The crops. This part describes, species by species, the results

Transcription

1 C The crops

2 The crops C This part describes, species by species, the results obtained for a group (not exhaustive) of cultivated species representative of the diversity of French agriculture. The species studied were chosen according to the availability of corres ponding crop models. C1 - Advantages and vulnerabilities C2 - Wheat C3 - Maize & Sorghum C4 - Grassland C5 - Oilseed rape C6 - Sunflower C7 - Grape vines This approach, relevant to industrial principles, allows us to go deeper into the characteristics of each species, be they physiological, agronomic or geographical. It is able to analyse, for each species, what effects, if any, climate change will have on the growth and development of these crops and to examine the main possibilities for adapting cultural practices. The ADVANTAGES and VULNERABILI- TIES section introduces and summarises the essence of the results, The ORGANIC FARMING section treats this method of production separately; its present development called for specific work in CLIMATOR. C8 - Forests C9 - Organic farming 140

3 Advantages and vulnerabilities of agricultural crops faced with climate change Frédéric Levrault C1 Advantages & vulnerabilities Through the action of its different components (temperature, radiation, rainfall) and by the effect of the increase in CO 2, climate change* will affect the growth* and development* of cultivated species. The effects on the plants will, depending on the scenario, be favourable to plant production (e.g. increased yields, reduced inputs, easier work, wider choice of crop rotations* etc.) or unfavourable (water stress*, reduced yields*, or an increase in their variability*). Here we propose to examine the mechanisms whereby these advantages and vulnerabilities will be manifested, thus affecting the relevance of the crops concerned. This notion of relevance encompasses agronomic, technical, environmental and organisational aspects. Hence this discussion goes beyond the concept of feasibility*, dealt with elsewhere in this book. Our analysis of the mechanisms of action of climate change is concerned with: temperature regime* and phenology*; high and low temperatures and physiological disorders; the evaporative demand* and irrigation requirements of summer crops; rainfall and available days for fieldwork*; the quality of products*; CO 2, water stress and production levels; the year-to-year variability* of the weather* and yields. We analyse separately these different mechanisms, referring the examination of the interactions to the sections concerned on The crops and The topics. Also, we do not include here the economic aspect of the suitability of crops, believing that assumptions about production costs and sale prices would be too hazardous in view of the time scale of the study (30-80 years). A Temperature regime and phenology The expected increases in temperatures over France (a mean warming of 2 C ± 0,6 C by 2050 for the A1B* emission scenario: cf. CLIMATE section) will increase the heat availability* for plants in general and cultivated species in particular. Thus the heat requirements for the complete development of a variety will, for a given location, be more rapidly and/or more often satisfied. This will have two consequences for farming: new crop possibilities in areas which at present are too cold; in present cropping zones, the advancement of harvest dates of annual crops will increase the length of periods between crops or widen the choice of crop successions. Green Book The crops Advantages and vulnerabilities Frédéric Levrault 141

4 C 1 Advantages & vulnerabilities New cropping opportunities For annual crops, these opportunities will appear in species with high heat requirements, which cannot now be grown in the coldest regions of metropolitan France. The case of maize For example, for the sites* of Mons and Mirecourt, representative of the northern plains of France (fig. 1 top), an early maize variety (Méribel), which at present only reaches maturity 2-7 years out of 10 because of low temperatures, should be able to be grown there in the near future (NF*) with maturity reached in 9 years out of 10. Figure 1: frequency of attainment of maturity of an early maize variety (Méribel) for the sites of Mons (top left), Mirecourt (top right) and Clermont-Theix (bottom). A1B* emission scenario All downscaling methods* All soils. Dotted grey: frequency recorded 8 years out of 10. Upper and lower edges of orange rectangles: minimal and maximal frequencies obtained by simulation. Bold black lines: mean frequency. For an average mountainous site such as Clermont-Theix, even colder than the two previous ones, the same trend appears (fig. 1 bottom), but more gradually, because the crop only becomes possible in the distant future (DF*) (maturity reached 9 years out of 10). The case of sunflower Another crop with high heat requirements, sunflower should also benefit from the increased heat availability, and following the same progression, firstly in the northern plains and then in the foothills. If one studies a late variety (Mélody), the access to these new production zones will take place in the same direction, but more slowly than in the previous case because of the higher temperature requirements for this sunflower variety. 142 Green Book The crops Advantages and vulnerabilities Frédéric Levrault

5 C 1 Advantages & vulnerabilities Figure 2: frequency of attainmment of maturity of a crop of a late variety of sunflower (Mélody) for the sites of Mons (top left), Mirecourt (top right), Clermont-Theix (bottom). A1B* emission scenario. All downscaling methods* all soils. Dotted grey: frequency recorded 8 years out of 10. Upper and lower edges of yellow rectangles: minimal and maximal frequencies obtained by simulation. Bold black lines : mean frequency. The results for sunflower were obtained with two crop models* (STICS and SUNFLO), the only difference being that the SUNFLO model predicted a greater increase in feasibility* in the NF. As they only treat the heat availability on the scale of the crop cycle, to arrive at a real agricultural application one should consider these results in relation to questions of choices of cropping plans, insertion into crop rotations*, and especially the required inputs (especially irrigation water). Green Book The crops Advantages and vulnerabilities Frédéric Levrault 143

6 C 1 Advantages & vulnerabilities The case of vines For viticulture, the increasing temperatures will steadily extend the growing regions to the whole of northern France. Meanwhile it will affect the choice of varieties in the present growing areas. Thus, our results show that at present, at Mons, grapes barely reach wine-making maturity naturally, chardonnay, currently grown in Burgundy and Champagne, only reached maturity in the recent past (RP*) 3 years out of 10 for the three methods of downscaling used. In the NF, Chardonnay will reach maturity 9 years out of 10. In the DF it will mature every year (fig. 3). Figure 3: frequency of attainment of wine-making maturity of grapes (chardonnay for Mons, grenache for Dijon and merlot for Rennes). A1B* emission scenario all soils. All planting densities. All downscaling methods*. Dotted grey: frequency recorded 8 years out of 10. Upper and lower edges of purple rectangles: minimal and maximal frequencies obtained by simulation. Bold black lines: mean frequency. At Dijon, grenache (typical of the south-east) cannot now be grown, which agrees with our simulation results (maturity reached in 3-7 years out of 10 according to the downscaling method used). In the NF, this variety reaches maturity in 9-10 years in 10. In the DF, it does so every year regardless of the downscaling method used (fig. 3). Finally at Rennes, merlot (typical of Bordeaux) cannot be grown at present: our models indicate that in the RP maturity should be reached in 2 years out of 10 at best, according to the downscaling method. In the NF, this variety will reach it in 8-9 years in 10. In the DF, it will mature every year (fig. 3). For the sites at Mons and Rennes, with the varieties used, climate change appears quite sudden, as from the NF the feasibility levels will be much higher than those of the RP. New thermal areas will gradually take shape for each variety. However the areas actually cultivated will be determined by other considerations, involving especially problems of the quality and typical character of the wine. Also, one should bear in mind that these simulations were for vineyards on the plains; the case of vines grown on hillsides with particular microclimates was not considered here. 144 Green Book The crops Advantages and vulnerabilities Frédéric Levrault

7 Accelerated rates of development C The more rapid passing of phenological stages*, due to the increase in temperature, will result in a fairly generalised advancement of harvest dates. This early release of land will affect both monocultures and crop rotations. 1 Advantages & vulnerabilities Monocultures For monocultures, we will witness a prolongation of the period between crops. Thus, in the typical case of the monoculture of soft wheat in the Paris basin (Versailles), we find this prolongation for both varieties (Charger and Soissons) and the three soil depths studied (fig. 4). Depending on the downscaling methods used, we find a prolongation of the period between crops of days between the RP and the NF, of 9-12 days between the NF and DF, and of days between the RP and DF. From now until the end of the century the length of the period between crops will go from slightly less than 75 days to about 100 days. Figure 4: length of the period between crops of a soft wheat monoculture for the Versailles site. Recent past: top left. Near future: top right. Distant future: bottom A1B emission scenario all downscaling methods, WT, QQ et ANO Two varieties three soil types. Left and right edges of the green rectangles : minimal and maximal simulated periods; white bars: means. The lengthening of the period between crops, combined with the stimulation of mineralisation of the soil organic matter by the warming (cf. ORGANIC MATTER section), other things being equal, increases the risk of leaching of nitrates from bare soils. On the other hand the expected reduction in rainfall and drainage (the PERCOL* variable, cf. WATER section), at this time of year will reduce the risk of nitrogen being carried into the sub-soil. The close examination of these two opposing mechanisms remains to be carried out, to find out how the quality of aquifers will be affected (cf. WATER section). If required, know-how already exists for installing trap crops* (phacelia, mustard, etc.) or catch crops* (sunflower, soya etc.), which will be facilitated by the higher temperatures, but subject more than previously to water shortage. Green Book The crops Advantages and vulnerabilities Frédéric Levrault 145

8 C 1 Advantages & vulnerabilities A crop rotation In rotations, the advancement of harvest dates will make possible certain crop successions in regions where they are currently impossible because the harvest of the previous crop is too late to allow timely sowing of the following crop. Thus at Colmar, in a maize-wheat-rape rotation (fig. 5) an early maize variety, Méribel, (preceding soft wheat) reached maturity in the RP 0-2 years out of 10 depending on the downscaling method, whereas in monoculture this same maize variety reaches maturity in 8 years out of 10 for the same period. The difference is explained by the need to harvest the maize crop earlier in the rotation, which does not allow it to reach the required stage for harvest naturally (farmers have to use energy-intensive drying techniques). In the NF, early maize in rotation will reach maturity in 8-10 years out of 10, whilst it will do so every year in monoculture. In the DF, early maize will mature every year, both in monoculture and rotation, and whichever downscaling method is used. One can conclude that because of the advancement of the maize life cycle, it will be harvested in good conditions, both for monoculture and rotations, without the need for artificial drying. Figure 5: frequency of reaching natural maturity for a maize crop at Colmar in the hard wheat-maize-soft wheat rotation for the three downscaling method: ANO, WT and QQ A1B emission scenario. 146 Green Book The crops Advantages and vulnerabilities Frédéric Levrault

9 B High and low temperatures and physiological disorders C Climate change will manifest itself, among other ways, by more frequent hot days and rarer cold days. These high or low temperatures have particular effects on the behaviour of cultivated plants and consequences for their yields. More hot days High temperatures, if they occur during grain filling, have physiological effects (competition between photosynthesis, photorespiration, increased night respiration, stomatal closure) called heat stress*, which reduces yield. This phenomenon is most severe for wheat and sunflowers (cf. TIMING section). For small grain cereals, it is thought that these mechanisms, which reduce yield, begin to take effect above 25 C. Yet our results (fig. 6) show that in future the maximum daily temperatures will exceed this threshold more frequently during the months of April-June. 1 Advantages & vulnerabilities Figure 6: changes in the number of heat stress days (daily tmax >25 C) between April and June at Avignon (top left), Toulouse (top right) and Versailles (bottom) during the 21 st century. A1B emission scenario: QQ downscaling method. Thus at Avignon, the number of heat stress days (i.e. the number of days when heat stress occurs) between April and June fluctuates from 8 to 53 in the RP, from 15 to 55 in the NF and from 31 to 69 in the DF. At Toulouse, this number varies between 2 and 29 in the RP, between 2 and 43 in the NF, and between 16 and 60 in the DF. At Versailles, it lies between 0 and 18 in the RP, between 1 and 23 in the NF and between 4 and 41 in the DF. The results obtained for these three sites using other downscaling methods (ANO and WT) were very similar. We should bear in mind that advancement or acceleration of phenology will result in partial avoidance* of these additional heat stress days, thus partially alleviating their harmful effects on yield (cf. TIMING section). Two ways of adaptation* thus appear: bringing forward sowing dates, and varietal improvement aimed at shortening life cycles and tolerance of high temperatures. Green Book The crops Advantages and vulnerabilities Frédéric Levrault 147

10 C 1 Advantages & vulnerabilities Fewer autumn frosts In the coldest French regions certain crop rotations currently suffer from frost damage on young winter crops. This is typically the case with rape grown in rotation, which is sensitive to autumn frost in certain conditions (before the 8-leaf stage or with excessive leaf area due to too much nitrogen). This frost sensitivity explains the low feasibility which our simulations show for the NF at Colmar for the two varieties studied (Olphi et Pollen): 1-2 years in 10. In the future these limitations due to autumn frost will become rarer. Consequently, in the NF, rape will reach maturity in 2-7 years out of 10 at Colmar. In the DF, maturity will be reached in 9-10 years in 10 (fig. 7). This diminution of autumn frosts thus constitutes a positive aspect of climate change. Figure 7: feasibility of growing a rape crop in a hard wheat-maize-soft wheat-rape rotation at Colmar using three downscaling methods: ANO, WT, QQ. A1B emission scenario. C Evaporative demand and irrigation requirements of spring-sown crops The marked reduction in spring and summer rainfall (cf. CLIMATE section) and the increase in potential evapotranspiration* will increase the irrigation needs of spring crops, whilst the advancement (accelerated phenology) will moderate this because of the avoidance effect. A typical example of this is that of irrigated maize. At a site with a soil of medium depth and with irrigation limited to 80% of the ETM*, our simulations show (fig. 8) a clear increase in irrigation between the RP and the NF: mm, depending on the variety and downscaling method, at Toulouse, mm at Lusignan, mm at Versailles, mm at Colmar, and mm at Mons. In the field this represents a range of one to four extra applications of water 148 Green Book The crops Advantages and vulnerabilities Frédéric Levrault

11 C 1 Advantages & vulnerabilities Figure 8: changes in irrigation requirements to 0.8 ETM of a maize crop during the 21 st century. A1B emission scenario- Mean, minimum and maximum of three downscaling methods mean for soil 1 fixed sowing dates. Late variety: Toulouse and Lusignan; early variety: Versailles, Colmar and Mons. This result is counterbalanced by the fact that our simulations (which include the advancement mechanism) are made for fixed sowing dates (constant throughout the period ). Yet in real agriculture, sowing dates have already been brought forward markedly, and will continue to be so throughout the 21 st century. It is as well however to remain cautious about a possible moderating effect (by avoidance) of the advancement of sowing dates on irrigation requirements, as our early results do not suggest this (cf. MAIZE-SORGHUM section). This adaptation to climate change through the advancement of sowing dates needs to be studied in more depth for a more complete picture of the trends in irrigation requirements. Incidentally all methods of adaptation need to be examined (sustainable water storage, irrigation control via a low ETR/ETM, reduction in the areas irrigated and substitution of species, earlier varieties, genetic improvement of the tolerance* of water stress) both within and between production zones, including the transfer of crops towards zones likely to suffer less in terms of water supply. Meanwhile the question arises of the change in natural water resources (surface and groundwater) in response to climate change. Recent work (the REXHYSS and IMAGINE 2030 projects) indicate more severe and prolonged low-water situations later in the year, which agrees with our own results on the field scale (cf. WATER section). These results should be followed through to find out what water abstraction possibilities exist for agriculture, both in spring and summer for irrigation, or in winter to replenish reservoirs intended purely or partly for irrigation. D Precipitation and days available for fieldwork The fairly general reduction in rainfall will affect the surface condition of the cultivated soils. To be precise, we should see an increase in the number of days (called the number of available days) when soil moisture is low enough to allow machinery to be used on the fields. Our simulations show in particular that in autumn (for preparations for sowing winter crops) this trend is particularly clear for the northern half of France. In fact, this number of available days is currently smaller in the north than in the south because of the higher rainfall in the north (fig. 9). Green Book The crops Advantages and vulnerabilities Frédéric Levrault 149

12 C 1 Advantages & vulnerabilities Rennes Versailles Mons Mirecourt Dijon Colmar Bordeaux Lusignan Clermont St-Étienne Available days (%): 0-50% 70-80% 50-70% 80-90% Toulouse % Avignon Figure 9: number of available days in autumn in the RP (circles, left), in the NF (circles, centre) and in the DF (circles, right). A1B emission scenario WT*: downscaling method. Same soil. Hence it is the facilitation of autumn work (harvesting maize, preparations for sowing winter crops) which will be affected by climate change, making it easier to manage the crops concerned. In other cropping situations, the number of available days will decline. This is particularly so for sowing spring crops: the lower rainfall expected in the south of France will reduce the number of suitable days for sowing, the lack of soil moisture preventing germination and satisfactory plant emergence. The use of starter irrigation is therefore likely to become widespread. E Quality of products The increased temperatures and advancement of phenology will for perennial crops have particular effects on the quality of products. Hence vines, whose harvest at present takes place after the annual temperature maximum, will mature earlier in the summer. This means that the berries will be exposed to higher temperatures, especially at night, which is known to be important for flavour. As shown in figure 10, all the present vineyards will experience this increase in night temperatures and will risk a decline in quality of the grapes unless some form of adaptation is introduced (cf. VINES section). Figure 10: change from 1950 to 2100 of the mean minimum daily temperatures during grape maturation (veraison to maturity) A1B emission scenario WT downscaling method. 150 Green Book The crops Advantages and vulnerabilities Frédéric Levrault

13 F CO 2, water stress and yield C The increase in atmospheric CO 2, the cause of climate change, stimulates the photosynthetic activity of C3* plants (in our study, maize and sorghum are the only C4 plants) which constitute most of the cultivated plants of temperate regions (the mechanism known as carbon* fertilisation), and hence the production of biomass. Conversely the reduction in the water sufficiency*, also a consequence of climate change, hinders crop production. The combination of these two factors will determine the trend upwards or downwards of yields, with results which will vary according to the crop, the locality, the period simulated (NF or DF) and the cropping schedules. These mechanisms, briefly alluded to here, are described in more detail in the YIELD section. Increased yields in the absence of increased water stress In the case of winter crops, less exposed to increased water stress, we should witness an increase in yields. This is typically what our simulations show for soft wheat monoculture. For example on a soil of medium depth and depending on the downscaling method used, the mean yield for a period of thirty years will increase in the NF by t/ha at Lusignan, at Bordeaux by t/ha, and at Rennes by t/ha. At Clermont-Theix the increase is even larger ( t/ha) due to the irregular maturity achieved in the RP of only 5-8 years out of 10. Carbon fertilisation and the increased temperatures are clearly the two main factors responsible for this yield increase. The advancement of developmental stages, a consequence of the temperature increase, also contributes to these increases, but indirectly: it allows the crop to avoid the increased water and heat stresses in spring and summer. Lastly, this yield increase is also partly explained by the reduction in excess winter rainfall, particularly in the west of the country. For grasslands the same trend is found. Very much associated with the downscaling method used, the yield increase between now and the end of the century will fluctuate between 8 and 22% for fescue and between 6 and 19% for perennial ryegrass. This increase in annual yield will result from a big increase in the daily growth of the plants in autumn and spring (due to increased temperature and CO 2 without any extra water stress) combined with a reduction in daily plant growth in summer due to increased water stress. A fall in yield with increased water stress Conversely, when the water stress becomes really severe, its harmful effects will cancel out the positive effects of carbon fertilisation and yields will fall. This is what we should find at certain locations for sunflowers and vines. Thus at Toulouse the mean yield of rainfed sunflower (in a sunflower-wheat-sorghum-wheat rotation) on a shallow soil will fall, depending on the downscaling method, by t/ha between the RP and the DF. This fall however will be localised: at Bordeaux and Lusignan, whose climate trend is similar to that of Toulouse, there is no sign of a fall. At the more northerly sites (the Paris basin for example) our results for the RP (maturity reached in 0-7 years in 10) prevent us from drawing conclusions. However one can expect that the lower water stress would allow the positive effects of CO 2 to cancel it out, thus resulting in an increase in yields. Vines, if not irrigated, could also be subject to a fall in yield in certain circumstances. Hence at Avignon, the yield of grenache on a soil of medium depth should vary by to 0.5t/ha in the NF, depending on the plant population and the downscaling method used. The same analysis for merlot at Bordeaux gives a variation between and 0.2t/ha. At Dijon, the yield of chardonnay should vary by to 0.4t/ha in the same study conditions. In the case of vines, the fall in yield is also due to the reduction in the duration of formation and filling of the grape berries. To sum up, the balance between favourable and unfavourable factors should only develop in a way which reduces yield in a limited number of situations: the south of France, perennial or annual rainfed crops, and annual crops sown in spring or summer. It is premature at this stage to define a frontier between regions which will or will not experience falling yields. 1 Advantages & vulnerabilities Green Book The crops Advantages and vulnerabilities Frédéric Levrault 151

