47 t ha−1 and cv. Hereward experienced the smallest yield loss of 1·27 t ha−1 (Table 2). The disease-induced yield losses observed for the cultivars were negatively correlated with their septoria leaf blotch resistance ratings (Y = –0·31X + 3·41; R2 = 0·39; P < 0·01). Disease-induced yield loss did not correlate with resistance ratings for powdery mildew, yellow rust, brown rust or eyespot. Despite the large differences in disease resistance and yield, none of the untreated cultivars had fewer GHG emissions than the treated cultivars (Table 2).
Disease significantly reduced straw yield in three out of six experiments from which data were obtained (Table 3). The greatest reduction in straw yield was observed in the experiment at ADAS Rosemaund in 2006, in which disease reduced straw yield from 8·10 to 6·86 t ha−1 (P < 0·01). Across the six experiments disease reduced straw yield from 6·73 to 6·14 t ha−1. Accounting for the reduction in straw yield made little difference to the GHG emissions associated with grain production. This was because only a small proportion of the GHGs per hectare were allocated to the straw, because of its low economic value compared with grain. However, disease significantly increased the GHGs associated with the production of baled straw. In the Rosemaund 2006 experiment, disease was estimated to increase GHGs from 87 to 105 kg CO2 eq. t−1 straw.
The effect of disease on grain protein was not quantified in the experiments reported here. It was calculated that increasing grain protein content from the default value of 11·5 to 12·5% increased GHG emissions per tonne of bioethanol from 1477 to 1496 kg CO2 eq. This small effect was the result of the lower bioethanol yield from grain with a higher protein content.
- Top of page
The experiments reported in this study, which encompassed a range of seasons, geographical locations and cultivars representative of UK wheat production, showed that disease reduced UK wheat yields by 1·78 t ha−1 (17·5%) across all cultivars. The amount of yield lost was negatively correlated with cultivar resistance rating to septoria leaf blotch. Each unit increase in resistance rating (1–9 scale) to this pathogen reduced disease-induced yield loss by 0·31 t ha−1. Some cultivars deviated from this relationship, probably because of differences in resistance to other diseases or differences in tolerance to septoria leaf blotch (Parker et al., 2004; Foulkes et al., 2006). Nonetheless these results illustrate the importance of resistance to this fungal pathogen in untreated situations.
The average untreated yield loss of 17·5% probably underestimated the true level of disease-induced yield loss because fungicides do not provide complete disease control. National disease surveys carried out on UK farms between 1989 and 1998 estimated that yield losses in fungicide-treated crops ranged from 1·89 to 6·15% and averaged 3·56% (Hardwick et al., 2001). Disease-induced yield losses between ‘disease-free’ yield and the uncontrolled yield were therefore estimated to average about 21% across cultivars.
GHG emissions associated with untreated yields were estimated at 386 kg CO2 eq. t−1 grain, decreasing to 327 kg CO2 eq. for treated yields and to 313 kg CO2 eq. if crops were disease-free. If the septoria leaf blotch resistance of all cultivars could be increased by one point then the GHG emissions associated with untreated yields would decrease by approximately 13 kg CO2 eq. to 373 kg CO2 eq. t−1 grain.
These estimates for GHG emissions probably underestimated those associated with commercial wheat production because yields within the national cultivar testing trials were generally about 2 t ha−1 greater than those achieved on-farm. The relative differences between the different scenarios were probably broadly representative of farm situations. The cultivars studied were the same as those grown commercially and the level of fungicide use in farm crops should result in a similar level of disease control, except possibly under high disease pressure.
If the disease-free yield could be achieved then about 4·70 Mt CO2 eq. would be associated with the production of the typical UK wheat output of 15 Mt. Current use of fungicides on farm is estimated to result in 4·91 Mt CO2 eq. associated with the production of the typical UK wheat output. If fungicides were not used then this would increase to 5·84 Mt CO2 eq. with current cultivars, or 5·59 Mt CO2 eq. resistance to septoria leaf blotch could be increased by one point (on a 1–9 scale). Therefore, controlling wheat diseases has the potential to save up to 1·14 Mt CO2 eq. per annum. For comparison, the total GHG emissions associated with all UK agriculture were estimated at 44·9 Mt CO2 eq. (Baggott et al., 2007).