14 C 1 Advantages & vulnerabilities G Year-to-year weather variability and yields An important determinant of the presence of a species in a cropping system* apart from its average yield and profitability is the long-term stability of yields under the local soil and climatic conditions. This large or small year-to-year yield variation can be quantified by means of the coefficient of variation, which is the ratio of the standard deviation* to the mean (of the yields) for a given period. Climate change should give rise to an increase in this variability (due to bigger yield differences from year to year), at least for certain crops. For rainfed sunflower for example (tab. 1), our results obtained for four sites show a clear trend to increasing yield variability. Over the whole century, the variability increases by 25% in the south (Toulouse), by 10% in the centre (Lusignan and Dijon) and remains constant in the north (Versailles). The present year-to-year yield variations, already increasing from the north to the south, seem certain to get worse. It is often this increased variability which makes the yield trends statistically non-significant. Site Toulouse Dijon Lusignan Versailles Period RP NF DF RP NF DF RP NF DF RP NF DF Mean yield (quintals [=100kg] per hectare) Coefficient of variation (%) Table 1: yield of rainfed sunflower (early variety Prodisol) in monoculture. Left-hand column: mean yield per period. Right-hand column: coefficient of variation. A1B* emission scenario WT downscaling method Soil 1 mean This increase in variability is not observed for all the crops. In vines, for example (tab. 2), our work shows that the year-to-year yield variability remains largely unchanged throughout the 21 st century. Site Toulouse (merlot) Dijon (chardonnay) Bordeaux (merlot) Avignon (grenache) Period RP NF DF RP NF DF RP NF DF RP NF DF Mean yield (quintals [=100kg] per hectare) Coefficient of variation (%) Table 2: yield of vines (variety depends on the site). Left-hand column: mean yield per period. Right-hand column: coefficient of variation. A1B emission scenario WT downscaling method Soil 1 for all Green Book The crops Advantages and vulnerabilities Frédéric Levrault

15 C Advantages & vulnerabilities Although the subject needs further work, one can nevertheless suggest that the crops most affected by increased yield variability will probably be the unirrigated spring-sown crops, especially those grown on shallow soils. Perennial and winter crops, and to a lesser extent grasslands, appear less subject to these year-to-year yield differences, for two main reasons: winter weather problems (very low temperatures or excess water) will become less frequent because of the warming and drying of the climate; their calendar timing protects them from the increased spring and summer droughts. 1 What you need to remember 3 Climate change, as predicted by the climate models, will in general cause neither a deterioration nor an improvement in farming possibilities. We should rather expect that a whole set of factors on which these possibilities depend should be modified. 3 The main favourable changes or advantages are: opportunities for new crops resulting from the higher temperatures, notably summer crops in the north of France and in the foothills; an acceleration of rates of development which will allow a partial escape from the increased water stresses and the increasing number of heat stress days in spring and summer; a lower soil moisture content in autumn which will lead to more days available for autumn work; fewer problems of autumn frosts with winter crops; an increase in yields when water stresses are avoided or made up for by growth in stress-free periods: winter crops, grassland and perennial crops. 3 The main unfavourable changes, or vulnerabilities, are: an increase in the duration of the period between crops in monoculture, which increases the risks of leaching and erosion; an increase in the number of heat stress days in spring, whose effect will be partially reduced by the advancement of cropping schedules; a reduction in yield in cases where aggravated water stress is not avoided: rainfed vines and sunflower; an increase in the water requirements of summer crops; an increase in the year-to-year variability of summer rainfed crops, especially sunflower; a risk of falling grape quality due to the advancement of the maturation period. Green Book The crops Advantages and vulnerabilities Frédéric Levrault 153

16 C 1 Advantages & vulnerabilities What needs further study 3 Can irrigated crop yields be maintained at their present level by combining a modest reduction in irrigation with the positive effects of carbon fertilisation and higher temperatures? 3 Will the shortening of growth cycles allow the introduction of extra catch crops? 3 What is the real impact of the increase in the number of heat stress days in spring? 3 What is the impact of the prolongation of the periods between crops on the quality of the water in aquifers? 3 To what extent will the shortage of spring rainfall affect the germination and emergence of spring-sown crops? To find out more Bates B.-C., Kundzewicz Z.-W., Wu S., Palutikof J.-P., éd, Le changement climatique et l eau, document technique publié par le GIEC. Secrétariat du GIEC, Genève, 236 p. Ciscar J.-C., Iglesias A., Feyen L., Goodess C.-M., Szabo L., Cristensen O.-B., Nicholls R., Amelung B., Watkiss P., Bosello F., Dankers R., Garrote L., Hunt A., Horrocks L., Moneo M., Moreno A., Pye S., Quiroga S., Van Regemorter D., Richards J., Roson R., Sonia A., Climate change impacts in Europe. Final report of the PESETA research project. EUROPEAN COMMISSION - JOINT RESEARCH CENTER. Chapter 3 : Agriculture impact assessment. Pages 43 à 49. European Environment Agency, Impacts of Europe s changing climate indicator-based assessment. Chapter 5.9 Agriculture and forestry. EEA report No 4/2008. Pages 135 à 148. Commission des Communautés Européennes, L adaptation au changement climatique : le défi pour l agriculture et les zones rurales européennes. Document accompagnant Le Livre Blanc sur l adaptation au changement climatique,13 pages. 154 Green Book The crops Advantages and vulnerabilities Frédéric Levrault

17 Climate change and the wheat crop: the main impacts Marie-Odile Bancal, Philippe Gate 2 CWheat A Some key elements for the soft wheat crop in France 3 West 3 Center-North 0 Wheat yield (t/ha) North-East Center-East 3 Wheat area (ha) South-West 94 South-East Figure 1: yields and areas sown with wheat in the CLIMATOR regions, with the proportion of durum wheat in red, and mean yields*. Means Source Agreste. Occupying about 4.8 million hectares (17% of the UAA), soft wheat is the highest-ranking of French arable crops. Half the area is situated in the centre-north (fig. 1). The mean yields are about 7 t ha -1 with a maximum of 7.5 t ha -1 in the north-east and a minimum of 3.7 t ha -1 in the southeastern region where it is usually replaced by durum wheat (cf. And durum wheat? ). Soft wheat is grown for its starch, and an improvement in its protein content is often sought for human and animal feeding. The diversity of industrial outlets with a primary processing (semolina and starch manufacture, milling) and often a secondary one (biscuits and bread) involves precise specifications to reach optimal quality at harvest which are specific for each industrial process. Wheat is a major crop in arable rotations. Its long life cycle, from October to July, its wide range of sowing dates and it relative tolerance to degraded environments make it an easy crop to grow, exploiting a wide range of environments; its sensitivity to heat and water stress* however limit its expansion into the southern zones. As to diseases, wheat is subject to numerous fungal diseases, both soil-borne (eyespot, take-all) and foliar (mainly Septoria leaf spot, rusts and Fusarium foot rot), which frequently affect yield and the technological, hygienic and nutritional quality of the harvest. It is also sensitive to waterlogging and has high nitrogen requirements. Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate 155

18 C 2 Wheat B Simulation protocol of CLIMATOR The cultivation of soft wheat has been studied in monoculture with three crop models*, CERES, STICS and PANORAMIX, and several cropping systems* with the STICS model, including a basic irrigated cereal system (MWRW) and two low-input systems (SWSgW) and ORG). Durum wheat was only studied in STICS simulations as part of the MWRW and SWSgW successions. Sites System Management Soils Variety 12 sites in mainland France including one at high altitude Monoculture (all models) Three successions (STICS): maize-soft wheat-rape-durum wheat (MWRW) sunflower-durum wheat-sorghum-soft wheat (SWSgW) peas-soft wheat-2 years of forage grass (ORG) Recommended management, sowing 10 Oct, 230 kg N ha -1 in 3 applications Sowing dates: 20 Sept, 1 Oct, 10 Oct, 20 Oct, 10 Nov, 1 Dec Irrigation for the MWRW system to 50% of water requirements (if necessary) Soil 1: deep silt soil with a high available water reserve (AWR) (226 mm) and average fertility Soil 2: rendzina with a low AWR (104 mm) and low fertility Soil 3: leached hydromorphic soil with a high AWR (218 mm) and high fertility Soissons (early) and Arminda (late) which differ in their earliness at the 1cm ear stage and duration of stem elongation Model CERES, PANORAMIX, STICS Climate SRES: A1B, ARPÈGE, downscaling: WT, QQ and ANO C Ecophysiological identity card of wheat in the climate change context Advantages Like all C3 plants, soft winter wheat and durum wheat are good for exploiting the increase in CO 2 concentration, with both a stimulation of photosynthesis* and a limitation of transpiration. The calendar timing of its developmental cycle is generally advantageous for avoiding summer water stress*, and this could be further enhanced by suitable varietal choice. Also, soft wheat has high vernalisation* requirements: to complete its floral development*; it requires a certain number of days when the mean temperature does not exceed 10 C (about days according to the varieties and the climate*). These vernalisation requirements depend on the variety and are a component of their earliness. Moreover, to flower, a vernalised wheat must be subjected to a long photoperiod, which gives a certain stability to the duration of the emergence-flowering period in relation to regional temperature regimes and climate change*. Thus the shortening of the growth cycle caused by climatic warming will be limited by these two requirements (vernalisation and long days) of soft wheat. The beginning of stem elongation (or the 1cm ear stage) is a critical stage as regards low temperatures at the moment when the apex passes from the buffered temperature of the soil to that of the air, and from a vegetative physiology to a reproductive one: the temperature threshold 156 Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate

19 C 2Wheat of 4.5 C for soft wheat is usually lethal for main culms during stem elongation (earlier, during the vegetative phase, one can assume a temperature of 25 C for the freezing threshold of the plants). This ear frost damage mainly affects the main stems which contribute about 50% to the final yield. Consequently, this occurrence of early freezing of the ears is regarded as a major factor in the feasibility* of growing the crop, even though tillering can partially compensate for the losses caused. If, as seems likely, these ear frosts should become rarer with climate change, it is important to reconsider the timing of the sensitive stages in relation to the occurrences of low temperatures. Still in relation to winter physiology, the reduction of waterlogging may also prove to be favourable in certain regions. Weaknesses Although soft winter wheat is less sensitive to summer water stress than spring-sown wheat, it is on the other hand very sensitive to heat stress towards the end of growth, which can occur above 28 C: above this threshold, there is an irreversible cessation of the growth of the grains and sometimes even abortion of grains beginning to fill. Also, above 25 C there may be a reduction in grain growth. More generally, one must consider the timing of the critical stages (stem elongation, flowering, grain filling) in relation to the occurrence of high temperatures and/or water stress, because it is these stresses brought about by climate change which could hinder the formation of wheat yield in the future. Note also that its vernalisation requirements make soft wheat sensitive to an increase in winter temperatures which, above a certain threshold, could become critical for its flowering. D Impacts of climate change on wheat growing Impacts on yield and its components The maps (fig. 2) show the geographical distribution of yield trends and their variability* for an early variety (Soissons) and a late variety (Arminda) of soft wheat, simulated with CERES. Compared with the RP* we see an increase in the yield of the variety Soissons (a mean increase of about 0.7 ± 0.5 t/ha for the NF* and of 1.2 ± 0.8 t/ha for the DF*), mainly in the east of the country. This increase in mean yield is nearly always accompanied by an increase in the variability of yields, indicated by the shade of grey of the circles in figure 2. Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate 157

20 C 2 Wheat Except for certain sites* (Bordeaux, Clermont and Saint-Étienne), the variety Arminda has a lower yield than the early variety (96% of its yield on average for the RP). It also shows a less clear trend towards an increase in yields: of about 2% and 6% for the NF and DF respectively, as against 9% and 14% for the early variety. Hence this all supports the view that for the late variety, the benefit of the increased CO 2 is counterbalanced by the larger effect of stresses. These results, obtained with the CERES crop model, agree in general with those of PANORAMIX. One should note however one difference with the STICS model which, by taking account of waterlogging, simulates a substantial increase in yield at all the sites towards the west (Rennes, Lusignan, Bordeaux). Standard deviation (t MS/ha) : < >3 Rennes Number: % Feasability Bordeaux Versailles Lusignan Mons Mirecourt Dijon Clermont St Étienne Colmar Standard deviation (t MS/ha) : < >3 Rennes Versailles Number: % Feasability Lusignan Bordeaux Mons Mirecourt Dijon Clermont St Étienne Colmar Toulouse Avignon Toulouse Avignon Yield( t/ha) Yield( t/ha) Figure 2: changes in yield (means and variability) and in feasibility (percentage given below the average yield persite) of the soft wheat crop for the early variety Soissons (A) and the late variety Arminda (B). CERES model, soil 1, WT climate. Lastly, figure 2 shows an increase in the feasibility of the crop with climate change at all the sites, due to the decline in ear frost risk. This trend is very clear in the east for both varieties. The feasibility, calculated in this way, is very high (>80%) at all the sites except for the high-altitude site of Clermont. This site is also distinguished by the fact that the early variety s feasibility falls from 63% to 50%, suggesting that the developmental shift due to climate change results in a more unfavourable timing of the frost-sensitive stage, whereas for the late variety the feasibility remains the same at about 75%. Due to winter waterlogging, the STICS model simulates low yields at the western sites, which often makes wheat growing unprofitable, even though physiologically feasible, in the NF. Climate change tends to remove this constraint. Table 1 gives yield projections* for the NF and the DF, site by site, with their level of significance. On average, we see a significant increase in yield of about 0.9 to 1t/ha in the NF and DF respectively, due to the effect of increasing CO Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate

21 C 2Wheat NF-RP DF-RP Avignon ns ns Bordeaux ns 0.9 Clermont ns ns Colmar Dijon ns 1.5 Lusignan ns ns Mirecourt Mons Rennes ns ns Saint-Étienne Toulouse Versailles All sites Table 1: changes in wheat yields simulated by CERES (soil 1, climate:wt). The significance of the changes in comparison with the year-to-year variability is denoted thus: Bold: p < 0.01; Italics: p < 0.05; Normal: p < 0.10; ns: non-significant. However the response is very variable between sites. Thus, no significant change is predicted for half the sites in the near future (Avignon, Bordeaux, Dijon, Clermont, Lusignan and Rennes) and for a third of the sites in the distant future (Avignon, Clermont, Lusignan and Rennes). For the other sites, the increase in yield is significant and varies from 0.9t/ha at Versailles to 2.4t/ha at Saint- Étienne in the near future and from 0.8t/ha to 2.6t/ha in two thirds of cases in the distant future. If we now examine the components of yield (fig. 3), we see that 70% and 55% of yield variability is related to that of the number of grains (NG/m 2 ) for the early and late varieties respectively. In both cases and for most of the sites, we find an increase in NG/m 2 with climate change. This suggests that the water and heat stresses resulting from climate change are not critical during the preflowering phases when this component is being formed: the advancement of stages partially limits the occurrence of stress (cf. TIMING section). On the other hand the sites differ as regards grain filling: for a given number of grains one can find differences in the 1000-grain weight (TGW) of up to 20%. Grain filling is more affected by water and heat stress towards the end of life for the late variety, but the longer duration of grain filling combined with the increase in CO 2 buffers these effects. This is illustrated by the concomitant changes in TGW and NG/m 2 of the two varieties which shows a larger fall in TGW for the late variety for the same NG/m 2. However it remains clear that the responses to climate change are very site-dependent: for example at Avignon, NG/m 2 and TGW are always limiting yield, but increase with climate change, whereas conversely at Mirecourt and Mons, with the likely climate during stem elongation improving, it allows the formation of a high NG/m 2 for the NF, although the conditions at grain filling preclude a maximal TGW. Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate 159

22 C 2 Wheat Figure 3: yield formation for an early variety (Soissons). Left, relation between yield and NG/m 2 and, right, the relationship between components for all sites. : RP, NF, DF. Impacts on the lifespan, the flowering date, and associated stresses Figure 4 illustrates, for the Mons site, the impact of climate change on the phenology of an early variety (PANORAMIX model) and its consequences for the occurrence of stress at critical phases in wheat growth: risks of frost at the start of stem elongation, water stress during stem elongation and grain filling, and finally, high temperatures (Tmax > 25 C) around flowering and during grain filling. DF WD=2 or 18mm P(FCD)= 0% HTD=5 SD=12 WD=52 or 96mm MONS Period NF RP WD=0 or 4mm P(FCD)=3% WD=0 or 5mm P(FCD) =7% HTD=4 SD=9 WD=22 or 70mm HTD=2 SD=8 WD=15 or 50mm Dates when developental stages occur Z30 - Z32 Z32 - Z55 Z55 - Z69 Z69 - Z75 Z75 - Z89 Z89 - Z92 P(FCD) = frequency of years with frost during stem elongation WD = water deficit (mm) if soil AWR is 193 or 104 mm during stem elongation (on left) and during grain filing (on right) HTD = number of days with Tmax > 25 C from heading to milk grain developmental stages SD = number of heat stress days during grain filling Rg = cumulative global radiation during stem elongation in MJcm 2 Figure 4: changes in phenology of wheat cv. Soissons with climate change and the associated climatic risks for the Mons site (PANORAMIX,WT, soil 1 for all, sowing date 10/10 ). 160 Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate

23 C 2Wheat Because of warming, flowering is brought forward at Mons by 9 and 17 days and the whole life cycle is shortened by 10 and 20 days in the near and distant future respectively. These trends are the same for all the sites with developmental shifts which are generally larger in the north than in the south (cf. TIMING section). For wheat the advancement effect is greater than that of shortening of grain filling, which moderates the yield loss. However these simple phenological shifts are also accompanied by increased occurrence of stress towards the end of growth: a doubling of the number of hot days around flowering, affecting the potential NG/m 2 formed before flowering (also increased because of the improvement in growing conditions during stem elongation), an increase of % in the number of heat stress days during grain filling, and lastly a doubling or more of the water deficit during grain filling in the distant future, reaching mm depending on the soil type. This is in agreement with the reductions in simulated TGW in the near and distant future. The water deficit during stem elongation also increases, to reach significant values (up to 18mm on a soil with a low AWR*) in the distant future. These increases in the mean water deficit are accompanied by large increases in the extreme values. Finally, climate change is not simply synonymous with deterioration in the conditions for wheat production. In fact, in addition to the increase in CO 2 which, depending on the model used, should compensate for the shortening of growth cycles, the frequency of ear frosts falls or remains unchanged while the incident radiation during stem elongation tends to increase. It is these factors which explain the increase in NG/m 2 mentioned earlier. We also find fewer problems of excessive water. Water sufficiency for wheat Figure 5: response of yield (a: left) of soft wheat to the variation in water sufficiency caused by climate change and variability due to the site (B: right). : RP*, : NF*, : DF*. Despite the advancement of developmental stages, the reduction in rainfall leads to a decline in water sufficiency (cf. WATER section) which is indicated by the blue arrow in figure 5a representing the change in ETR/ETRM* for most sites and the early variety Soissons. Also, the slope of the relations in figure 5a indicates that yield will respond more to this parameter in the future: we calculate that 60-80% of yield variations between sites is due to water sufficiency. On the other hand there is a positive effect: for the same level of water sufficiency, the yield increases in the NF and even more in the DF because of the beneficial effect of CO 2. Although less marked, we see the same thing for the late variety. However we should remain cautious about these conclusions as there is a large interaction between the effect of water sufficiency, the period and the site (fig. 5b): although most of the sites will experience increasing water stress with climate change, it may remain stable (as at Avignon) or even return to its reference value in the distant future after falling in the near future (at Colmar, Clermont, Saint-Étienne). This last effect is explained by the antitranspirant effect of CO 2. Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate 161

24 C 2 Wheat Wheat diseases At present the two commonest and most harmful leaf diseases of wheat (cf. HEALTH section), septoria leaf spot, predominant in the north of the Paris basin and near the Atlantic coast, and brown rust, a rampant disease in the south and south-west, are responsible for large yield losses. A reduction in their destructiveness of 20% and 15% on average in the case of septoria and brown rust respectively, is simulated for the DF. Although their biology is very different, the decline in the impact of both these diseases is due to the reduction in rainfall and the water sufficiency of the crop, and of course the yield increase with high CO 2. These effects result from two contradictory behavioural processes in the context of climate change. On the one hand the primary inoculum seems to be favoured in the NF and DF by the reduction in frequency of winter frosts (cf. HEALTH section) for septoria, and probably this is also the case for brown rust. On the other hand the number of days of infection and the maximum severity of the rust tends to decline, particularly in the DF. Septoria develops at later developmental stages, generally ending up with less damage. We should point out however that adaptation of the pathogenic fungi to climate change is not taken into account and so the risks are probably underestimated. Figure 6: changes in harmfulness of brown rust (yield losses as % of healthy yield; CERES) with the variety, the sowing dates and the periods of interest (RP, NF and DF). The choice of a combination of variety and sowing date is also a way of limiting disease damage. Thus figure 6 shows the advantage of late varieties in damage limitation, especially in the DF; similarly, and in agreement with current observations, late sowing dates also reduce damage. These data suggest the existence of ways of limiting future losses, even if the pathogen populations adapt. Also, as the sowing date affects both the yield potential (fig. 8) and the amount of damage, an optimum should be sought for each site. And durum wheat? Rarer than soft wheat, durum wheat represents only 1.4% of the French UAA, and 80% of it is grown in the south-eastern and south-western regions (fig. 1). The mean yield is 4.7t/ha with a maximum yield of 6.2t/ha in the centre-north and a minimum of 3.2t/ha in the south-east. The main outlet for durum wheat is for human food (couscous and pastry) with 65% exported. The quality requirements (technological and hygienic) are becoming stricter and stricter for export, in particular to north Africa. Nevertheless we find French growing areas and yields will remain unchanged. An increase in good quality yields is a major aim of the industry. In particular, the choice of variety is crucial in the control of fungal diseases. 162 Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate

25 C 2Wheat Compared with soft winter wheat, durum wheat has lower requirements for vernalisation, making it less vulnerable to increases in winter temperatures. On the other hand the plants freeze at higher temperatures ( 16 C). Although durum wheat is more sensitive to high temperatures (heat stress) than soft wheat, its shorter grain-filling period offers it better avoidance* possibilities. It has greater tolerance to water deficit due to stomatal closure at lower water potentials than for soft wheat, a capacity to extract soil water beyond the permanent wilting point, and also because of a lower leaf area index counterbalanced by an ability of the ears to carry out net photosynthesis. These properties explain why it is grown in the south of France. Comparison of the yield changes for soft and durum wheat for the SWSgW succession. Figure 7: comparison of durum wheat yield (Acalou) with that of soft wheat (Soissons) in a durum wheat-sunflowersoft wheat-sorghum succession with the STICS model. We see the same trend towards a yield increase with climate change for both species. Figure 7 compares the yield changes for durum wheat (cv. Acalou) and soft wheat (cv. Soissons) for each site. Although in general the durum wheat yield remains lower than that of soft wheat, we can see that at certain sites the two crops tend to converge in the NF. Hence on average climate change seems to benefit durum wheat more than soft wheat (attested by the slope of the regression line being greater than 1), which is in keeping with their respective physiological responses to water and heat stress. With a shorter lifespan, in particular for the grain-filling phase, the avoidance of high summer temperatures will be accentuated in durum wheat. Its lower leaf area index* and better capacity for extracting soil water will allow it to satisfy its water needs more easily. E Adaptation of varieties and practices faced with climate change The varietal response, by the timing of the growth cycle in relation to the climatic stresses, is an important lever for adapting cultural practices* to climate change. The sowing date in interaction with the variety is a second readily available means, provided that soil tillage and sowing are feasible (cf. TIMING section). For each period, we find a significant advantage to the earliest sowings, from 5% to 25% depending on the sites, for the current period, which is within the range of variation normally observed. Figure 8a shows an example of the response of yields to sowing date as % of the yield calculated for the current period and the earliest sowing date. The yield falls with the delay in the sowing date, especially so the earlier the variety. This effect is more than compensated for in the NF and DF by the advancement of stages and the increase in CO 2 for sowing dates from 20 September till Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate 163

26 C 2 Wheat 10 November. On the other hand the sowing of 1 st December results in lower or equal yields to those calculated for the RP with the same sowing date. Of the stress indicators shown previously (fig. 4), the water deficit around stem elongation increases dramatically beyond 18mm in the DF for the latest sowing dates, and with a variability which diminishes by half (fig. 8b). Figure 8a: effect of sowing date on changes in wheat yield with climate change for the Versailles site (CERES, soil 1, climate WT). Figure 8b: effect of sowing date on changes in the water deficit of wheat during stem elongation with climate change for the Versailles site (PANORAMIX, soil 1, climate WT). The same changes with climate change and sowing date are found for the other sites, but with very variable range between sites: small for Bordeaux, Lusignan, Mons, Rennes and Toulouse, a similar one at Versailles, Mirecourt, Saint-Étienne and Dijon, and finally a large one for Avignon, Clermont and Colmar. Finally, a study carried out independently on the optimisation of emergence conditions has shown, for the sites of Colmar and Toulouse, that sowing too early would impair the establishment of the stand because of the increasingly dry conditions in early autumn. All this suggests that a combination of variety and sowing date could be chosen for each site so as to avoid early stress (at emergence) and late stress (during grain filling) which would assure a yield level at least as good as at present. The comparison of different cropping systems including wheat (fig. 9) shows no significant difference between the systems. We should bear in mind, however, that since these simulations do not take account of biotic limitations, the yield of monoculture wheat is no doubt overestimated. Figure 9: comparison of various cropping systems for wheat yield at Toulouse and Versailles. 164 Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate

27 C 2Wheat One can see in figure 9 the increase in variability which accompanies the increase in yields in the future. One also sees that supplementary irrigation of the MWRW system, which was hardly worthwhile in the RP, will be clearly more profitable in the future, with water stress becoming as severe as, if not worse than, the nitrogen stress. The latter varies with the form of fertilisation (organic or mineral), but also with the previous crop (i.e. whether it is leguminous, and the period between crops). We see that organic wheat is not favoured by climate change (cf. ORGANIC FARMING section). F Uncertainty associated with the simulations carried out The section UNCERTAINTY and VARIABILITY analyses the causes of the yield variation in wheat. Here are its main conclusions. As regards the climate, the variability created by the models of general circulation is of the same order as that created by the downscaling methods for a given SRES scenario: these methods differ mainly in their prediction of the variability of the future climate, which has quite a big effect on yield, whose simulation is based on a large number of phenomena which have a threshold or are discrete, including, in particular, the occurrence of developmental stages. The second source of uncertainty arises from the different formalisms used in the crop models. In the case of wheat, the dynamic crop models STICS and CERES, although they differ quantitatively in their output* variables (yield etc.), predict the same changes, both in magnitude and variability. Conversely the static model PANORAMIX, although based on diagnostic variables including the effects of weather, does not simulate the same changes or the same variability (cf. YIELD and MODELS sections). We should add that, aside from the modelling approach used, failure to take account of certain processes (the nitrogen balance for PANORAMIX, excess water for CERES and PANORAMIX) partly explains the differences in behaviour between the models. The variability generated by the simulation options constitutes the third source of uncertainty. Of the three factors analysed for Toulouse and Colmar (heat stress or not, CO 2 or not, and photosynthesis calculated by a mechanistic model [Farquhar] or a more empirical one), only the CO 2 had a quantitative effect on the model outputs. This result leads us to include the effect of CO 2 on yield in all the models (cf MODELS section). However, it is probable that the choice of an early variety tends to minimise the heat stress. Lastly, the variability generated by the different practices is analysed and compared with climate change in the case of the two crop models (CERES and PANORAMIX) for the NF (because of the small effect of CO 2 ). In general the yield variability calculated by CERES is about 1.5 times that calculated by PANORAMIX. This is because CERES takes account of ear frost damage and nitrogen nutrition and explains why the year-to-year variability constitutes half of the total variability for this model. According to PANORAMIX, the effects of site, variety, soil and sowing date are greater than those of climate change. According to CERES, only the soil effect exceeds that of climate change. In spite of this big difference between the models, our results suggest that the choice of soils and varieties in combination with the sowing date is a way to at least limit the risks from climate change. Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate 165

28 C 2 Wheat Figure 10: proportion of the variance of yield attributed to various source depending on the model. 166 Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate

29 C 2Wheat What you need to remember For wheat, a crop with a long lifespan, moderately sensitive to summer water stress, the increase in temperature results in an advancement of the developmental stages and a shortening of the life cycle which limits some of the stresses: ear frost, water stress during stem elongation on a deep soil, shortening of stem elongation without affecting the available radiation. This set of advantages tends to allow an increase in the number of grains per m 2 for a large number of sites. Also, fungal diseases tend to decline in the near and distant future (cf. HEALTH section). In fact the general reduction in rainfall and hours of leaf wetness would tend to reduce the infection potential and the dispersal of the current main diseases, despite the temperature increase. Coupled with the reduction in excess winter water, this improvement in disease likelihood makes it possible to envisage more wheat growing near the Atlantic seaboard. However, although the advancement of stages allows some avoidance of late stresses during grain filling, the water sufficiency of the crop declines with climate change and the risks of heat stress increase in the northerly zones. Although these harmful effects of climate change are often compensated for by the increase in CO 2, the increase in the year-to-year variability suggests a prudent approach and the search for alternative solutions. Among these we should mention the possibility of amplifying the avoidance effect by bringing forward the sowing date or the choice of early varieties, provided that sowing can be done in soil which is moist enough to allow rapid emergence and crop establishment. The risk of ear frost constitutes another limitation to early sowing or growing wheat at high altitudes. Finally, the number of available days* to sow the crop increases with climate change in the different regions, which suggests greater flexibility in managing successions. What needs further study 3 The CLIMATOR project has approached the possible effect of climate change on crops in various ways. Apart from the uncertainty associated with the models, already mentioned above, several uncertainties arise from our lack of knowledge about ecophysiological mechanisms to understand the effects of climate change. For example the quantitative laws for the actions of extreme events (e.g. high temperatures) are often poorly understood and confused with other abiotic stresses. Similarly, the occurrence of rainfall and its variability over time influences the disease risk (through effect thresholds), and the efficacy of different practices (e.g. the efficiency of nitrogen applications, inter alia). 3 The choice of recommended practices and of soils standardised for all the regions does not allow us to really tackle the improvement of wheat quality in the future, while the predicted yield increase could be accompanied, on average, by a reduction in wheat protein content; here again, the regional aspect and spatial distribution will only really be able to be approached by taking account of soil variability between regions and by adapting practices to the local yield potential. 3 Finally, a better knowledge of the adaptation of physiological mechanisms to climate change, both from the point of view of the plant and from that of its enemies and friends (soil-borne or aerial), will also be necessary to plan realistically for the future. Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate 167

30 C 2 Wheat To find out more Brisson N., Mary B., Ripoche D., Jeuffroy M.-H., Ruget F., Nicoullaud B., Gate P., Devienne-Barret F., Antonioletti R., Durr C., Richard G., Beaudoin N., Recous S., Tayot X., Plenet D., Cellier P., Machet J.-M., Meynard J.-M. and Delécolle R., STICS: a generic model for the simulation of crops and their water and nitrogen balances. I. Theory and parameterization applied to wheat and corn. Agronomie 18, Number 5-6, Brisson N., Gate P., Gouache D., Charmet G., Oury F.-X., Huard F., Why are wheat yields stagnating in Europe? Field. Crop. Res., submitted. Gate P., Analysis of a large set of results from France. L écophysiologie du blé. Eds ITCF. Gate P., Blondlot A., Gouache D., et al, Impacts of climate change on wheat growth in France. What solutions and actions to undertake? OCL - Oléagineux, Corps Gras, Lipides 15(5), Green Book The crops Wheat Marie-Odile Bancal, Philippe Gate

31 Climate change and maize and sorghum grain crops: the main impacts Nadine Brisson C3 Maize & sorghum A Some key aspects of growing maize and sorghum in France aize yield (t/ha) aize area (ha) West 81 South-West Center-North 55 North-East 60 Center-East South-East Sorghum yield (t/ha) Sorghum area (ha) West South-West Center-North North-East Center-East South-East Figure 1: yields and areas sown with maize in the CLIMATOR regions with the proportion of grain maize in red Means Source Agreste. Figure 2: yields and areas sown with sorghum in the CLIMATOR regions. Means Source Agreste. The importance of maize in France (3.1 million hectares for grain plus silage, i.e. more than 10% of the UAA) reflects its importance on the world scale since, with wheat and rice, it is one of the most cultivated plants in the world. In France, maize is the second largest crop after soft wheat. It is no exaggeration to say that maize is behind a veritable farming revolution which took place during the second half of the 20 th century: an increase in area, diversification of outlets, genetic progress and stimulation of economic exchanges of agricultural products. Grain maize represents slightly more than 55% of the area sown with maize and 40% is situated in the south-west. After falling by about 10% between 2004 and 2008, maize acreages are increasing. The main outlets for grain maize are animal feed (about 30%) and export (about 40%, but very variable depending on seasonal weather). Next comes use on the farm of production and corn meal production; use of maize for energy (bioethanol production) is currently still quite small. Since the drought of 2003, maize has been accused of being the most water-consuming crop: how much so exactly? 40% and 7% of the areas of grain maize and silage maize respectively are irrigated, and 50% of the irrigated area is devoted to maize. However for several years the areas of irrigated maize have undergone big variation: a fall of about 15% between 2004 and 2006, followed by an increase of about 14% between 2006 and However we should not forget that the whole of the irrigated area of France only represents about 6% of the UAA. Compared with this giant of spring crops, grain sorghum seems like a small crop, with hectares for grain and hectares for forage in 2009, mainly situated in the south-west and centre-east. Only 25% is irrigated. Although at present animal feed grain constitutes its main outlet, it potentially possesses advantages comparable to those of maize for silage (its energetic value being that of an average quality maize and its nitrogen value being slightly higher than that of maize) and for the production of bioethanol and of biomass. On the other hand, efforts towards genetic improvement of sorghum are smaller than those for maize. Green Book The crops Maize & sorghum Nadine Brisson 169

32 C 3 Maize & sorghum Grain maize is mainly grown in monoculture: on the scale of the whole of France, half the maize area is in monoculture, but in the main producing departments, this figure ranges from 70 to 90%. On the other hand, sorghum is grown as part of cereal rotations and often on soils with a lower water reserve then maize. The growing conditions for sorghum, harsher than those of maize, explain much of the lower yield of the crop: 6t/ha as against 9t/ha (expressed at standard water content) for the national average for maize. B Simulation protocol of CLIMATOR Both crops are simulated with the STICS model. The maize crop is simulated in the context of two high-input systems, a monoculture and a four-year rotation*, always irrigated to 80% of its water requirement (or ETM*), whereas the sorghum crop is simulated for a rainfed rotation. Sites Systems 12 metropolitan sites including one at high altitude 3 systems: maize monoculture (IM), maize-soft wheat-rape-durum wheat rotation (MWRW), sunflower-soft wheat-sorghum-hard wheat rotation (SWSgW) Management Soils Variety Model Maize: irrigation to 80% of needs, 200kg N ha -1 in 2 applications, incorporation of stubble Sorghum: rainfed, 120kg N ha -1 in 2 applications, incorporation of straw Common to both crops: sown on 10 April, should be harvested before 1 October in rotation Soil 1: deep silty soil with a high available water reserve* (226mm) and average fertility Soil 2: rendzina with a low available water reserve (104mm) and low fertility Soil 3: leached hydromorphic soil with a high available water reserve (317mm) and high fertility Maize: Méribel (short-term) and DKC5783 (long-term) Sorghum: Friggo (short-term) and Fulgus (long-term) STICS Climate SRES: A1B*, ARPÈGE, downscaling methods*: WT*, QQ* and ANO* Some specific trials were also done for the Colmar and Toulouse sites: calculation of sowing dates to optimise the growth of the seedlings (adequate temperature, absence of frost, soil water content not too high or too low); simulation of irrigated sorghum. C Ecophysiological identity cards of maize and sorghum in the context of climate change Their similarities Maize and sorghum are plants with C4 metabolism, which restricts the exploitation of atmospheric CO 2 in terms of both photosynthesis and stomatal closure compared with C3 plants like wheat, sunflower and oilseed rape. Moreover their summer growth cycle, very similar, is driven solely by temperature, which makes them very sensitive to shortening of the duration of their growth phases caused by warming. For the same reasons, one can however base some hopes on earlier sowing and escape* when faced with water deficit. 170 Green Book The crops Maize & sorghum Nadine Brisson

33 Their differences C 3 Sorghum posses undeniable advantages of adaptation to drought, both because of its root system, which is deeper than that of maize, and its sparser foliage. However, because of earlier plant improvement work, maize has the advantage in terms of yield potential. D Impacts of climate change on the maize and sorghum crops, their physiology and management In their current production context, which we try to reproduce by an irrigated maize monoculture and a rainfed sorghum rotation, the yield* levels of the two crops are very different and difficult to compare. However, in view of the potential uses for sorghum, which are similar to those of maize, it seems essential to us to look at both crops side by side as regards their behaviour in the face of climate change* to envisage all the components of adaptation. Production levels Without changing the variety, grain maize yields in the main current production zones (southwest, Poitou-Charentes, Limagne, Alsace) is destined to fall in spite of irrigation (tab. 1). This fall, of about 0.5 1t ha -1 in the near future (NF*) could exceed 1.5t ha -1 in the distant future (DF*). However the northern edge of this zone (Versailles, Rennes) is not affected by this fall, with possibly even increases at Versailles, foreshadowing possible new growing regions for grain maize in the coolest areas (Mirecourt, Mons, Clermont-Theix). NF Sites Maize Sorghum Maize Sorghum Avignon -1.0 ns -1.5 ns Bordeaux Clermont 2.3 ns Colmar -0.8 ns -1.3 ns Dijon -0.3 ns Lusignan ns Mirecourt Mons Rennes ns 2.2 ns 4.0 Saint-Étienne -1.2 ns -1.6 ns Toulouse Versailles Table 1: changes in yields (in t ha -1 ) of maize and sorghum with their current management (irrigated maize monoculture, rainfed sorghum in a cereal rotation, long-term varieties at Toulouse, Bordeaux and Avignon, and short-term elsewhere). The significance of the change in comparison with the year-to-year variability is denoted as follows: Bold (p < 0.01), Italics (p < 0,05), Normal (p < 0,10), ns (non significant). For sorghum the downward trend is less marked. Although it remains very serious in the southwest in the NF, the DF offers possibilities of an increase everywhere except at Toulouse. As for maize, the border regions, currently too cool, appear as future zones for sorghum growing. DF Maize & sorghum Green Book The crops Maize & sorghum Nadine Brisson 171

34 C 3 Maize & sorghum Figure 3 shows that in spite of this better behaviour of sorghum, irrigated maize has the advantage in terms of yield, in particular for the sites* with low summer rainfall, like Toulouse. However, we see the difference between the two crops diminishing, making it appropriate to improve the growing conditions for sorghum. Figure 3: comparison of the change in mean yield of irrigated maize and rainfed sorghum when it is significant (cf. tab. 1). Phenology The increase in temperatures gives rise to a substantial advancement* of flowering, similar for both crops, of about 7-10 days in the NF, and days in the DF according to the sites and the climate downscaling method. This advancement of flowering results in a shift in the grain-filling phase during the months of June, July and August, i.e. the hottest months of the year. Consequently the duration of grain-filling will also be shortened, bringing forward the harvest even more and resulting in a fall in yield for irrigated maize. The advancement in harvest date will be much greater in the north with, for example for maize monoculture, harvesting a month earlier in the NF and a month and a half earlier in the DF at Versailles, Colmar, Dijon, Lusignan, Mons and Rennes, whereas in the south the maize harvest will be only two weeks earlier in the NF and one month in the DF. Sorghum will also mature earlier, but more so due to drought, being rainfed, (causing an increase in the crop temperature and a shortening of the grain-drying phase), so that the advancement of the harvest is 5-7 days greater in sorghum than in maize. Hence, despite a slightly longer theoretical physiological grain-filling duration for sorghum, it will nearly always be harvested before maize. Water For the two crops the water problem is not dealt with in the same way, at least in their current production context. For maize we analyse the change in the quantities of water used to irrigate the crop, whereas for sorghum, which is rainfed, we need to analyse the change in the water stress* suffered by the plant. The change in the irrigation requirements of maize will result in two contradictory effects. In fact because of the fall in the future climatic annual water balance (P ET0) (of about 100mm in the NF and mm in the DF compared with the recent past (RP*)), these additional water requirements will increase, tending to increase the irrigation water needs. However, assuming that the varieties remain unchanged, the irrigation period will be considerably shortened due to the shortening of the growing period caused by the higher temperatures, which will tend to restrict this increase in irrigation. Figure 4 shows that this restriction will have only a small effect in the NF, and that we must expect an increase in irrigation proportional to the climatic water deficit of about 50% or 40mm on average for the growing period. 172 Green Book The crops Maize & sorghum Nadine Brisson

35 C 3 Maize & sorghum Figure 4: change in mean irrigation over each period for a maize monoculture for soil 1, related to the annual rainfall (P) predicted by the QQ and WT downscaling methods* (scenario SRES A1B). In the DF the phenological* restriction will become more important and it might compensate for the increase in the climatic water deficit: this is predicted by the WT downscaling method. Here we are confronted by major uncertainty* in the long term, but this should not mask the certainty of a large increase in irrigation requirement in the medium term. Figure 5: comparison of the impact* of climate change on irrigation compared with other sources of variation. Two analyses were carried out separately for the two time periods NF-RP and DF-RP. If one compares this change with the other main sources of variation in the quantities of water applied as irrigation (fig. 5), it is substantial: it corresponds to 2/3 of the year-to-year variability, but also to half of the geographical variability and to 2/3 of the sampled soil variability in the project, suggesting possible crop relocations. More detail is given in the IRRIGATION and WATER sections. As to sorghum, it is undeniable that the water sufficiency of the crop will deteriorate in the NF (fig. 6), but, for the same reasons as before, that in the DF on certain sites we can hope for an improvement by the escape effect which will compensate for the increase in climatic water deficit. This optimistic scenario is the result of using the weather type (WT) downscaling method. The other method (QQ) predicts further deterioration of the water sufficiency for sorghum. Green Book The crops Maize & sorghum Nadine Brisson 173