In the UK, 2123 t fungicide a.i. are applied per annum, using on average 2·9 spray applications per hectare (Garthwaite et al., 2005). This amounts to 0·058 Mt CO2 eq., assuming 3·9 kg CO2 eq. is associated with the production of each kg a.i. (Lal, 2004) and 8·8 kg CO2 eq. is associated with the field operations for each application (Williams et al., 2006). If fungicides were not used then an additional 0·93 Mt CO2 eq. would be emitted to produce 15 Mt wheat. This estimate assumes that the additional wheat area required would displace other annual crops or temporary set-aside/fallow land in the UK, which would result in little change in GHG emissions. However, if additional wheat area, or additional area of other crops to substitute for wheat grain were to displace grassland or forest land, then additional GHG emissions would occur primarily as a result of changes in carbon storage in the soil and vegetation.
It has been estimated that converting grassland or forest into annual crop land releases 1·7 to 29·2 t CO2 eq. ha−1 per year over a 25-year period (carbon reporting within RTFO). Replacing forest in equatorial regions is estimated to release the most GHGs. For illustration, if half of the additional wheat land resulting from the non-use of fungicides displaced UK set-aside or arable crops and the other half displaced UK grassland, then the additional estimated GHGs associated with producing 15 Mt wheat would rise to 1·92 Mt CO2 eq. If all of the additional wheat requirement was met by crops which displaced equatorial forest then the GHG cost would rise to 6·49 Mt CO2 eq. The latter calculation assumes crops with similar yields in the UK and equatorial forest regions. It is clear that the estimate is sensitive to assumptions about demand for grain and how it might be met. If part of the current demand for UK grain supply to global markets could be met by alternative production systems, increased productivity or crop species with lower GHG emissions per unit of production, then the net impact of uncontrolled disease would be ameliorated. Nonetheless this analysis indicates that the use of fungicides provides a significant net saving in GHG emissions associated with wheat production. This advantage must be balanced against any negative environmental impacts which may be associated with their use.
A further reduction in emissions would occur if diseases could be controlled below the levels found in commercial crops currently. Improved control could be achieved through cultivars with more effective or durable disease resistance, or by improved fungicide treatment. The GHG implications of these two approaches differ. The net benefit from improved disease resistance would be sensitive to any associated yield penalty. ‘Yield drag’ may occur directly, through the physiological costs to the host of resistance responses, indirectly, through deleterious genes closely linked with resistance loci, or by diversion of finite plant breeding resources from selection for yield traits. In organic or resource-poor farming systems where fungicide treatment is not available, the benefits to yield of improved resistance are likely to outweigh any yield drag and, therefore, provide a substantial net benefit. In systems where fungicide treatment is widely used, the GHG benefits of improved disease resistance could potentially accrue by reducing the amount of fungicide required to achieve a given level of disease control, or by reducing the amount of disease remaining after a given level of fungicide input. However, the GHG benefits of either may be outweighed by even a small yield penalty associated with disease resistance.
With currently available fungicide active substances, the dose which would minimize GHG emissions per tonne of grain substantially exceeds the dose required to minimize the economic unit costs of production (because the GHG cost of fungicides is very small in comparison to the yield benefit, whereas their economic cost is significant). However, an increase in fungicide use, besides being economically deleterious, would be limited by regulatory constraints on the maximum total dose which can be applied. An increase in the number of fungicide applications to each crop would increase the risk of fungicide resistance (Brent & Hollomon, 2007). New, more effective, active substances could provide a net benefit to emissions, provided the GHG costs associated with their manufacture were not substantially higher than those for existing products. In practice, the crucial aim must be to maintain effective disease control, despite evolution in pathogen populations towards virulence and fungicide insensitivity, by integrating disease resistance and fungicide treatment.