36 C 3 Maize & sorghum Figure 6: changes in the water sufficiency of sorghum, defined at the ratio ETR*/ETM during the period from flowering to harvest. In this context, the soil can play an important role, especially to avoid preflowering drought, which is prejudicial to grain initiation. Hence, at most sites, the yield difference between the soils with a large water reserve (soil type 1) and those with a small reserve (soil type 2) increases. Nitrogen stress: should nitrogen fertilisation of maize be questioned? The quantity of nitrogen absorbed by the maize plants in monoculture tends to fall slightly in the NF and by about units in the DF. This change is due to the reduced growth of the plants due to the shortening of the growing period and to the increased leaching of nitrates by the increased irrigation which prevents the crop from benefiting from the extra mineralisation. However in the context of this cropping system* where the incorporation of crop residues is small, this nitrogen supplement provided by mineralisation is modest and decreases over time. Nevertheless our results show that there is no significant fall in the nitrogen nutrition index at flowering (which remains at 0.75 on average), indicating that there is no need to increase nitrogen fertilisation*. Conversely for sorghum, no significant trend is visible; the nitrogen nutrition is compatible with the yields obtained. We should point out however that sorghum has access to a supply of nitrogen from mineralisation which is larger than for maize monoculture because the straw of all the crops in the rotation is recycled. The absence of irrigation avoids the leaching of this nitrogen, but the drought prevents some of the available nitrogen from being absorbed. On the scale of the rotation Maize will be able to be rotated with winter crops in the main production zones, like Alsace, thanks to the increased temperatures which allow maize harvest before wheat sowing, and in the humid south-west (Aquitaine), due to the reduced risk of water-logging for winter crops (cf. WHEAT section). Table 2 shows that the yields of maize in rotation become, in the NF, fully competitive with monoculture. This result is explained partly by the better nitrogen status of the crops in rotation, to which one may add (although it has not been modelled) the better state of health. Note that the very low yield of maize in rotation at Colmar for the NF is due to the impossibility of the crop reaching the harvest stage with dry enough grain before planting the following wheat crop. In reality farmers resort to artificial drying to avoid this problem. 174 Green Book The crops Maize & sorghum Nadine Brisson

37 C 3 Maize & sorghum Colmar Bordeaux monoculture rotation monoculture rotation Yield (t ha -1 ) Plant nitrogen at harvest (kg N ha -1 ) RP 7.9 (0.9) 1.4 (0.4) 10.5 (1.2) 6.9 (0.6)* NF 7.2 (0.5) 7.7 (0.6) 9.6 (0.9) 8.4 (0.9) DF 6.6 (0.5) 7.0 (0.7) 8.9 (0.9) 9.8 (1.2) RP 214 (18) 227 (30) 201 (20) 205 (29) NF 209 (22) 258 (38) 187 (18) 264 (36) DF 199 (19) 250 (50) 193 (29) 275 (53) Table 2 : comparison of yields and nitrogen absorption (means and standard deviations) of maize in monoculture and in rotation with a short-term variety at Colmar and a long-term variety at Bordeaux except in the case marked * (short-cycle variety). E What possible adaptations? Earlier sowing This change, already perceptible for fifteen years, could continue, as shown in figure 7, at a rate of about one day every four years for maize, whichever downscaling* method is chosen. For sorghum, which does not benefit from possible starter irrigation like maize, earliness of sowing in the DF will be limited by the dryness of the seed-bed. Without changing the variety, the results obtained with both WT and QQ downscaling methods show that this advancement will have no influence either on the yields or the amounts of water applied as irrigation (to maize), or to the water stress suffered by the crop (sorghum). In other words, the escape effects expected from earlier sowing will not be there! This result is very well explained by a close examination of the climatic data: the increase in temperature goes hand in hand with the decline in rainfall (cf. CLIMATE section) which is accompanied by an increase in radiation which in turn causes an increase in potential evapotranspiration. Yet in C4 plants like maize and sorghum, this increase is not counterbalanced by an antitranspirant effect of CO 2. Figure 7: calculation of the earliest sowing dates optimising planting conditions in terms of temperature and soil moisture (the horizontal line corresponds to the prescribed date used in the standard exercise in the project, 10 April) and the effect on maize yield for the sites of Colmar and Toulouse (soil 1, WT downscaling method). Changing varieties The increase in heat availability* allows one to envisage growing grain maize and sorghum where it was not previously possible (e.g. Mons, Mirecourt and even Clermont-Theix in the DF), but it also offers the possibility of using longer- and longer-term varieties (tab. 3). Green Book The crops Maize & sorghum Nadine Brisson 175

38 C 3 Maize & sorghum SITE AVI BOR CLE COL DIJ LUS MIR MON REN STE TOU VERS PERIOD MAIZE DKC DKC -- MER MER MER MER MER DKC MER SORGHUM FUL FUL -- FRI FRI FRI FRI FUL -- PERIOD MAIZE DKC DKC -- DKC DKC DKC MER MER MER DKC DKC MER SORGHUM FUL FUL -- FUL FUL FUL FRI FRI FRI FUL FUL FRI PERIOD MAIZE DKC DKC MER DKC DKC DKC DKC DKC DKC DKC DKC DKC SORGHUM FUL FUL FRI FUL FUL FUL FUL FUL FUL FUL FUL FUL Table 3: varieties tested in the CLIMATOR project which optimize the heat availabilities for a sowing on 10 April for the two varieties tested in the project: long-term (DKC5783 and FULGUS) and short-term (Méribel and Friggo). -- : when the crop cannot complete its cycle in 8 years out of 10 due to winter cold or the need to sow a cereal crop (in the case of sorghum). RP NF DF Figure 8: effect of changing maize varieties on yield and irrigation for 3 sites (soil 1, WT downscaling method). However it is to be feared that irrigation will not increase and will not allow this possible lengthening of the growing period. Figure 8, showing the results from the same climatic dataset, shows the real benefits in terms of yield at Colmar and Versailles, with a modest cost in additional water: 4-5t ha -1 of extra yield for 10-20mm of extra water. These figures can be compared with the water savings made with an escape policy in the south-west: at Toulouse (fig. 8), by sowing short-term varieties, one would lose 3-4.5t ha -1 while saving about 60mm of water. Irrigating sorghum We have seen that, in its current production context (without irrigation, in rotation), sorghum is not at all competitive with maize; this is why we have made an exploratory trial of sorghum growing at Toulouse, with the same irrigation conditions as maize (80% of requirements) and with restricted irrigation (50% of requirements). It appears (fig. 9) that the yield of sorghum irrigated at 80% gradually approaches that of maize, whereas there was a difference of about 2t ha -1 in the RP. Also sorghum needs less water, using about 50mm less than maize. Note however that it is slightly less stable than maize, being subject to greater year-to-year variability. Moreover, sorghum withstands water rationing very well since, at 50% of its water requirement (about 45mm less), the fall in yield is less than 1t ha -1. Hence its hardiness gives sorghum several advantages which favour it over maize when faced with climate change. 176 Green Book The crops Maize & sorghum Nadine Brisson

39 C 3 Maize & sorghum Figure 9: comparison of simulation of an irrigated monoculture sorghum crop (cf. WATER section) (at 80% and 50% of its water requirements) with a maize monoculture for yield and water requirement (WT downscaling method and soil 1). Geographical location of crops and exploitation of soils In view of the limitation of water resources (cf. WATER section), the grain maize crop in the southwest is under serious threat. Indeed an average threshold value of 250mm per year (which corresponds, in fact, to variability of irrigation from 120 to 380mm according to the year) will be exceeded at Toulouse from the NF and at Bordeaux in the DF (with the QQ scenario). Conversely, climate change offers opportunities in the more northerly zones where grain maize could be grown with fully competitive yields (fig. 10). Sorghum, already grown mostly in the south-west (Midi-Pyrénées + Aquitaine + Poitou-Charentes account for 65% of the sorghum area), could bene fit from climate change to expand into the regions of the centre-east and west in the NF and the whole of France in the DF. It could also replace maize in the south-west, needing less water. Standard deviation (Tha -1 ) < > Mons Standard deviation (Tha -1 ) < > Mons Rennes Lusignan Bordeaux Versailles Toulouse Mirecourt Colmar Dijon Clermont St Étienne Avignon Rennes Bordeaux Lusignan Toulouse Versailles Mirecourt Dijon 5.1 Clermont St Étienne Avignon Colmar Yield (Tha -1 ) 7,0 7,5 8,5 9,0 9,5 10,0 10,5 11,0 8,0 Yield (Tha -1 ) Figure 10: change in yield of irrigated monoculture maize (left) and rainfed sorghum (right) when they can be grown (feasibility* > 80%) with the Wt downscaling method and soil 1. Regarding the soils, it could be worthwhile to exploit certain waterlogged soils (like soil 3) which for a small loss of yield (due to shallow rooting) to save a significant amount of irrigation water: for example 1.5t ha -1 of yield for 40mm of water at Toulouse between soils 1 and 3. Conversely, not confining sorghum to shallow soils could allow it to express its yield potential, which is far from the national average: for example at Toulouse, the mean difference between soils 1 and 2 is more than 3t ha -1. Green Book The crops Maize & sorghum Nadine Brisson 177

40 C 3 Maize & sorghum Changes in the cropping system If there has to be a reduction in the maize area, it could be made to the benefit of rotations which are favoured by climate change and offer a better nitrogen regime for the plants (cf. WATER section to see the effect of aquifer recharge). Moreover we have shown that sorghum makes good use of irrigation and hence it would be quite profitable to introduce supplementary irrigation. F Uncertainties We have already pointed out the uncertainty introduced by the methods for downscaling the climate (cf. UNCERTAINTY and VARIABILITY section) which is a major source of variation in the climatic data and especially in the physical coherence between the variables. It affects especially the sign of the change in irrigation requirements in the DF: the WT method predicts decreasing requirements between the NF and DF, whereas the QQ method predicts gradually increasing requirements. We wanted to see if this difference between the methods could cast doubt on the finding that a long-term variety would use no more water than a short-term one at Colmar. By only examining the medians, it seems in fact that one arrives at this conclusion (fig. 11), but the value which is exceeded in 2 years out of 10 increases by 40mm in the DF and with it the risk of exceeding the 250mm threshold. This example underlines the importance of considering the year-to-year variability in studies of the impact of climate change. Figure 11: comparison of irrigation statistics at Colmar with a long-term and short-term maize variety using the QQ downscaling method. The points correspond to the 2 and 8 deciles and the median. It also shows the importance of considering the timing of the growth cycle in relation to climatic stresses, and not just its length. 178 Green Book The crops Maize & sorghum Nadine Brisson

41 C 3 Maize & sorghum What you need to remember 3 Irrigated grain maize monoculture will be greatly disadvantaged by climate change, especially in the current production zones. The main reason is its timing as a summer crop, which, without a change in varieties, means that the grain-filling period will be shortened, leading to yield losses of about 1 to 1.5t ha -1 for the NF and DF respectively. The second reason is the water deficit which calls for about 40mm extra irrigation on average in the NF. In the DF, the change in irrigation needs is more uncertain and it could be that the developmental restriction compensates for the increase in water deficit. For rainfed sorghum, the yield reduction, due to the deterioration in water supply, is less marked, with even the prospect of an increase in the northern zones. For both crops, opportunities are apparent in the zones presently at the northern boundary of the maize area, but also in the new zones in the north-east or at high altitude. 3 Earlier sowing by one day every four years on average will be possible, but it will not reduce the irrigation requirements of maize or the water stress suffered by sorghum. On the other hand the use of longer-term varieties appears essential for maintaining, or indeed increasing, yields in the northern zones for a small amount of extra water. This could make up for the use of short-term varieties in the south-west to save water. 3 Changes in cropping systems can be envisaged: the inclusion of maize in a cereal rotation, which becomes feasible almost everywhere without the need for grain drying, and irrigation of sorghum which, eventually, could be as productive as irrigated maize. 3 More than for other crops, it seems important to consider geographical relocation of the maize and sorghum crops and growing them on different soils. 3 In the future, the much earlier maturity of maize crops will allow greater drying in the field and big savings on drying costs, which are currently very high. Green Book The crops Maize & sorghum Nadine Brisson 179

42 C 3 Maize & sorghum What needs further study 3 In view of the acuteness of the water problem and findings from previous work (in particular Lacroix et al., 2009), it seems important to study the climatic uncertainty exhaustively by analysing all the scenarios. Likewise it would be interesting to use another crop model to reveal the uncertainty due to imperfect knowledge, associated with the representation of the mechanisms in the models. 3 Other intermediate technical hypotheses between the extreme cases treated in the project could be studied: semi-early varieties and supplementary irrigation for example, for both maize and sorghum. 3 Collaboration with plant breeders would be good in order to improve sorghum yields. To find out more Amigues J.-P., Debaecke P., Itier B., Lemaire G., Seguin B., Tardieu F. et Thomas A., (2006) - Sécheresse et agriculture. Réduire la vulnérabilité de l agriculture à un risque accru de manque d eau. Expertise collective scientifique, synthèse du rapport, INRA Editions. 72 p. Brisson N., Launay M., Mary B. and Beaudoin N., (2009) - Conceptual basis, formalisations and parameterization of the STICS crop model, Quae (Eds), Versailles, 297 p. Brisson N., Gervois S., Diaz R., Benoit M., (2007) - Changements climatiques et pratiques agricoles : Regards vers le passé et le futur, In Seminary STICS, mars 2007, Reims, France. Lacroix B., Ruget, F. Lorgeou, F., Souverain F., (2009) - Impact du changement climatique sur maïs grain et maïs fourrage. Questions posées et pistes d adaptation. Changement climatique : conséquences et enseignements pour les grandes cultures et l élevage herbivore : colloque de restitution du projet ACTA, 22/10/2009, Lobell D.-B., Asner. G.-P., (2003) - Climate and management Contributions to recent Trends in U.S. Agricultural Yields, Science 299 : Ruget F., Bethenod O., Combe L., (1996) - Repercussions of increased atmospheric CO 2 on maize morphogenesis and growth for various temperature and radiation levels. Maydica 41, Tardieu F. et Zivy M. (2006) - Amélioration génétique de la tolérance des cultures à la sécheresse. In: Amigues J.-P., Debaecke P., Itier B., Lemaire G., Seguin B., Tardieu F. et Thomas A. (eds). Sécheresse et agriculture. Réduire la vulnérabilité de l agriculture à un risque accru de manque d eau. Expertise collective scientifique, rapport, INRA (France). 380 p + annexes. 180 Green Book The crops Maize & sorghum Nadine Brisson

43 Climate change and grassland: the main impacts Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux 4 CGrassland A Some figures for current French production Center-North West Grassland yield (t/ha) Grassland area (ha) South-West Center-East North-East South-East Figure 1: yields and grassland areas in the CLIMATOR regions. Means Source Agreste. In 2007, temporary grasslands (sown grasslands of less than 6 years duration and containing at least one grass species) occupied 2.8 million hectares, or about 10% of the French UAA (Agreste, 2009). These areas constitute an essential component of forage systems. For the maintenance of their herds, farmers must produce enough quality hay to satisfy their needs over the whole year. Apart from its production function, grassland also plays a role in protecting the environment by limiting leaching, providing biodiversity, acting as a sink for carbon, and detoxifying soil pollutants. It is an important component of the landscape. Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux 181

44 C 4 Grassland B Simulation protocol in CLIMATOR The meadows simulated are grass, mowed and adequately fertilised. Sown in September, they remain in place for the five following years, and are rainfed. Sites Systems Management Soils 12 sites in metropolitan France including one at high altitude 1 system: ley sown 9 September and cut four times a year for six years, including that of sowing No irrigation, 200kg N ha -1 yr -1, all biomass exported with a residual biomass of one tonne per hectare. Soil 1: 226mm AWR RU, 140cm deep, surface texture SL (sandy loam) Soil 2: 92mm AWR, 60cm deep, surface texture SCL (sandy clay loam) Soil 3: 287mm AWR, 210cm deep, surface texture SL (sandy loam) Species Models Climate Perennial ryegrass and tall fescue STICS and PASIM SRES: A1B, ARPÈGE, downscaling: WT and QQ 182 Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux

45 C Identity card of the grassland system in the context of climate change* Impacts expected from climate change on grassland, its behaviour and management Three major climatic variables modified by climate change have a significant impact on the physiology of the plants: the CO 2 concentration of the air (denoted [CO 2 ]), the temperature, and the amount of water available to the crop. Their impacts on the variables of interest simulated as part of this study are summarised in figure 2. C 4 Grassland Figure 2: conceptual scheme of the impact of climate change on grassland. Climate change, simulated using various methods and with the chosen scenarios, reduces summer rainfall and is associated with an increase in radiation and ET0, which reduce more or less severely soil moisture and leaf production (Durand, 2007), with a consequent immediate impact on yield. The variation in soil moisture and drainage are also reduced, affecting the nitrogen leaching regime. On the other hand the increase in CO 2 concentration and temperature tend to increase production, notably in the wet period. The phenology* of temperate grasslands is generally accelerated by the rise in temperature by about 4-5 days per degree of warming (Chuine et al., 2006). Also, all the processes contributing to plant production have optimal temperatures higher than the means currently observed (Durand et al., 1999); morphogenetic activity should therefore be accelerated by global warming. Depending on the balance of favourable and unfavourable phenomena, particularly in spring (the period of maximum production) yield increases or falls depending on the site* and the level of sensitivity ascribed to them by the crop models*. This results in varying degrees of storage of carbon and nitrogen in the soil organic matter. Finally, the changes in the moisture and temperature conditions in the soil affect the activity of micro-organisms and mineralisation*, finally modifying in turn the availability of nitrogen for the plants and plant growth*. Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux 183

46 C 4 Grassland Ecophysiological identity cards of ryegrass and tall fescue in the context of climate change Similarities These two forage grasses are plants with C3* metabolism, which allows them to exploit the increase in atmospheric CO 2 in terms of both photosynthesis and stomatal closure (hence water economy). Moreover, their very similar annual growth cycle is essentially driven entirely by temperature, making good use of the prolongation of the growing season at the beginning and end of winter, caused by the warming. Differences Fescue has undeniable advantages in adaptation to drought, firstly because of its deeper root system than that of ryegrass, but also because of leaf and root senescence much more sensitive to drought. In the humid zones to the west, the feeding value of ryegrass and its persistence due to its high tillering capacity) favour it, whilst fescue is very common in the meadows of the centre and centre-west. D Impacts of climate change on grassland yields All the processes previously described are precisely accounted for with a variable influence depending on the phenomena considered in the mathematical simulations of the STICS and PASIM models, from the weather records provided (cf. CLIMATE section). Year-to-year yield variability* First and foremost, it is essential to bear in mind that the biggest source of variability of grassland yields*, exceeding that of climate change, is the year-to-year variability (fig. 4) which is simulated very clearly with the STICS model and with PASIM (although the variation is less with the latter). Water sufficiency The degree of satisfaction of water requirements of a crop is measured by ratio of real evapotranspiration (ETR*) of the crop in real conditions (dependent on the quantity of water available in the soil) to the maximal evapotranspiration (ETM*) which represents the quantity of water evapotranspired by the same crop grown with non-limiting water supply. Thus, the ETR/ETM ratio characterises the moisture state of crops (cf. WATER section). The two simulations PASIM and STICS agree in showing a decline in the water status of grasslands on an annual scale for both species, and more severe in summer (fig. 3) The decline in the water status can be explained quite logically by an increased demand due to the increased yields under the effect of the increase in [CO 2 ] and to the increased temperature and radiation (cf. CLIMATE section) which increase the evapotranspiration of the canopy. Meanwhile, the reduction in rainfall, especially in summer when the demand is greatest, amplifies the water stress. Thus the results show a decline in the water sufficiency* of grasslands which is greater on a shallow soil, which has a lower available water reserve* than that of a deep soil. According to the simulations of the two models, the beneficial effect of the CO 2 on stomatal closure and the fall in the real evapotranspiration are not enough to compensate for the increase in droughts and temperatures which increase the water demand (Tubiello et al., 2007). 184 Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux

47 C 4 Grassland Annual distribution of yield One of the main results of this study is the clear change in the distribution of the yield throughout the year. The trends show in fact an acceleration of production in winter and spring in the future until the available soil water is completely exhausted. In summer there is then a fall in the daily yield due to the more frequent drought episodes, which last until the autumn. Thus climate change should accentuate the problem of overproduction of herbage in spring and lack of forage in summer. Figure 3: annual change in daily production over the course of the year. Example of fescue grown at Rennes on a shallow soil, A1B scenario simulated with STICS and the QQ climate downscaling method. Level of yield and quality For perennial ryegrass, the two models give similar results. STICS however calculates higher yields than those of PASIM for fescue, probably because of the greater sensitivity of the root and leaf growth functions in STICS, which expresses more strongly the differences between the parameters assigned to each species. The intensity of the trend in response to climate change also varies between the two methods of simulation. Although both models predict a trend towards increasing mean yields for both downscaling methods (QQ and WT) in the distant future, the magnitude of this change varies with the species, soil and site. Figure 4: average change in annual dry matter yield on a shallow soil between the RP, the NF and the DF, over all 12 studied site for fescue (in red) and ryegrass (in blue), simulated with STICS (left) and PASIM (right) for the QQ downscaling method (solid) and WT (hatched). The standard deviations were calculated by taking the mean of the year-to-year standard deviations of each station. Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux 185

48 C 4 Grassland Hence, over the whole century, whereas the simulations with the QQ method predict a mean increase of 22% (PASIM 12%) compared with today for fescue and 19% (PASIM 10%) for ryegrass, those made with the WT method predict, in parallel with lower rainfall, a smaller increase, with only 8% (PASIM 13%) for fescue and 6% (PASIM 11%) for ryegrass. Thus, although the weather patterns originating from the two different downscaling methods result in similar trends, one expects quite wide ranges of variation (reflecting the sensitivity of the STICS model to the variations in the climatic data). However the trends obtained agree with observations reported from different experiments, done under natural conditions and in CO 2 -enriched atmospheres, which predict very variable yield increases for ryegrass, between 0 and + 34% (Lüscher et al., 1998). Nitrogen content of the harvest The Nitrogen Nutrition Index (NNI) represents the nitrogen nutritional status of a plant stand. Its value is the ratio of the observed nitrogen concentration and the critical nitrogen concentration, which is a function of the biomass of the stand (Lemaire et Gastal, 1997). Even for the RP, a period for which the yields are lowest despite a simulated mineral nitrogen fertilisation of 200kg ha -1 yr -1, the model always predicts a nitrogen deficit over the year (mean annual NNI<1). In other words, the nitrogen application is not enough to reach maximum yield, which is in agreement with earlier studies (Lemaire et Gastal, 1997) for which the optimum nitrogen level was reached with fertilisation of about kg N ha -1 for fescue for a single summer cut (as against 40kg N ha -1 in the case of our simulations). Figure 5: mean nitrogen content (%) over thirty years of fescue (black) and perennial ryegrass (white) in forage harvested at Lusignan, simulated by STICS from the climatic data of the A1B scenario, by the QQ downscaling method. Although on this basis the models are more difficult to parametrize, notably because of the dynamics of the soil organic matter, there is a positive interaction between the water supply level and that of nitrogen nutrition which is reflected in an amplification of the effect of the nitrogen on the growth of the dry matter when the water supply becomes optimal. The increase in summer drought in the future risks reducing the amount of nitrogen available to the grassland and thus reduces its summer growth and nitrogen content. However there is no clear general trend in the NNI: the negative and positive effects balance one another out. Climate change could however be accompanied by a marked reduction in nitrogen concentration if [CO 2 ] and yields increase. In the case of simulations made with constant fertilisation levels between 1970 and 2100, the increase in yields is bound to result in a fall in the crop nitrogen concentration due to the dilution effect. The fall in the nitrogen concentration for perennial ryegrass is on average 27%, which moreover agrees with the trials carried out by Soussana et al. (1996) who 186 Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux

49 C 4 Grassland observed a fall of about 25-33% in similar experimental conditions. Environmental balance Grassland, like all perennial crops, consumes water over a long period, which in turn depletes the layer of draining water. However, with time and as for all crops, the reduction in rainfall is accompanied by a smaller reduction in drainage. The quality of this water does not generally present any problems with grassland for cutting. The simulation results show that leaching* is practically nil on the deep soil, both for fescue and perennial ryegrass, because of its depth and its clay content which allow it to store a large quantity of water. On a shallow soil, leaching is greater, and higher for perennial ryegrass (which is shallower rooting) than for fescue. On the scale of a year, the results for all the sites and situations show nitrate concentrations in drainage water always below the drinking water standard for groundwater, fixed at 50mg.l -1. However the trend indicates a risk of it worsening in the regions already affected by the problem (Poitou-Charentes) and especially in the zones where the soils are shallower. An adjustment (downwards) of fertilisation therefore seems necessary, no doubt to the detriment of the nitrogen nutrition status of the crop and its nitrogen content. Finally, grasslands can play a role in carbon storage in soils. All will depend in the end on the organic matter balance of the soil (cf. ORGANIC MATTER section). Minor regional differences The increase in fescue yields is often greater at the stations situated in the east of France (fig. 6). This finding is attributable to the smaller reduction in annual rainfall predicted for the east than for the west and south of France, and confirms that water is the factor which will limit forage production even more in these regions in the future. In the case of perennial ryegrass, the calculations show a smaller regional difference in the yield trends Mons Mons 2 7 Rennes Versailles Mirecourt Colmar Dijon 8 28 Rennes Versailles 4-1 Mirecourt Colmar Dijon B/A C/A Lusignan Bordeaux Clermont St Étienne B/A C/A Lusignan 8 16 Bordeaux Clermont St Étienne Toulouse Avignon 5 22 Toulouse Avignon Evolution (%) Evolution (%) Figure 6: change (in %) in yields of fescue for the periods NF and DF compared with the RP period. The case of the CNRM series on a shallow soil. Left, STICS; right, PASIM. Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux 187

50 C 4 Grassland What changes in practices should be considered? Our simulations were made with fixed nitrogen fertilisation rates every year. The increase in yield due to climate change does not seem to require any adaptation of fertilisation, except to maintain high values of protein content in the forage. Climate change should therefore assure the needs for high annual dry matter production. Particularly during the last fifty years, varietal improvement has resulted in increasing the yield of grasslands in summer and autumn, levelling out the production throughout the year. This genetic improvement has, for example, improved the yield of perennial ryegrass by 0.29 t ha -1 on average every ten years (Sampoux et al., 2009), i.e. about three times more than the largest effect of climate change simulated in this study. However locally, the increase in summer water stress could be alleviated by supplementary irrigation. 188 Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux

51 C 4 Grassland What you need to remember 3 As a perennial crop, the growth of grassland continues throughout the year. Yet when the main factor limiting its growth is the temperature, global warming should extend the period of production and thus allow it to be used (for cutting or grazing) earlier in spring and later at the end of the year, provided that the fields are workable. Conversely, a reduction in the summer yield is predicted, which could accentuate the imbalance in summer compared with spring and thus increase the early use of stored forage which adds a lot to the farmers workload. 3 Hence climate change will have a moderate and rather positive impact on the levels of forage production in France. The results of this study (within the limitations explained previously) generally predict an increase of about 5-20% in forage production between now and 2100 with, however, a deterioration in the water and nitrogen status of grasslands and a slight trend to lower nitrogen contents in the crop for ryegrass. 3 Livestock systems, based on the exploitation of the herbage, directly dependent on the climatic effects which determine both the growth of the herbage and the constitution of the forage stocks, (Lemaire and Pflimlin, 2007), must adapt not only the management of the grasslands (especially fertilisation) but also the farm, through different management of the production, with probably more and more diversification of forage production (other herbacious species, legumes, sorghum etc.), in order to assure sufficient production in view of the climatic uncertainty. The extension of the production period into the winter and at the beginning of spring will probably result in a revision of schedules for grazing and the management of the reproduction of the herd. What needs further study 3 The present climate models* cannot reliably predict extreme events such as heatwaves or, conversely, intense frosts. Hence it is difficult at present to predict the effects that these extreme weather events, which may become more frequent, will have. For example in the case of heatwaves, one might imagine that they would have a significant effect on the mortality of grasslands, due to the combined effects of high temperature and drought. However these crops seem to be very resistant to these stresses, although their yield would be seriously reduced during longer periods than in a normal year. Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux 189

52 C 4 Grassland To find out more Agreste (2009) - Statistiques agricoles annuelles 2006, 2007 définitives et 2008 provisoire. [En ligne], Disponible sur : Chuine I., Morin X., Roy J., Salager J.-L., Sonié L. and Staudt M. (2006) - Fonctionnement d une communauté végétale soumise à un réchauffement atmosphérique et à une sécheresse édaphique. Programme GICC Gestion et Impact du Changement Climatique Séminaire de restitution, mai 2006, Paris. Durand J.-L. (2007) - Les effets du déficit hydrique sur la plante : aspects physiologiques. Fourrages, 190, Durand, J.-L., Schäufele R. and Gastal F. (1999) - Grass leaf elongation rate as a function of developmental stage and temperature: morphological analysis and modelling. Annals of Botany, 83: Lemaire G. and Denoix A. (1987b) - Dry matter summer growth of tall fescue and cocksfoot in western France. II Interaction between water consumption and nitrogen nutrition. Agronomie, 7 (6), Lemaire G. and Gastal F. (1997) - On the critical N concentration in agricultural crops. N uptake and distribution in plant canopies. In: Lemaire G. (Ed.), Diagnosis of the Nitrogen Status in Crops. Springer-Verlag, Heidelberg, pp Lemaire G. and Pflimlin A. (2007) - Les sécheresses passées et à venir : quels impacts et quelles adaptations pour les systèmes fourragers? Fourrages, 190, Lüscher A., Hendrey G.-R. and Nösberger J. (1998) - Long-term responsiveness to free air CO 2 enrichment of functional types, species and genotypes of plants from fertile permanent grassland. Oecologia, 113, Sampoux J.-P., Métral R., Ghesquière M., Baudouin P., Bayle B., Béguier V., Bourdon P., Chosson J.-F., De Bruijn K., Deneufbourg F., Galbrun C., Pietraszek W., Tarel B. and Viguié A. (2009) - Genetic improvement in ray-grass (Lolium perenne) from turf and forage breeding over the four past decades. XVIII th Meeting of the Eucarpia Fodder Crops and Amenity Grasses Section. Book of Abstracts, p. 84. Soussana J.-F., Casella E. and Loiseau P. (1996) - Long-term effects of CO 2 enrichment and temperature increase on a temperate grass sward. II. Plant nitrogen budgets and root fraction. Plant and Soil, 182, Tubiello F.-N., Soussana J.-F. and Howden S.-M. (2007) - Crop and pasture response to climate change. Proceedings of the National Academy of Sciences of the United States of America, 104 (50), Green Book The crops Grassland Jean-Louis Durand, Frédéric Bernard, Romain Lardy, Anne Isabelle Graux

53 Climate change and the oilseed rape crop: the main impacts Nadine Brisson 5 COilseed rape A Some key aspects of the oilseed rape crop in France Center-North Rapeseed yield (t/ha) West Center-East North-East Rapeseed area (ha) South-West South-East Figure 1: areas sown with rape in the CLIMATOR regions. Means Source Agreste. With more than a million hectares devoted to it, oilseed rape is a major crop in France, 4 th in importance after wheat, maize and barley. More than half of this area is in the centre-north. Mean yields average about 3 t ha -1 and are highest in the north-east, where the growing period is long provided that the crop is not frost-damaged. Since 2002, the areas sown with rape have increased, whereas the yields have tended to stagnate. Rape is grown for its oil and protein. The outlets for human and animal food compete with industrial outlets, especially the biofuel industry (the production of biodiesel). The fraction of the area destined for biofuel production went from 30% in 2005 to 60% in 2008; which explains the increase in area. Rape is a major crop in cereal rotations. Its long growing period, from September to June, and its ability to absorb a lot of nitrogen are both advantages for reducing nitrate leaching. Freeing the field at the start of summer, it allows early sowing and creates good sowing conditions for the winter cereals which follow. After harvest, volunteer plants often appear which can help to limit nitrogen losses (as a green manure). As to pests and diseases, rape has many enemies (Phoma, Sclerotinia, Oidium, weevils etc.) which need careful control. Green Book The crops Oilseed rape Nadine Brisson 191

54 C 5 Oilseed rape B Simulation protocol in CLIMATOR The rape crop has been studied as a component of a four-year intensive arable rotation* based on irrigated cereals. Sites System Management Soils Variety Model Climate 12 sites in metropolitan France including one at high altitude a single rotation: maize-soft wheat-rape-durum wheat rainfed, sown 20 August; 180kg N ha -1 in 2 applications Soil 1: deep silty soil with a high AWR* (226mm) and average fertility Soil 2: rendzina with a low AWR (104mm) and low fertility Soil 3: leached hydromorphic soil with a high AWR (317mm) and high fertility Olphi (short-term) and Pollen (long-term) which differ in their duration of stem elongation STICS SRES: A1B*, ARPÈGE, downscaling method*: WT*, QQ* and ANO* C Ecophysiological identity card of rape in the context of climate change Advantages As a C3 plant, rape can exploit the increase in atmospheric CO 2 concentration, which, in the absence of other constraints, will stimulate photosynthesis*. It is very sensitive to frosting of the foliage during its winter phase which can occur once the minimum temperature falls below 4 C. The lethal temperature is about 15 C: these frost damage risks will be markedly reduced by the increase in minimum temperatures. Moreover the calendar timing of its growth cycle (September to June) confers escape* properties on rape in relation to the shortage of summer rainfall towards the end of growth, which, combined with its deep rooting, make it a crop resistant to drought. However the photoperiodic and vernalisation* requirements of rape (it is a long-day plant) will tend to restrict the shortening of the spring phase of the cycle. Rape, which is also reputed for its good nitrogen absorption properties, resulting from both its high requirements and good root system, should be able to make use of the extra nitrogen coming from the stimulation of mineralisation* of the soil organic matter. Even more than now, if introduced into the rotation, rape should restrict the loss of nitrogen by leaching during its own life cycle. Weaknesses The difficulties of sowing rape towards the end of summer risk being exacerbated by drought and can lead to deferred sowing, to the detriment of the crop. It all depends In spite of its good nitrogen-absorbing capacity, rape will leave the soil very high in mineral nitrogen in the autumn. Indeed one can expect that the amount of mineral nitrogen arising from the mineralisation of the organic matter in the soil left bare by the crop should be increased, as rape frees up fields earlier and earlier, and the soil will be subjected to high temperatures, even though the action of these temperatures will be diminished by the drought. This constitutes an advantage to stimulate the start of the following crop, sown in early autumn (e.g. wheat), but also a problem in terms of pollution if the soil remains bare for too long, such as when a spring-sown crop follows. 192 Green Book The crops Oilseed rape Nadine Brisson

55 C 5 Oilseed rape D Impacts of climate change on the rape crop in a cereal rotation Changes in yields Standard deviation (Tha -1 ) < > Mons Rennes 4.1 Bordeaux Lusignan 3.6 Toulouse Versailles Mirecourt Colmar Dijon Clermont St-Étienne Avignon 3.8 Yield (Tha -1 ) Figure 2: change in yield* of oilseed rape over all sites, rotation: MWRW, cv: Olphi, soil 1, sown 20 August, WT downscaling method. On the map shown in figure 2, the yield levels of the recent past (RP*) are an indication of the ease of planting the crop at the end of summer (sowing fixed everywhere at 20 August) and of the frost danger at the sites*; hence they are higher near the Atlantic coast. We will try especially to examine their changes: in general, rape yield tends to increase. However this increase is particularly large at the sites near the eastern boundary (Mirecourt, Colmar, Dijon, Saint-Étienne) and in the mountains (Clermont), traditionally colder in autumn and winter. Hence it is the reduction in winter frost which seems, first and foremost, to account for the yield increases of rape. In the traditional arable zones, we see that yields at Versailles, Mons, Rennes and Toulouse show little change. Green Book The crops Oilseed rape Nadine Brisson 193

56 C 5 Oilseed rape In table 1 we see the significant benefit which the eastern regions may derive from the rape crop. NF-RP DF-RP Avignon ns ns Bordeaux ns ns Clermont Colmar Dijon Lusignan ns 0.8 Mirecourt Mons ns ns Rennes ns ns St-Étienne Toulouse ns ns Versailles ns ns All sites Table 1: changes in the yield of rape (soil 1, climate: WT). The significance of the change compared with the year-toyear variability* is denoted as follows: Bold (p < 0.01), Italics (p < 0.05), Normal (p < 0,10), ns (non-significant). Winter frost The reduction in winter frost damage to rape crops is undoubtedly the most positive effect of climate change*. In fact whereas frost-kill of rape has occurred 3 years in 10 at Clermont-Theix and Mirecourt and 2 years in 10 at Colmar, Dijon and Saint-Étienne in the RP, this risk will be less than 1 year in 10 everywhere in the near future (NF*), and finally will become very rare in the distant future (DF*). These results agree for all the climate downscaling methods. Phenology Taking the example of Versailles, figure 3 explains the changes in the duration of the developmental phases of rape. The autumn-winter phases, when emergence and vernalisation occur, are lengthened, because of drought in the case of emergence and the temperature increase in the case of vernalisation. Figure 3: partitioning of the change in the duration of developmental phases of rape at Versailles, represented by the 10-year running means. 194 Green Book The crops Oilseed rape Nadine Brisson

57 C 5 Oilseed rape Conversely the stem elongation leading to the flowering of rape is considerably shortened, whereas the reproductive phase is very little affected. Why? The stem elongation phase takes place in conditions which are warmer for two reasons: because of the later development due to delay in vernalisation and because of the higher temperatures. Later on, the calendar advancement of the reproductive phase more or less balances out the increase in temperature. These changes do not take place steadily throughout the century, but seem to be concentrated over the NF period. Figure 4: partitioning of the duration of developmental phases of rape for three sites over the three periods of interest (error bars = standard deviations*). On the scale of the whole growing period (fig. 4), if a shortening should occur, it is limited to about twenty days and applies only to the stem elongation phase. The greater spread of emergence, accompanied by greater year-to-year variability, will harm the establishment of the crop. Rape and nitrogen The stimulation of mineralisation of the soil organic matter (cf. ORGANIC MATTER section) would suggest an increase in nitrogen absorption. On the contrary, nothing clear-cut is apparent. Figure 5: comparative changes in yield and nitrogen absorption at Rennes (right)and at Versailles (left) as10-year running means. Green Book The crops Oilseed rape Nadine Brisson 195

58 C 5 Oilseed rape Although there is a good correlation between yield and nitrogen absorption, it seems that the difference between the change in the two variables increases from 2050 (fig. 5). This phenomenon is particularly clear at Versailles where the absorption remains constant whereas the yield increases at the end of the century; however it does occur at Rennes too. The important question is whether this represents a limitation of yield by nitrogen which would then limit the positive effect of CO 2. Figure 6: relations between yield and nitrogen absorption for all the sites. More generally, figure 6 shows clearly that for the same yield, the nitrogen accumulated in the plant will be less. This is due to the water deficit during the vegetative part of the life cycle which limits absorption by the plant, whereas the soil mineral nitrogen becomes more and more abundant due to the stimulation of mineralisation by the warming. To illustrate these phenomena, here are the mean values for the mineral nitrogen content of the soil (in kg N ha -1 ) and for the nitrogen nutrition index (NNI) for the Versailles site on 1 st April, at about the time of flowering: RP(50, 0.8), NF(80, 0.7) and DF(100, 0.6). In other words, an increase in the nitrogen fertilisation* would not help to correct this nitrogen deficiency within the plant. Water stress Figure 7: risk of a dry seed-bed (top 10cm) for each period, based on a threshold of 10% gravimetric water content. 196 Green Book The crops Oilseed rape Nadine Brisson

59 C 5 Oilseed rape The accentuation of water deficits, for rape, will affect the vegetative phase and in particular the establishment of the crop. For example at Versailles (fig. 7), the moisture content of the seed-bed will not only be lower (falling from 12% in the RP to 10 and then 9% in the NF and DF respectively on average), but the nature of its variation will be different because of different occurrences of storms at the end of summer. This will affect decisions about sowing. Thus one can show that, assuming that the farmer accepts a 45% risk of soil surface drying, sowings should be deferred by more than a month in the NF. As for the DF, things seem to become even more uncertain, with an acceptable window at the end of August/beginning of September, but which involves risks for the survival of the plants due to the increased risks of drought during the month of September. Comparable phenomena can be demonstrated for the other sites. Figure 8 confirms that the water sufficiency* of rape during the grain-filling phase is hardly affected by shortage of rainfall. The duration (about 50 days) and the advancement* (about 20 days) resulting from the increase in temperatures explains this result. In this respect the behaviour of rape is much better than that of wheat (and even more so than other crops), confirming that it would be pointless to apply supplementary water during grain filling. Standard deviation (%) < >3 99 Rennes Toulouse Mons Versailles Mirecourt Dijon Lusignan Clermont Bordeaux St-Étienne Avignon Colmar ETR/ETM (%) Figure 8: changes in the satisfaction of water requirements (in %) during the grain-filling phase. MWRW, cv. Olphi, soil: 1, WT downscaling method. The effect of the soil available water reserve (AWR) is mainly to satisfy the water needs of the plant during the vegetative growth phase, which influences the yield via the number of grains. Thus the differences in yield between the soil with a low AWR (soil 2) and soils with a higher AWR (soils 1 and 3) tend to increase with climate change: from about 0.7t ha -1 in RP, they increase to 1t ha - 1 in NF and 1.2t ha -1 in the DF. On the scale of the rotation As expected, we find an increase in the quantity of mineral nitrogen present in October when the wheat crop is sown, whatever the preceding crop. However, although this increase is similar in size in the near future for both rape and maize as preceding crops (~ + 25kg N ha -1 ), it is twice as much for rape in the distant future (~ + 70 as against ~ + 35kg N ha -1 ). The presence of a vegetative cover after a rape crop thus proves to be more and more important to limit the risks of leaching*. On the other hand, this extra available nitrogen is beneficial to the following wheat crop. Green Book The crops Oilseed rape Nadine Brisson 197