The amount of atmospheric CO2 incorporated into crop biomass (grain and straw) ranges from about 22 to 33 t CO2 ha−1, or 2·7 to 2·9 t CO2 t−1 grain (Table 3), assuming that carbon makes up 44% of plant tissue. Approximately half of the carbon fixed is in the straw. In this work it was assumed that the amount of carbon fixed by the crop has a negligible net effect on atmospheric CO2, because the carbon is re-emitted when the plant product is consumed, burned or decomposes. This is certainly true over a long time-scale. Over shorter time-scales of a few years it is possible that straw incorporated into the soil could lock up significant amounts of carbon (King et al., 2004).
This paper estimates GHG emissions associated with the production of wheat using fungicides ranging from 267 to 388 kg CO2 eq. t−1 grain. This compares with estimates of 220 kg CO2 eq. t−1 (Mortimer et al., 2004) and 800 kg CO2 eq. t−1 (Williams et al., 2006). The estimates of Mortimer et al. (2004) were lower because of a lower estimate of N2O emissions (Mortimer et al., 2003). Williams et al. (2006) estimated N2O emissions using the method of the Intergovernmental Panel on Climate Change (1997). This method predicts N2O emissions that are more than 50% greater than the method used in this present study (De Klein et al., 2006). Williams et al. (2006) also estimated GHG emissions for bread wheat, which has a greater N fertilizer requirement and lower yield than the mixture of feed, biscuit and bread wheats described in this study. It is estimated that the GHG emissions associated with bread wheat are about 25% greater than for feed and biscuit wheats. Nonetheless, it is clear that there is considerable uncertainty when estimating N2O emissions (Kindred et al., 2007). For example, it was reported recently that the method described in De Klein et al. (2006) may significantly underestimate N2O emissions (Crutzen et al., 2007). As the study reported here used conservative estimates of the amount of GHGs associated with N fertilizer use, the impact of disease on GHG emissions may have been underestimated.
In this study it was assumed that nitrogen fertilizer use is unaffected by disease. However, it is possible that disease control may affect the economic optimum for N. For example, less nitrogen fertilizer may be required for a following crop if disease reduces the amount of nitrogen removed in the grain and nitrate leaching losses over winter are low. There are also likely to be effects of N supply on the severity of biotrophic diseases (e.g. Neumann et al., 2004). Further work should investigate whether these interactions with nitrogen have a significant impact on the GHG implications of disease.
In the absence of disease control in the UK, the GHG emissions associated with each tonne of animal feed and wholemeal flour are estimated to increase by the same amount as for whole grain (59 kg CO2 eq.). This is because the processing yield of these products from grain is close to 100%. Disease will result in a greater increase in GHGs per tonne for products with processing yields of less than 100% and for which the by-product has a lower economic value than the primary product (e.g. bioethanol or white flour). For example, the GHG emissions per tonne of bioethanol were estimated to increase from 1235 to 1421 kg CO2 eq. t−1 when disease was not controlled, equating to 46·3–53·2 kg CO2 eq. GJ−1 ethanol, respectively. Compared to fossil-derived petrol, which gives emissions of 85 kg CO2 eq. GJ−1 (carbon reporting within RTFO), bioethanol from wheat in this study reduced GHG emissions by 46% for each MJ of energy produced. If disease is not controlled then the GHG savings from bioethanol decline to 37%.
Disease may have a small effect on the yield of bioethanol per tonne of grain via effects on grain protein content (Smith et al., 2006): both rusts and powdery mildew have generally been observed to decrease grain protein content (Dimmock & Gooding, 2002). For example, reducing leaf rust infection using fungicides increased grain protein from 12·9 to 13·5% dry matter (Clare et al., 1990) whereas leaf and glume blotch diseases generally either have no effect or increase grain protein content (Dimmock & Gooding, 2002). For example, disease control using fungicides increased grain protein content by an average of 0·66% (Clark, 1993). However, altering protein content by 0·6% was estimated here to change GHG emissions by just 1%, so disease-induced changes to protein appear to be of relatively little importance.
This study suggests that disease control substantially reduces GHG emissions per tonne of wheat grain, animal feed, flour or bioethanol produced. Similar analytical methods could be applied to assess the GHG cost of disease in other climatic zones and crop species which have different diseases, agronomic inputs and yields.