60 C 5 Oilseed rape E Possible adaptations A slight preference for short-term varieties Figure 9: comparison of yield of the varieties Olphi and Pollen (WT climate downscaling method, soil 1). It is very difficult, on the basis of our results (fig. 9), to declare a real preference for one or other of the varieties. It seems however that the variety Olphi, with a shorter life cycle due to its rapid stem elongation (60 degree-days less than that of the variety Pollen) will be slightly favoured in future. This result is explained by the better absorption of N during the vegetative phase, allowed by this slight escape phenomenon, which favours the formation of grain number. Dealing with the sowing problem Figure 10: influence of the soil on the duration of rape emergence and its year-to-year variability. The seriousness of the problem of sowing rape calls for adaptations*. We have already pointed out the delay in sowing, which in the NF will have to be deferred by days. But in view of the climate change dynamics in autumn, this tactic will not suffice in the DF when starter irrigation could be needed. One must also consider the influence of soil type. Light-textured soils, hence rather drying, will exacerbate the emergence problems. Thus between soil 1 and soil 2, with the lightest texture, the difference in the duration of emergence, already significant in the RP, will increase both in terms of its mean (going from 5 to 8 days in the NF and DF respectively) and in variability* (fig. 10). Supplementary irrigation in spring to improve oil quality Depending on the status of rape in the rotation and the outlets envisaged, it may be important to maintain the quality* of the oil. If so, one or two supplementary irrigations in spring will in future be very profitable. 198 Green Book The crops Oilseed rape Nadine Brisson

61 C 5 Oilseed rape F Uncertainties The uncertainties* which exist for the climate forecasts, and which are well represented by the three downscaling methods WT, QQ and ANO, create a string of uncertainties about the value of the agronomic results. They confirm, by statistical analysis, that certain small changes obtained with the WT method (fig. 1) are not significant: i.e. at Rennes, Toulouse, Avignon and Mons. However these uncertainties do not detract from the general trends, as Figure 11 illustrates: in fact, neither the conclusions about the problems of emergence nor those about the fall in nitrogen absorbed per tonne of rape produced are changed. Figure 11: influence of the downscaling method on the agronomic results for a) duration of emergence (to be compared with figure 10) and b) relation between nitrogen absorbed for all the sites for each period (to be compared with figure 6). Green Book The crops Oilseed rape Nadine Brisson 199

62 C 5 Oilseed rape What you need to remember 3 Rape has advantages when faced with climate change. With a reduction and then the disappearance of killing frosts in winter, the north-east and centre-east will become suitable for the crop. Protected by its photoperiodic and vernalisation requirements, the rape life cycle will remain long enough to avoid loss of yield. As part of a rotation, rape as a previous crop will provide more nitrogen for winter cereals. 3 Although its phenology*, coupled with a good root system, allows it to escape droughts during grain filling (even with long-term varieties), rape will however be seriously affected by droughts in early life, particularly at sowing. This aspect is its weakest feature faced with climate change. These autumn droughts threaten not only the establishment of the crop but also its nitrogen absorption during the vegetative phase and the quality of the oil at harvest, especially in its traditional growing regions in the centre-north. This shortage of nitrogen absorption prevents the rape from benefiting adequately from the increase in atmospheric CO 2. To escape somewhat from these nitrogen absorption difficulties, it will be preferable to grow varieties with rapid stem elongation. The western regions, rather less affected by this problem, appear as refuge zones for growing rape. 3 One can however envisage deferring sowing and using one-off irrigations to stimulate emergence and nitrogen absorption. 3 Although the uncertainty about the future climate introduces variation into the figures obtained, it does not invalidate the trends and conclusions above. What needs further study 3 It is essential to consider a specific study on the changes in the pests and diseases of rape under the effect of climate change, since a possible change in the number of treatments, and the sustainability of the crop, depend on them. 3 The solutions for adaptation have only been sketched out; they need to be refined. Other solutions may be possible, perhaps with the help of genetics, such as the capacity of the seeds and plantlets to withstand deeper sowing. 3 In terms of methodology, it would be good to use another crop model* to include the agronomic uncertainty in the results. To find out more CETIOM, 1988 Colza : Physiologie et élaboration du rendement. 158 p. CETIOM, Rencontres annuelles du CETIOM sur le Colza, Paris, 30/11/ p. Colnenne C., Meynard J.-M., Roche R., Reau R., Effects of nitrogen deficiencies on autumnal growth of oilseed rape. Europ. J. Agronomy 17 : Dejoux J.-F., Meynard J.-M., Reau R., Roche R., Saulas P., Evaluation of environmentally-friendly crop management systems based on very early sowing dates for winter oilseed rape in France. Agronomy 23 : Green Book The crops Oilseed rape Nadine Brisson

63 Some aspects of climate change and the sunflower crop Lydie Guilioni, Nadine Brisson, Frédéric Levrault 6 CSunflower A Some key features of the sunflower crop in France West Center-North Sunflower yield (t/ha) Sunflower area North-East Center-East South-West South-East Figure 1: yields and areas sown with sunflower in the CLIMATOR regions. Means Source Agreste. Occupying roughly hectares in France, i.e. 2.5% of the national UAA, sunflower is the source of the second largest oil produced and consumed in France after rapeseed oil. It is also used in animal feed in the form of cattle cake or forage, as well as for plant-derived chemicals (agricultural solvents, biolubricants, biofuel). Sunflower is a spring-sown crop with moderate water requirements, whose agricultural yields in each CLIMATOR zone* vary between 2.1 and 2.7t ha-1 (mean ). It can grow under quite restrictive conditions of water supply, makes good use of supplementary irrigation, is useful for controlling weeds in cereal rotations* and benefits the soil structure. At present it is commonest in the south-west zone where it occupies 7% of the UAA (mean ). Global warming could extend the possibilities of growing it in the parts of the country which at present are cooler. Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault 201

64 C 6 Sunflower B Simulation protocol in CLIMATOR Sunflower was studied as a rainfed crop grown in two ways: one purely didactic way as a monoculture with two models* (STICS and SUNFLO), and in a four-year rainfed arable rotation associated with wheat and sorghum (STICS model). Sites System Management Soils Variety Model Climate 12 sites in metropolitan France including one at high altitude Monoculture (SS) and rotation (SWSgW): sunflower-soft wheat-sorghum-hard wheat Rainfed, sown 10 April as normal, 60kg N ha -1 at sowing, harvested before 30 September in rotation and on 31 October in monoculture Soil 1: deep silty with a high AWR* (226mm) and average fertility Soil 2: rendzina with a low AWR (104mm) and low fertility Prodisol (short-term) and Mélody (long-term) SUNFLO (SS) and STICS (SS and SWSgW) SRES: A1B*, ARPÈGE, downscaling methods*: WT*, QQ *and ANO* Some prospective trials on sowing dates were also done for just the two sites of Toulouse and Colmar: test of a range of sowing dates widened to 1 March to 30 May with the SUNFLO model; calculation of sowing dates so as to optimise the growth of the seedling (high enough temperature, absence of frosts, soil moisture content not too high or too low) with the STICS model. C Ecophysiological identity card in the context of climate change Advantages Sunflower is a C3 plant and this aspect of its photosynthetic physiology allows it to make good use of the increase in atmospheric CO 2 concentration. Furthermore, sunflower possesses several adaptive mechanisms to drought which, in the future conditions of water shortage, confer on it several advantages. In particular it can keep its stomata open longer under conditions of limited water supply, which prolongs growth when water stress begins to set in. At the same time, sunflower greatly reduces its leaf area by restricting the growth of its foliage or by accelerating its senescence. This mechanism reduces yield in unfavourable water conditions, but gives sunflower great plasticity when better conditions return. Weaknesses Because of its summer growth cycle, sunflower, like all crops, will be subjected to advancement of developmental phases by increasing temperatures, but it will also be disadvantaged by a reduction in the seed-filling phase, since the advancement of the phase in midsummer will amplify the warming it experiences. We may fear that the seed-filling conditions may also have an adverse effect on the formation of seeds whose embryogenesis is disturbed above 32 C. Moreover, since it is usually rainfed, good planting conditions are essential to the success of the crop, yet it is probable that moisture conditions in the upper soil layers will be less favourable in the future. 202 Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault

65 C 6 Sunflower D Impacts of climate change on the sunflower crop Yield Whereas in the present production zones average yields are changing little, we see a big expansion of the growing regions for sunflower between the recent past (RP*) and the near future (NF*) into the north of France. By analysing the results in more detail, we see that the maps in figure 2, representing two extremes in terms of soil conditions, show that a soil with a good reserve will allow a small increase in average yield. However the year-to-year variability* increases. Standard deviation (Tha-1) < > Rennes Bordeaux Lusignan Toulouse Mons Versailles Mirecourt 3.6 Dijon 3.6 Clermont 3.8 St-Étienne Avignon Colmar Standard deviation (Tha-1) < > Rennes Bordeaux Lusignan Toulouse Mons Versailles Mirecourt Dijon Clermont 1.5 St-Étienne Avignon Colmar Yield (Tha -1 ) Yield (Tha -1 ) Figure 2: change in yield of sunflower in rotation on all the sites (STICS model) for soil 1 on the left and soil 2 on the right and WT climate, with the varieties Mélody at Bordeaux, Toulouse and Avignon and Prodisol elsewhere. If one considers the year-to-year variability (tab. 1), one notices that, for the traditional sunflowergrowing sites, the yield does not change significantly and that the general increase takes place near the northern boundaries or in the hills: Clermont-Theix, Mirecourt, Mons, Versailles and Rennes. However the number of days on which photosynthesis* functions without temperature restriction increases everywhere: it increases for example from 25 in the RP to 40 in the NF and then to 50 in the distant future (DF*) at Colmar, and from 40 in the RP to 50 in the NF and 60 in the DF at Toulouse. Thus the crop is subject to major limitations. Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault 203

66 C 6 Sunflower NF-RP DF-RP Avignon ns ns Bordeaux ns 0.4 Clermont Colmar ns ns Dijon ns ns Lusignan ns ns Mirecourt Mons Rennes St-Étienne ns ns Toulouse ns ns Versailles All sites Table 1: changes in yields of sunflower in rotation (soil 1, climate WT). The significance of the change compared with the year-to-year variability is denoted as follows: Bold (p < 0.01), Italics (p < 0.05), ns (non significant). Phenology Without changing varieties, we see an advancement* of all the growth stages: faster emergence, earlier flowering and harvest. This advancement is closely related to the increase in mean annual temperature (fig. 3): it varies from 4 to 6 days/ C for flowering and from 7 to 12 days/ C for harvest, depending on the downscaling method and the variety (being more marked with the long-term variety). Figure 3: advancement of sunflower harvest (cv Prodisol) for the WT downscaling method. The NF-RP and the to DF-RP. correspond to If we look carefully at the duration of growth stages (fig. 4), we see that the phases of emergence and seed-filling are shortened more than the vegetative growth phase. Thus the role of this latter phase in yield formation (establishment of number of seeds) is enhanced and if it is adversely affected by early drought or uneven establishment this will have bigger effects on yield. We see that at Toulouse the shortening of the emergence period reaches a threshold (cf. section E) due to the drying of the seed-bed, which acts in the opposite direction. 204 Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault

67 C 6 Sunflower Figure 4: partitioning of the growth cycle for Toulouse (cv Mélody) and Saint-Étienne (cv Prodisol) with STICS and the WT downscaling method. It is this new phenology* which allows the crop to expand into the north in cereal-based rotations, as shown in figure 5. Standard deviation (%) < > Rennes Bordeaux Lusignan Toulouse Mons Versailles Mirecourt Dijon Clermont St-Étienne Avignon Colmar Feasability (%) Figure 5: feasibility of growing Mélody in a SWSgW rotation using WT downscaling. Water stress The sunflower crop, traditionally rainfed, will be confronted by increasingly severe water stress throughout its growing period, especially in its traditional sites like the south-west. At the beginning of growth, although emergence does not seem to be too threatened (even if somewhat delayed; cf. previous section) since in April the seed-bed retains enough moisture for germination, the growth of the young plant will be. In fact the risks of seeing the soil (fig. 6: 10-30cm layer) dry out beyond the permanent wilting point increase, reaching 25% in the DF at Toulouse. Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault 205

68 C 6 Sunflower Figure 6: soil drying during sunflower sowing at Toulouse. During the seed-filling phase (fig. 7) there is a fall in the water sufficiency (the variable ETR/ ETM*: cf. WATER section), already very low in the example of soil 2, which is quite typical for sunflower crops. However we see in the DF escape* effects, by shortening of the growing period on certain sites like Bordeaux (where there is an increase in water sufficiency between NF and DF), which will benefit yield (cf. YIELD section) and thus suggest a method of adaptation* by bringing forward sowing dates (cf. section E). A displacement of the growing zone towards the north, made possible by the warming, will offer a solution to this serious water stress problem (fig. 7). Standard deviation (%) < >16 Mons Rennes Bordeaux Lusignan Toulouse Versailles Mirecourt 44 Dijon Clermont St Étienne Avignon Colmar ETR/ETM (%) (flowering-harvesting) Figure 7: change in the satisfaction of water requirements (%) during the seed filling phase. SUNFLO, cv. Prodisol in the north, and Mélody in the south, soil 2, WT climate downscaling method. 206 Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault

69 C 6 Sunflower Heat stress Yield formation in sunflower is optimal within a certain temperature window. In fact, at the beginning of the process, floral initiation requires minimum temperatures above 15 C, whereas embryogenesis, the formation and filling of seeds are disturbed if maximum temperatures exceed 32 C. Yet the change in these two heat stresses is not symmetrical (fig. 8), i.e. the first falls little (or late) whereas the second increases considerably (for the fixed sowing dates) at the sites in the south. Figure 8: changes in heat stresses at Toulouse and Colmar as medians (middle values) and as 2 and 8 deciles (extreme values). Left, low temperature stress during floral initiation and right, high temperature stress during seed filling. QQ climate downscaling method. Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault 207

70 C 6 Sunflower E Possible adaptations Choice of variety The increase in temperature will gradually allow growth cycles* to be lengthened by using longer-term varieties. For the two extremes of earliness tested in this project, we see (tab. 2) that the replacement of Prodosil by Mélody takes place sooner at certain sites (Saint-Étienne) than at others (Dijon, Lusignan), but that by the end of the century all the sites (except the one in the mountains) will be suitable for growing a variety similar to Mélody. SITE AVI BOR CLE COL DIJ LUS MIR MON REN STE TOU VERS RP MEL MEL -- PRO PRO PRO PRO MEL - NF MEL MEL -- PRO PRO PRO PRO PRO PRO MEL MEL PRO DF MEL MEL PRO MEL MEL MEL MEL MEL MEL MEL MEL MEL Table 2: varieties tested in the CLIMATOR project which optimise temperature availabilities for a sowing on 10 April for the two varieties tested in the project: a long-term one (MÉLODY) and a short-term one (PRODISOL). -- : when the crop cannot complete its growth cycle more than 8 years out of 10. WT downscaling method. Figure 9 shows that these varietal choices will allow yields to be maintained or even increased. Figure 9: yield of sunflower at Dijon, Lusignan and Saint-Étienne by optimising the choice of variety. SUNFLO, soil 2, WT climate downscaling method. Choice of sowing date Another way to lengthen the growth cycle is to bring forward the sowing date (fig. 10). The advancements tested here do not take account of the technical problems of sowing, nor the success of establishment, but simply the consequences on yield. Compared with the recommended date of 10 April, the advancement appears more worthwhile at Toulouse than at Colmar, especially in the DF when in certain years (2 out of 10) it can increase yield by 80 % and in particular limit yield losses in unfavourable years. On the other hand delaying sowing after the recommended date is always harmful at Toulouse and will become so in the DF at Colmar. 208 Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault

71 C 6 Sunflower Figure 10: effect of sowing date on yield (median, 2 and 8 deciles) for the variety Prodisol, simulated with SUNFLO (soil 2,WT climate downscaling)at Toulouse and Colmar. These results are partly explained by the escape from water stress during seed filling at Toulouse (fig. 11), but the main explanation is to be found elsewhere: the advancement of sowing mainly improves the water supply during the vegetative growth phase and later allows escape from heat stress towards the end of growth. Figure 11: effect of sowing date on the water sufficiency during seed filling (median, 2 and 8 deciles) for the variety Prodisol, simulated with SUNFLO (soil 2,WT climate downscaling) at Toulouse (left) and Colmar (right). However technical problems, such as excessive seedbed water content, also influence the sowing date, as do constraints associated with the physiology of germination and the growth of the seedling before emergence, so that advancement of the sowing date is restricted (fig. 12): at Colmar, sowings could be brought forward on average from 20 to 5 April and at Toulouse from 2 April to 20 March. We see in the DF a trend towards later sowing caused by drying out of the seed-bed. Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault 209

72 C 6 Sunflower Figure 12: calculation of sowing date optimising the establishment of sunflower (STICS, soil 1,WT climate downscaling) at Colmar and Toulouse. F Uncertainties The results put forward in this study are tainted with uncertainties* (cf. UNCERTAINTY and VARIABILITY section). We can work out the agronomic uncertainty by analysing the major differences between the two models used (STICS and SUNFLO: fig. 13). Figure 13: boxplots (with extreme values, medians, 2 and 8 deciles) to compare the STICS and SUNFLO models at Toulouse for the deep soil (1) and the shallow soil (2). The periods are identified by the letters (A = RP, B = NF, C = DF). STICS is much more variable and sensitive to water shortage. This sensitivity operates, in particular, by accelerating the senescence of the foliage or by limiting its growth in the vegetative phase. SUNFLO is much less severe, as it assumes this senescence induced by water stress to be negligible. The result is lower yields for STICS for a shallow soil, whereas for a deep soil the two models are on average very close. STICS is also more sensitive to drought in the DF. However in general we find with both models very little change in yield at Toulouse. We have also tested the climatic uncertainty through the three downscaling methods. It appears that this uncertainty is not significant for the estimation of the advancement of phenological stages, whichever model is used. As regards yield estimation, we have tested the climatic uncertainty with the most sensitive model, i.e. STICS (fig. 14), and this analysis shows that for the whole of France the main signal provided by the climate trend is much greater than the uncertainty associated with this trend. 210 Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault

73 C 6 Sunflower Figure 14: comparison of mean yields over all the sites for the 3 downscaling methods. STICS model soil 1. What you need to remember 3 Sunflower is a crop which responds rather well to climate change. Its expansion to the north as part of cereal rotations will be possible in the NF. In the current production zones one should expect little change on average without changes in practice, the positive effects of the increase in atmospheric CO 2 compensating for the negative effects of water stress. However the increase in year-to-year variability, due largely to drought during the vegetative phase, could be reduced by the use of starter or supplementary irrigation (cf IRRIGATION section). The gradual choice of long-term varieties and the advancement of sowing offer prospects of yield increases, but not in every case. What needs further study 3 The interactions between the effects of high temperatures and those of CO are poorly 2 understood and could limit the benefits of the latter. Other effects of climate change were not considered, such as the effects of weeds and sunflower pathogens. The adaptation of practices was only partly analysed for two sites: a more systematic study would be desirable. Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault 211

74 C 6 Sunflower To find out more Casadebaig P., Debaeke P., Exploring genotypic strategies for sunflower drought re-sistance by the means of a dynamic crop simulation model, in Proceedings of the 17th International Sunflower Conference. Casadebaig P., Debaeke P., Lecoeur J., Thresholds for leaf expansion and transpira-tion response to soil water deficit in a range of sunflower genotypes. European Journal of Agronomy 28, Debaeke P., Bertrand M., Évaluation des impacts de la sécheresse sur le rendement des grandes cultures en France. Cahiers Agriculture 17 : Green book The crops Sunflower Lydie Guilioni, Nadine Brisson, Frédéric Levrault

75 Climate change and viticulture: the main impacts Philippe Pieri 7 CGrape vines A Some key features of the grapevine in France Center-North West Vine yield (t/ha) Vine area (ha) South-West North-East Center-East South-East Figure 1: wine-growing areas in the CLIMATOR regions. Means Source Agreste (see also onivins.fr/pdfs/statiques/localisation_du_vignoble_français.pdf). The importance of the grapevine in the French agricultural economy is considerable. In fact the vine is one of the main farm crops in terms of gross value, with about 9 000M of gross profit for the whole country (including in Champagne and in Bordeaux). It is a crop which enjoys a high added value compared with other farm products. The viticultural sector is also one of the main agricultural employers. Production is segmented into industries: table grapes, appellation contrôlée wines (VDP), other wines and wines for making brandy. All contribute strongly to national exports. The grapevine is a perennial crop, often the only crop on specialised farms. Production systems are extremely varied, and frequently integrated with wine-making. This is a crop for which the added value due to the quality* is genuine and indeed crucial. Yet the quality of the product is very sensitive to seasonal weather ( the vintage effect and, to a certain extent, the soil effect ( terroir )). Considerable sensitivity to climate change is thus expected. The recent past shows moreover noticeable changes which are attributed to climate change* (advancement of phenology* and harvest dates, increases in sugar concentrations, increasing frequency of vintage years, problems with phenolic maturity etc.). Often (but not always) the economic results are unconnected with the yield. Crop yield and quality depends on diverse varieties (or cépages), adapted to the climate of the region and to the type of production. In the present production systems, vines exploit poor soils since, although a summer crop, they are relatively economical with water. The grapevine also plays a very important role in the formation of the landscape. The vine and wine are thus the basis of noteworthy touristic and cultural activities. Green Book The crops Grape vines Philippe Pieri 213

76 C 7 Grape vines B Simulation protocol in CLIMATOR The results presented are based on simulations of vines grown on flat, weed-free land, rainfed or with occasional irrigation. Sites System Management Soils Varieties Model Climate 12 sites in metropolitan France including one at high altitude two systems: one rainfed (V) and one irrigated (VI) combined with two planting densities: 0.3 and 0.9pl/m 2 Irrigation is applied only when needed, satisfying 30% of the water requirements; 15kg N/ha are applied annually. Soil 1: common deep silty arable soil with a high AWR* (226mm) Soil 2: calcareous brown shallow soil with a low AWR (73mm) Soil 3: leached brown soil of average depth and AWR of 150mm Chardonnay, Merlot and Grenache, which differ in their earliness at flowering and the sugar content required at harvest STICS, BHV, BOTRYTIS (cf. HEALTH section) SRES: A1B*, ARPÈGE, downscaling*: WT*, QQ *and ANO* C Ecophysiological identity card of the vine in the context of climate change Advantages In the current production systems the grapevine is a crop reputed for its tolerance* of water shortage. The generally low LAI* and the arrangement in rows of crop canopy help to limit transpiration. Moreover the root system of well-established vineyards is often deep. All the vines in production are grafted and much of the adaptability of the rootstocks will potentially be exploitable. The quality of the grape and of the wine are generally favoured by moderate water stress*. The recent trend towards more frequent vintage years is attributed to this effect. Climate change could lead to an automatic resolution of the current problems of limiting vegetative vigour and yields. It could also reduce present problems with disease control and thus lead to a reduction in inputs (cf. HEALTH section). Current cultural practices tend to aim to raise the temperatures experienced by the plants: cultivation on exposed slopes, leaf thinning, warming of the soil surface etc; they could if necessary be reversed. More generally, there is room to manoeuvre due to a very great capacity for adaptation*: the variety, the rootstock, density, weed control, irrigation, fertilisation and other cultural operations. We can also expect a reduction in the frost risk to buds and young shoots in spring provided that the phenological advancement* does not bring budburst into a more frosty spring period. Vines being a perennial crop, climate change should favour photosynthetic activity after harvest and the rebuilding of reserves. Lastly, as a C3 crop, the vine responds to the increase in CO 2 concentration, in particular by the antitranspirant effect of CO 2 ; one can expect an improvement in the water use efficiency (WUE) during droughts. Problems Development is related to temperature, so for a summer crop which matures after the annual maximum temperature, climate change should result in an advancement of all the phenological stages and thus in maturation in much hotter, drier conditions. 214 Green Book The crops Grape vines Philippe Pieri

77 C 7 Grape vines Also, high temperatures have known adverse effects on photosynthesis and maturation, especially on the secondary metabolism of polyphenols and flavour precursors. In other words one is running a high risk of quality degradation due to high temperatures during maturation. D Impacts of climate change on the grapevine crop, its physiology and management Phenology and feasibility Figure 2: change in flowering (a) and harvest (b) dates simulated by BHV for 4 wine-growing sites. Merlot variety, A1B climate, WT downscaling* method. Phenology being related to cumulative temperatures, climate change results in a general advancement of the different developmental stages (fig. 2), whichever variety or downscaling method is chosen (cf. TIMING section). For the flowering date, this change is not accompanied by a noticeable change in the year-to-year variability*. For the harvest date, a similar change is observed with a tendency for lower variability; sites* where the variety used not to reach maturity become suitable for production, or become so more regularly, which also leads to a lowering of the year-to-year variability. For the current wine-growing areas, the entire growth cycle will be brought forward by days between the recent past (RP*) and the distant future (DF*). Green Book The crops Grape vines Philippe Pieri 215

78 C 7 Grape vines Standard deviation (%) < >45 Mons Standard deviation (%) < >45 Mons Rennes Versailles Mirecourt Colmar Rennes Versailles Mirecourt Colmar Dijon Dijon RP NF DF Bordeaux Lusignan Clermont St-Étienne RP NF DF Bordeaux Lusignan Clermont St-Étienne Toulouse Avignon Toulouse Avignon Feasability (%) <4 <6 <8 <2 >8 Feasability (%) <4 <6 <8 >8 <2 Figure 3: feasibility*, 30-year means and standard deviations, Merlot variety, A1B climate, WT downscaling method (left) and QQ (right). The change in the feasibility of growing the crop in different places using different varieties is due simply to the advance in the phenology and hence the expansion into the north of potential production (fig. 3) (cf. GEOCLIMATOLOGY section). The reduction in the variability in the DF will allow, for example, sites like Colmar, Versailles or Rennes to consider growing Merlot with no more risk than is currently the case at Bordeaux. This change is the same for both downscaling methods presented in figure 3. Yield levels Figure 4: relative change in yield levels for 4 wine-growing sites as 10-years moving averages (the yields in absolute values are very different), dry systems, STICS model. The yield levels and their simulated changes are extremely variable depending on the growing systems* considered (variety x soil x density x irrigation or not) which involve combinations of diverse limiting factors. For the examples presented (fig. 4), the fall observed at Avignon and Toulouse can be attributed to the effect of an increased risk of water stress, while the increase at Dijon and Bordeaux is due to a more favourable developmental timetable, which allows the exploitation of the increasing CO 2 concentration. 216 Green Book The crops Grape vines Philippe Pieri

79 C 7 Grape vines Water sufficiency The level of water sufficiency* of the vine is evaluated by the mean ETR*/ETM* ratio during the period from flowering to maturity (cf. WATER section). This ratio is related deterministically to the water balance during the same phenological period. We find that, despite the increase in climatic water deficit, the water sufficiency of the vine does not suffer a general deterioration: only the sites at Toulouse and Saint-Étienne show a clear fall in the ETR/ETM ratio. The future trend in the ETR/ETM is small for the Avignon site where the climatic water balance is already very negative. Likewise at Bordeaux the change remains negligible despite a strong drying of the climatic water balance (P ET0), but which should still have a few summer showers so that the conditions are not too severe. Figure 5: mean ETR/ETM during the period from flowering to maturity as a function of the annual climatic water balance indicator (P ET0), 30-year means, periods RP ( ), NF ( ), DF ( ), A1B climate, WT downscaling method, Merlot variety, common soil 1(left)and soil 2 (right), density 0.9, model: BHV. As a 30-year mean, the ratio ETR/ETM is correlated with the annual climatic water balance indicator P ET0 (fig. 5) once the latter falls below 200mm. This indicator thus constitutes a good predictor of the change in the level of satisfaction of water needs for the vine. The new sites where winegrowing becomes possible in the future tend to be situated on the plateau above the P ET0 value of 200mm and therefore seem to be protected from a fall in water sufficiency. Mons appears to be a particular case where an improvement appears because of a favourable climatic interaction. The form of this general relation ( a plateau then a fall related to the fall in P ET0) is modified to some extent by aspects of the growing system. Thus the effect of the available water reserve of the soil is dominant and results in a general displacement of the level of the ETR/ETM ratio, without any clear change in the slope of the linear part of the relation. The planting density has a smaller effect because the lower transpiration of the vines is largely balanced out by the increased evaporation from the soil surface. On average, the variety also has quite a small effect on the ETR/ETM ratio, the displacement of the floweringmaturity developmental period being insufficient to significantly alter the water status, which results from long-term cumulative water flows. The variety effect is thus smaller when the soil water reserve is larger, and plays a buffering role. Green Book The crops Grape vines Philippe Pieri 217

80 C 7 Grape vines Heat stress and quality index Figure 6: consequences of climate change on the advancement of the harvest date (left) and on the rise in the mean minimum temperature during maturation (veraison-maturity) (right) for 4 sites and the traditional growing systems associated with these sites, as 10-years moving averages. As the vine is a summer crop which is harvested after the summer maximum temperatures, the advancement of the phenological calendar (fig 6 left) brings forward the maturation period to earlier in the summer, hence into generally hotter and drier conditions. The phenological advancement thus exacerbates the effect of atmospheric warming, and the actual temperatures experienced by the plants and the grapes during maturation should increase well above the mean heating of the atmosphere (fig. 6 right). Moreover studies have shown the relevance of an indicator called night coolness index for the quality of the harvest, particularly from the point of view of flavour (Tonietto and Carbonneau, 2004). The predictable trend of this indicator (QUAL: mean minimum temperature during maturation) shows a drastic change in maturation conditions towards a degradation in the quality of the grapes (fig. 5 right). Although simple, this index is a good indicator of the possible harmful effects of climate change on grape quality, and in particular on the content of phenolic compounds. This conclusion is reinforced by the problems of delayed phenolic maturity observed in recent years. Environmental balance: pesticides and water As regards the consumption of pesticides, the impacts of climate change will be contrasting. The simulations for Botrytis on Merlot indicate that the level of epidemic risk, and hence the number of treatments, will fall for the southern sites where this variety can already be grown, but will be high for the northern sites where it will become feasible to grow it. By extrapolation to all the fungal diseases, the use of fungicides should therefore decline in the current wine-growing regions but will be considerable in the new ones (cf. HEALTH and GEOCLIMATOLOGY sections). As to the water balance under vines, climate change will result almost everywhere in a significant fall in the return of water to the environment, i.e. to the deep groundwater. This aquifer recharge and its future trends are evaluated by the PERCOL* variable, which quantifies this cumulative annual flux over the agricultural year, i.e. from 1 October to 30 September (cf. WATER section). Figure 7 : mean PERCOL over the year (mm) as a function of the annual climatic indicator of water balance (P ET0) (mm), 30-year means, period PR ( ), NF ( ), DF ( ), climate A1B, WT downscaling method, variety Merlot, common soil 1 (left) and soil 2 (right), density 0.9, model BHV. 218 Green Book The crops Grape vines Philippe Pieri

81 C 7 Grape vines PERCOL varies with the annual climatic indicator of water balance (P ET0), with a linear relationship when P ET0 is positive or slightly negative and a levelling out when PERCOL approaches 0, its physically impassable lower limit (fig. 7). As for water sufficiency, P ET0 thus constitutes an easy-touse explanatory variable for predicting the effects of climate change on PERCOL. In climatic water balance conditions which are not (or barely) negative, the slope of the relation is close to 1. When P ET0 becomes negative, only the Avignon site stands out, probably because of the high intensity of autumn rains, a feature of the Mediterranean climate. The impact of climate change is then small even if P ET0 changes considerably, as for example at Toulouse, and is negligible at Avignon. This relationship is also influenced by the different parameters of the cropping system, mainly the available water reserve of the soil: a larger reserve enables water to be stored which will be evaporated or transpired into the atmosphere, thus reducing PERCOL (cf. WATER section). Due to the interplay of various compensating factors, plant density and variety have no significant influence on this variable. Uncertainty and ranking The uncertainties attached to the long-term climatic predictions* are evaluated by the differences which appear between the weather patterns derived from the different downscaling methods (WT, QQ and ANO) for a given SRES scenario (A1B). Through agronomic impact models, they in turn affect the uncertainties applying to the agronomic results. For a variable which is hardly affected by these models, like QUAL, these uncertainties are directly apparent. The WT method for example tends to reduce the intensity of the change due to climate change (fig. 8). However the general trend of the variable always remains consistent. The risk of the degradation in grape quality is therefore clearly confirmed. Figure 8: change in QUAL (30-year means and standard deviations) as a function of the downscaling m ethod (QQ, WT and ANO) for several wine-growing sites. Variety: Merlot; model: BHV Other variables, arising from the crop models, result from complex and cumulative processes which are liable to interact. In this case, the overall uncertainty (much less predictable) also incorporates the uncertainty produced by the impact models. An analysis of variance thus allows all the combinations of uncertainties to be ranked, and their interactions to be quantified. The results show, for example for the variable ETR/ETM, the importance of the interactions and the large effects of the model (fig. 9). These results lead us to treat with caution this variable, on which the effect of climate change is quite small. Green Book The crops Grape vines Philippe Pieri 219

82 C 7 Grape vines Figure 9: analysis of variance of the variable ETR/ETM (expressed as % of the associated variance) for the changes over RP-NF and RP-DF. Breakdown into main effects and interactions. The year effect within each 30- year period represents the residual variance. Adaptation and optimisation The procedure consists of examining, for a given place, which combination of variety, soil, planting density and [rainfed vs irrigated] could maintain constant (δ = 0 as a 30-year mean with RP as the reference period) or barely affect certain variables of interest (FLO*, REC*, ETR/ETM, PERCOL, QUAL) considered separately. The results of this tentative optimisation are still preliminary, but they can give quantified indications of possible margins for manoeuvre. Figure 10: total annual quantity of irrigation water (mm) needed and sufficient to maintain ETR/ETM at at least 0.3 as a function of the annual climatic water balance indicator (P ET0) (mm), 30-year means, RP ( ), NF( ), DF ( ), climate A1B WT, variety Merlot, soil 1 (left) and soil 2 (right). Irrigation is a way of maintaining the water sufficiency of the plants, provided that water remains available. The change in irrigation water requirements was simulated by the STICS and BHV models, based on a system which is economical with water (buried drip irrigation, ETR/ETM maintained at 0.3 or more). Unsurprisingly, climate change clearly led to an increase in irrigation requirements, especially for the driest climates and the soils with the lowest AWR (fig. 10). The predictive value of the annual climatic water balance indicator (P ET0) is again confirmed since, once they depart from 0, the volumes of irrigation water needed vary linearly with the deficit. 220 Green Book The crops Grape vines Philippe Pieri

83 C 7 Grape vines Figure 11: test of the effect the cropping system (in its broad sense) on the water sufficiency (ETR/ETM) from four traditional systems associated with four wine-growing sites, as 10-years moving averages. The black line corresponds to the present (RP) level of water sufficiency. STICS model. The possible adaptation strategy to climate change of the cropping systems was tested in a very rudimentary way for the water sufficiency criterion (ETR/ETM) at sites representative of a viticultural tradition (fig. 11). At Dijon and Avignon, irrigation allows the water sufficiency to be maintained at least at its current level, whereas at Bordeaux the gain remains very marginal; a change in density is then much more effective (but probably disadvantageous in terms of quality and revenue). At Toulouse, starting with a low-density system, no method is able to maintain the initially very high level of water sufficiency; even a change in genotype. However this reduction in the water sufficiency could in this case be beneficial to the quality. Moreover, the maturation conditions everywhere alter in a way which is unfavourable to quality (fig. 6 and 8), a change which is further exacerbated by growing an earlier variety and only temporarily improved by choosing a late variety (fig. 12). Figure 12: mean minimum temperatures during maturation (veraison-maturity) at Toulouse for the three varieties as 10-years moving averages. Green Book The crops Grape vines Philippe Pieri 221

84 C 7 Grape vines What you need to remember 3 For the grapevine, probably the clearest effects of climate change are those expected on the phenology and hence also on the feasibility of growing the crop, whose area of distribution should expand towards the north. Between the recent past and the end of the 21 st century, the entire growing period will be brought forward by days, and sites like Rennes or Versailles will become perfectly suitable for viticulture. 3 As a direct consequence of the phenological advancement, large and harmful impacts are also predictable on the conditions during maturation of the grapes and hence on wine quality, especially in terms of flavours and polyphenols. In fact for the traditional wine-growing sites, increases in temperature during the maturation period of about at least 5 C are expected by the end of the century. 3 The water status of vineyards in general should undergo a very moderate reduction in the water sufficiency, but with a very clear reduction in aquifer recharge. Diagnostic relations established with the climatic indicator of water deficit (P ET0) make it easy to assess the impact of climate change on the vine. The sites and simulated weather patterns where P ET0 changes most result in the strongest impacts on the water status of the vine. 3 To bring forward the development and the change in the conditions of maturation, it will be very difficult to escape the harmful effects of climate change. To a certain extent, an adaptation of the growing systems and techniques could be envisaged (using northfacing slopes, cooling by irrigation, abandoning deleafing etc.), as well as adaptation of the genetic material. For the water status, numerous forms of action are available. Among them, irrigation has the advantage of not having to change the whole production system, provided of course that water remains available. 222 Green Book The crops Grape vines Philippe Pieri

85 C 7 Grape vines What needs further study 3 Although the main impacts of climate change on the grapevine crop have been well quantified, at least as trends, certain aspects of the physiology and of the response to climate change have not been treated, mainly due to a lack of models simulating these effects. As regards the physiology of the crop, one may cite for example the effects of high temperatures on photosynthesis, growth and development, the effects of the relation between vegetative vigour, yield and quality, the dynamic adaptation of rooting, the exposure to frost risks, the increased post-harvest growth and hence the increased allocation to reserves. In terms of adaptation, one may mention the effects of cover cropping, fertilisation, defoliation and other cultural operations, including the adaptation of the variety and of the rootstock. 3 Certain aspects like maturation and the climatic and microclimatic determination of quality are still poorly understood and hence the evaluation of the impact of climate change in this domain is still partial and imperfect. In general the effects of high temperatures on photosynthesis, growth and development could mitigate or magnify the results obtained in relation to the advancement of phenology. 3 Lastly, the assessment of the possibilities of adaptation is still incomplete and may be considered as simplistic since it does not take account of all the possible interactions. Everything in fact is potentially adaptable: (climate) x variety x soil x density x (rainfed/ irrigated). 3 However to successfully evaluate the impact of climate change and optimise the choices will be crucial for a long-term perennial crop with high establishment costs: current plantings will still be active in The difficulty in taking decisions is due to the imperfect ranking of the uncertainties, which is superimposed on the evaluation of the long-term economic risk. 3 Socio-economic aspects have not been tackled, and could limit the agronomic adaptation: a very restrictive regulatory framework, the weight of tradition (the notion of the terroirs [ascribing a wine product or wine type to a certain climate, soil, topography, exposure etc.] and of typicity ) and a strong market segmentation. Green Book The crops Grape vines Philippe Pieri 223

86 C 7 Grape vines To find out more Downey M.O., Dokoozlian N.K., Krstic M.P., Cultural Practice and Environmental Impacts on the Flavonoid Composition of Grapes and Wine: A Review of Recent Research. Am. J. Enol. Vitic. 57: Duchêne E., Schneider C., Grapevine and climatic changes: a glance at the situation in Alsace. Agron. Sustain. Dev. 25: Jones G.V., White M.A., Cooper O.R., Storchmann K., Climate change and global wine quality. Clim. Change 73: Lobell D.B., Field C.B., Cahill K.N., Bonfils C., Impacts of future climate change on California perennial crop yields: Model projections with climate and crop uncertainties. Agric. Forest Meteorol. 141: Schultz H.R., Lebon E., Modelling the effect of climate change on grapevine water relations. Acta Hort. 689: Seguin B, de Cortazar IG., Climate warming: Consequences for viticulture and the notion of terroirs in Europe. Acta Hort. 689: Tonietto J. & Carbonneau A., A multicriteria climatic classification system for grape-growing regions worldwide. Agric. Forest Meteorol., 124: Green Book The crops Grape vines Philippe Pieri

87 Some aspects of climate change and forests in metropolitan France Nathalie Breda, Alexandre Bosc, Vincent Badeau 8 CForests A Some key features of forests in mainland France (means ) The metropolitan French forest covers 15.7 million hectares, or 28.6% of the country or about half of the agricultural area. Its area has increased since the middle of the 19 th century. It is scattered randomly over the country (fig. 1 left) after historically having become re-established in zones unwanted for agriculture. The mean annual production is 6m 3 /ha/yr and the volume of standing timber in French forests is 2.37 billion m 3 (2005). This volume has increased by 28% in 19 years (27 million m 3 for an average year between 1986 and 2005) (fig. 1 left). During the same period, the volume per hectare went from 138m 3 /ha to 160m 3 /ha; the increase in volume however is not due to the increase in area, but to the harvesting rate being slower than the growth. This increase in standing timber increases the competition in mature stands. Figure 1: changes in the volume of standing timber in production forests over 20 years by region (in millions of cubic metres); the background colour illustrates the change as a percentage; the histograms the volumes on the two dates (left); and areas devoted to agriculture and to forest by CLIMATOR region( right). The majority of the French forest is privately-owned and deciduous. In fact privately-owned forest represents three quarters of the area of the metropolitan forest, or 11 million hectares. The forests of maritime pine, included in CLIMATOR, are mainly private cultivated forests for which we will consider choices of cultural practices (rotation lengths). The French forest is mainly composed of broad-leaved species (cf. table 1). Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau 225

88 C 8 Forests Source IFN, 2008 Area of forests for timber production (1000s of ha) Standing volume (millions of m 3 ) Gross annual production (millions of m 3 ) Deciduous (69 %) (62 %) 55.8 (55 %) Conifers (total) of which (30 %) 905 (38 %) 45.3 (45 %) maritime pine (7 %) 179 (7 %) 11.1 (11 %) fir, spruce, Douglas fir (11 %) 456 (19 %) 23.1 (23 %) Total Table 1: species composition of the French forest. Constraints to metropolitan French forest production The metropolitan French forest is at present confined to the poorest soils in terms of mineral fertility and physical limitations (proportion of coarse fractions for example) and hydrodynamic ones (waterlogging). In CLIMATOR, we only deal with water balance limitations and not those due to possible mineral deficiencies. B Simulation protocol in CLIMATOR The impacts* of climate change on the French forest were assessed using a set of simulations based on 5 criteria. Three models* of the analysis of the impacts of climate on the forests They use specific approaches (tab. 2), previously used within the forest project CARBOFOR (2010). BILJOU (stand water balance), GRAECO (integrated water, carbon, growth balance with forestry management) and EVOLFOR (a niche model). Note that these models are conceptually different and not comparable. Three types of canopy Deciduous trees, conifers with a high leaf area index* (firs, spruce, Douglas fir etc.) and plantations of maritime pine with a herbaceous undergrowth. This restricted selection of canopy types does not represent the whole of the French forest, and especially not the features of Mediterranean woods, mixtures of deciduous and coniferous woods, mountain forests and short-rotation plantations for biomass production. 226 Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau

89 C 8 Forests The climate* To the standard configurations (emission scenario A1B*, ARPÈGE model and three downscaling methods*) were occasionally added all the other climatic configurations described in the CLIMATE section, i.e. seven additional series. Model Approach Canopy type analysed in CLIMATOR Key variable or other in CLIMATOR Time step BILJOU Stand water balance (1) Deciduous trees, LAI = 6, season from day of year 122 to 300 (2) Non-pine conifers, LAI = 6 Aquifer recharge (PERCOL*) Specific (ETR*/ET0*): water deficit Daily GRAECO Integrated balance of water, carbon and growth with forestry management Maritime pine + undergrowth (LAI simulated from 7 years) Yield (trunk only) Aquifer recharge (PERCOL*) Water sufficiency (ETR*/ETM*) Hourly (reconstituted from daily) EVOLFOR Niche model (1) Beech (2) Holm oak (3) Biogeographical groups Specific: probability of presence (cf. GEOCLIMATOLOGY section) possible displacement of crops Period Table 2: comparison of features of the forest models used in CLIMATOR. The soil To the common soil (soil 1) were added four forest soils (tab. 3), whose characteristics were obtained from the INRA-Orleans database Infosol (cf. MODELS section). For the pine, soils 2 and 3 are similar in terms of available water reserve but are distinguished by their hydrodynamic properties at depth. Soil 1, with a high available water reserve*, is not currently found under maritime pines, but was used for comparison with the other crops studied in CLIMATOR. Soils Soil 1 brown Soil 2 brown Soil 3 fersiallitic Soil 4 brownish Soil 5 leached Depth (cm) AWR (mm) OM (%) BILJOU Coniferous forest with a high leaf area index Deciduous forest GRAECO Maritime pine forest with undergrowth Table 3: characteristics of soils used in the forest simulations. Three forestry rotations for the cultivated pine forest Long-term rotation for timber (PFL, 80 years); medium-term for unhewn timber (CFM, 50 years); or short, for biomass (PFS, 28 years). The commonest rotation used at present is the medium-term one (50 years). The way the simulations were carried out was specifically to represent a long rotation of a perennial crop: Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau 227

90 C 8 Forests the behaviour of the stand over the course of a year results from the seasonal weather conditions, the canopy structure (its LAI*) and the water status of the soil, resulting from the integration of its behaviour in preceding years; the behaviour of a forest canopy in a given year is calculated as the mean of the behaviour of n stands simulated in parallel, representing all the ages found in the course of the forest growth cycle; this way of operating reproduces the buffering effect of trees vis-à-vis year-to-year weather variation. C Ecophysiological identity card of forests in the context of climate change Advantages Metropolitan French forest species are C3* plants and because of this they respond to the increase in atmospheric CO 2 concentration, which stimulates their photosynthesis*. Due to their perennial nature, trees store atmospheric CO 2 more sustainably than herbaceous plants, grasses or crops. For certain species, like the oak, there is also an antitranspirant effect of CO 2 which confers better drought resistance. In general, forest stands deploy various strategies which make them tolerant of water shortage: a deep rooting system, a better extraction capacity than annual plants, and strong stomatal regulation. Also, the year-to-year behaviour of the canopies in response to water stress (e.g. reduction in the leaf area index) is a long-term advantage to assure the survival of the stand. Problems The high leaf area indices of the high-yielding* species result in high water consumption and considerable interception of precipitation by the foliage which slows down soil rewetting. They make the tree yields sensitive to the vagaries of the weather which, by reducing these high leaf area indices, cause yield losses which can persist even after the disappearance of the stress. The antitranspirant effect of the atmospheric CO 2 does not exist for certain species, such as maritime pine and beech. As regards adaptability*, the scope for action by the forester is small: the frequency of intervention is limited by the length of the rotations; there is little possibility of substituting species, due to lack of knowledge, regulations, industry requirements etc. How the canopies studied in CLIMATOR differ Their phenology*: the bare period for deciduous trees between October and April when no water is used; the low leaf area index of maritime pine (below 2.5), high in conifers other than pines (6-10), average to high in deciduous trees (4-10). Their capacity to intercept precipitation, which is higher for conifers at an equivalent leaf area index. D Impacts of climate change on the forest, its water balance, its yield and its distribution Water balance Intensity of water deficit Water stress* is characterised by the cumulative water deficit over the growing season of deciduous trees or over the whole year for conifers. This deficit is calculated as the sum of the differences between the water content for each day and 40% of the AWR (Granier et al., 1999). The intensity of the water deficit increases over the course of time, whatever the soil, the climatic projection* or the site* (fig. 2). This decline in the water balance of forests is particularly marked for coniferous forests (fig. 2 right) from the near future. Note than this decline is hardly noticeable for the mountain foothill station (Clermont-Theix). 228 Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau

91 C 8 Forests Standard deviation < >30 Standard deviation < >30 Mons Mons Rennes Versailles Mirecourt Rennes Versailles Mirecourt Colmar Colmar Dijon Dijon PERIOD A B C Bordeaux Lusignan Clermont St Étienne PERIOD A B C Bordeaux Lusignan Clermont St Étienne Toulouse Avignon Toulouse Avignon STRESS STRESS Figure 2 : changes in the intensity of the water deficit for a deciduous forest (left) and a coniferous forest (right) for the recent past (A), the near future (B) and the distant future (C). Aquifer recharge (cf. WATER section) Deciduous forest canopies return more water to the aquifers than canopies of evergreen conifers. A great deal of water is returned to the environment, mainly outside the growing season, particularly under deciduous forest. We know moreover that this drainage water is of good quality, because deposits, pesticides and other inputs have been intercepted, conferring a high environmental value to this type of canopy. Water consumption of forests and climate change The ETR*/ETM* ratio falls from the recent past to the distant future, whatever the water balance model or soil. The year-to-year weather variability and uncertainties due to the downscaling method are greater at Toulouse than at Colmar (fig. 3). Although the two models BILJOU and GRAECO do not calculate exactly the same variable (one calculates ETR/ETM and the other ETR/ET0), the trends in water consumption for the three periods are comparable, i.e. falling due to drought. Figure 3: mean, standard deviation and year-to-year range of ETR (LAI modelled) /ETM, simulated with GRAECO (G, red) for a plot of maritime pine and of ETR (LAI fixed) /ET0 simulated with BILJOU (B, blue) for a plot of coniferous trees with a high leaf area index, for the sites of Toulouse and Colmar for three time periods. Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau 229

92 C 8 Forests Effect of climate change on the yield of pines The yield of pines falls slightly in the near future and strongly in the distant future (tab. 4). Although these falls are significant, they are quite small in relative terms (4.6% in the near future, 11.3% in the distant future). The high significance of these results compared with those obtained for other non-perennial crops is due to the integrative capacity of tree growth. Much of the uncertainty associated with these results appears to be due to the method of climate downscaling (fig. 4). The weather-type method used by CERFACS leads to a smaller yield loss than the two other methods. NF-RP DF-RP Avignon (-7.4%) (-17%) Bordeaux (-3.0%) (-7.1%) Clermont (-0.9%) (-3.8%) Colmar (-3.5%) (-6.7%) Dijon (-3.7%) (-15%) Lusignan (-13%) (-22%) Mirecourt (+2.5%) (-4.3%) Mons (-6.0%) (-12%) Rennes (-4.3%) (-9.7%) Saint-Étienne (-5.4%) (-16%) Toulouse (-12%) (-23.5%) Versailles (-4.0%) (-7.4%) All sites (-4.6%) (-11%) Table 4: changes in yields of maritime pine (soil 3, climate WT) between the recent past (RP) and the near future (NF) or distant future (DF). The significance of the change compared with the year-to-year variability is denoted as follows: Bold (p < 0.01), Italic (p < 0.05), Normal (p < 0.10). Figure 4: mean, standard deviation and year-to-year yield range(tonnes dm/ ha/yr) of maritime pine simulated with GRAECO for the three time periods and the three climate downscaling methods (A: ANO, C: TT, Q: QQ) at Bordeaux for soil 3 and an average rotation. 230 Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau

93 C 8 Forests For a given forest rotation, the mean yield for each climatic period is mainly governed by the mean annual temperature; then by the annual rainfall (fig. 5). For forests, unlike certain crops, the temperature increase has a negative effect on yield by increasing the transpiration of the trees via the increase in the moisture deficit of the air. This stress is exacerbated by the concomitant reduction in rainfall, leading to more severe water stresses in the future climates. On balance for forests, of the climate changes expected, only the increase in CO 2 is likely to prove beneficial. Figure 5: change over the 12 sites ( A1B scenario, QQ downscaling, Soil 3) in the yield of maritime pine (CFM) as a function of the mean annual temperature (A) and of the mean annual rainfall (B). For each station, the three connected points represent the three periods recent past (RP*), near future (NF*) and distant future (DF*), where DF is represented by the triangle. Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau 231

94 C 8 Forests Changes in potential climatic zones for forest species The niche model was used here on groups of species and on two reference species (beech and holm oak). The use of different downscaling methods and climate models allowed us to evaluate the climatic uncertainty associated with the projected forestry zones. Mons Mons Rennes Versailles Mirecourt Dijon Colmar Rennes Versailles Mirecourt Dijon Colmar PERIOD A B C Bordeaux Lusignan Clermont St Étienne PERIOD A B C Bordeaux Lusignan Clermont St Étienne Toulouse Avignon Toulouse Avignon FEASIBILITY (%) FEASIBILITY (%) Figure 6: changes in the probability of the presence, or feasibility, (expressed as %) of beech (left) and holm oak (right); A1B scenario, WT downscaling; the soil type is not taken into account. Figure 6 illustrates the probability of the presence of beech (left) and holm oak (right) between the recent past and the near and distant futures. The simulations show a big change in the potential distribution of species. We see a progression towards the north of the Mediterranean and Aquitaine climatic components at the expense of the oceanic, continental and mountainous influences. These trends modify the boundaries of the potential areas of distribution of species. As an example, beech tends to regress in the near future, while conversely the holm oak expands, especially in the distant future, to the north of a line from Bordeaux to Saint-Étienne. Whatever the site or weather pattern, the probabilities of the presence of beech diminish in the near and distant future, in conformity with the results presented in figure 6, indicating a possible regression of the species. The signal is however strongly disturbed by climatic uncertainty (fig. 7 top). For the holm oak (fig. 7 bottom) in the near future, the probability of its presence increases in the southern half of the country (from Lusignan to Avignon). This increase becomes generalised in the distant future, corresponding to a movement to the north of the climatic limit of the holm oak. As for the beech, the climatic uncertainty is very high (except in the case of Avignon). For the most northerly sites the uncertainty is greatest, with probability differences varying from 0 to 0.9. For the beech, as for the holm oak, the effect of the A2 scenario is greater than that of the A1B scenario, whose effect is itself greater than that of B1. As regards downscaling methods, the QQ method has the strongest impact and the WT method the weakest. The method of anomalies (scenario A1B) has a smaller influence on the probability of the presence of both species than the QQ method applied to scenario B Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau

95 C 8 Forests Figure 7: differences of probabilities of the presence between periods A-B (blue) and A-C (red) calculated for Beech (above) and Holm oak (below) from three IPCC* scenarios and three downscaling methods (i.e. six weather patterns). The sites are classified in increasing order of probabilities from the NF period (values below the graphs). Figure 8 shows the possible change in the main biogeographical domains, i.e. the main flora/ climate balances such as they are seen by the species composition of French forests. Although it is not possible to assign a species to a domain unequivocally, it is possible to divide the country into five large domains: Mediterranean, south/atlantic, north/atlantic and north-east, plus a mountainous domain which can be subdivided more finely into three groups (Badeau et al., 2010). The results for the groups of species show a transition of landscapes towards more Mediterranean characteristics (an extension of red and orange colours) and a decline in north-easterly and mountainous characteristics (green and blue colours). As for the beech or the holm oak, the climatic uncertainty is very strong and particularly visible in the northern half of France. Depending on the downscaling methods, the sites of Colmar, Mons, Mirecourt and Dijon change either towards more Mediterranean characteristics (QQ) or, at the opposite extreme, towards more mountainous characteristics. For biogeographic groups as for the individual species, and with a constant scenario (A1B), the effect of downscaling methods is very strong. Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau 233

96 C 8 Forests Figure 8: probability of the occurrence of 7 biogeographical groups for the three studied periods. A1B scenario, QQ method (above); WT method (below). Group of Mediterranean species (red), south/atlantic (orange), north/atlantic group (yellow), north-east group (green), mountainous group (3 shades of blue). 234 Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau

97 C 8 Forests What adaptations of practices can be envisaged? In view of the results of the study, several ways can be suggested, mainly aimed at forms of forestry using less water: a preference for deciduous trees rather than conifers; cultural practices aimed at lower leaf area indices. One might also consider planting exotic species or choosing genotypes able to maximise yield and drought resistance, or even opt for mixtures of new species and natural regeneration. What you need to remember 3 Forest, both deciduous and coniferous, can be affected by climate change in the near future and very significantly (anywhere in France) in the distant future. Soil and atmospheric drying are the main constraints. Unlike all the other cropping systems, it is not possible to improve the water supply by irrigation. Temperature effects on phenology do not lead to adequate escape* strategies, unlike annual crops. Temperature acts adversely via an increase in atmospheric dryness. The sole beneficial effect is that of the increase in atmospheric CO 2 concentration which improves photosynthesis, but without compensating for the negative effects of the increase in water stress. The biggest source of uncertainty* is associated with the weather patterns (emission scenarios* and downscaling methods). Soil properties (depth, AWR) do little to alleviate the impact of the reduction in rainfall. Forests, especially if deciduous, are among the plant covers which return the most water to the aquifers, even if rainfall is reduced by climate change. This return takes place mainly in the winter. The three modelling approaches used (water balance, mechanistic growth model, niche model) all result in the same conclusions. What needs further study 3 Our modelling tools could be improved in various ways. For example it would be desirable to introduce mortality thresholds in the forest models, related to vulnerability to extreme weather events* and to the risks of withering associated with weather/pest interactions. One can also consider modelling the seasonal and year-to-year dynamics of forest LAI in response to drought. Finally, the extension of the niche model approach, currently applied to the presence/absence of species, to growth potential (radial growth for example) would allow more subtle responses. Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau 235

98 C 8 Forests To find out more Badeau V., Dupouey J.-L., Cluzeau C., Drapier J., Le Bas C. (2010) - Climate change and the biogeography of French tree species: first results and perspectives. In Forests, Carbon Cycle and Climate Change, Denis Loustau (Ed.), Edition Quae, Versailles, ISBN Bréda N., Granier A., Huc R., Dreyer E. (2006) - Temperate forest trees and stands under severe drought: a review of ecophysiological responses, adaptation processes and long-term consequences. Annals of Forest Sciences, 63, 6, DaviI H., Dufrene E., Francois C., Le Maire G., Loustau D., Bosc A., Rambal S., Granier A., Moors E. (2006) - Sensitivity of water and carbon fluxes to climate changes from 1960 to 2100 in European forest ecosystems. Agricultural and Forest Meteorology 141(1): Granier A., Bréda N., Biron P., Viville S. (1999) - A lumped water balance model to evaluate duration and intensity of drought constraints in forest stands. Ecological Modelling, 116, Loustau D. (éditeur) (2010) - Forests, Carbon Cycle and Climate Change. Denis Loustau (Ed.), Edition Quae, Versailles, ISBN Loustau D., Bosc A., Colin A., Ogee J., Hendrick D., Francois C., Dufrene E., Deque M., Cloppet E., Arrouays D., Le Bas C., Saby N., Pignard G., Hamza N., Granier A., Bréda N., Ciais P., Viovy N., Delage J. (2005) - Modelling the climate change effects on the potential production of French plains forests at the sub regional level. Tree Physiology, 25, Loustau D., Pluviaud F., Bosc A., Porte A., Berbigier P., Deque M., Perarnaud V. (2001) - Impact of a regional 2xCO 2 climate scenario on the water balance, carbon balance and primary production of maritime Pine in Southwestern France. Models for the Sustainable Management of Plantation Forests. Bordeaux, European Cultivated Forest Institute (EFI sub division) EFI Proceedings numéro 41D. Porté A. (1999) - Modélisation des effets du bilan hydrique sur la production primaire et la croissance d un couvert de pin maritime (Pinus pinaster Ait.) en lande humide. Ph.D. Thesis, Univ. Paris XI, Orsay, France, 160 p. 236 Green Book The crops Forests Nathalie Breda, Alexandre Bosc, Vincent Badeau

99 Analysis of an example of organic arable farming Nadine Brisson C9 Organic farming A Some key figures for organic farming in France West Center-North Types of organic farms Vines, fruits and vegetables Arable crops and vines Arable crops and vegetables Arable crops Livestock Area (ha) % UAA South-West North-East Center-East South-East Figure 1: organic farming in the CLIMATOR zones showing the main products of organic farms and the areas concerned (in ha and in % of the UAA). With 2.12% of the UAA devoted to organic farming (2008 data), France is well below the European average, which is in the range of 4-5%, similar to the fractions which our German and Spanish neighbours devote to it. The best examples are Italy with 9% and Austria with 13%. After a period of stagnation between 2002 and 2006, areas are again increasing. As figure 1 shows, there is a big disparity between regions, with fruit-growing, wine-growing and livestock regions in the lead. It illustrates a big difference between crops. For example 5.9% of the fruit-growing areas and 3.3% of the winegrowing areas are organic, whereas for the arable crops, this figure is only 1%; it is 2.8% for forage crops. We also find that organic cereals are often grown on mixed cropping-livestock farms. Green Book The crops Organic farming Nadine Brisson 237

100 C 9 Organic farming B Simulation protocol in CLIMATOR We only studied one organic system, based on a four-year rotation* composed of peas, wheat and a forage grass grown for two years, fertilised with livestock manures. We assumed that this cropping system* was coupled with livestock, and that some of the residues are exported. Hygiene problems (weeds and diseases) were not considered. Sites System Management Soils Variety Model Climate 12 sites in metropolitan France including one at high altitude a single rotation: peas-wheat-fescue-fescue rainfed, 40t/ha of manure in two applications during the course of the rotation; wheat straw exported; fescue sown in autumn after the wheat harvest Soil 1: deep silty soil with a high AWR (226mm) and average fertility Soil 2: rendzina with a low AWR (104mm) and low fertility Soil 3: leached hydromorphic soil with a high AWR (317mm) and high fertility Bacarra (pea), Soissons (wheat), tall fescue STICS SRES: A1B*, ARPÈGE, downscaling*: WT*, QQ *and ANO* C Will climate change alter the relationship between organic and conventional farming in terms of yield? Wheat If we analyse wheat yields, the things which differentiate the organic system from other arable* rotation systems are the previous crop (peas in our case) and the absence of mineral fertilisation, replaced by organic manuring twice every four years. Figure 2: change in the yield of wheat in rotation, as means and variability for the periods of the recent past (RP*), near future (NF*) and distant future (DF*) for two sites: Versailles (left) and Toulouse (right). The abbreviations ORG, MWRW and SWSgW refer to the three cereal rotations studied in CLIMATOR. (cf. AGRICULTURE section). Soil 1, WT downscaling method. Figure 2 shows that, for wheat yield, organic management does not differ significantly from other methods of management in rotation (MWRW, SWSgW) with identical sowing dates. It is interesting to note that organic wheat is not favoured by climate change*, but, as for other cropping systems, it will experience greater year-to-year variability in the future, with bigger differences between low and high yields. Hence our results indicate that the occurrence of yields exceeding 6 t ha -1 will increase, which is explained by better results from organic fertilisation due to warming. On poorer soils, drier or wetter, the wheat crop, on average, exploits the advantages offered by climate change less well, whichever the system. We observe however that on these poor soils, the yield variability increases in the DF for the organic crop, whereas it tends to diminish for the conventional crop (fig. 3). 238 Green Book The crops Organic farming Nadine Brisson

101 C 9 Organic farming Figure 3: comparative changes in organic and conventional wheat yields for 3 soil types at Versailles and the WT downscaling method. Here the conventional wheat crop is a wheat monoculture, sown early. Fescue In terms of forage production, the difference between the two systems (compared for the same situation, i.e. the yield in the year following sowing) is not significant. On the forage sites of Lusignan and Mirecourt (fig. 4), the small advantage of the conventional system tends to disappear towards the end of the century (DF). Figure 4: comparative changes in organic and conventional forage yields for soil 1 at Lusignan and Mirecourt with the WT downscaling method. At other sites, like Rennes or Toulouse, the increase in drought in the DF weakens the organic system whose yield in dry years is very low (fig. 5). Figure 5: change in organic forage yield at Rennes for soil 1 and the WT downscaling method. Green Book The crops Organic farming Nadine Brisson 239

102 C 9 Organic farming D Improvement in the quality of organic wheat In view of the slight tendency to an increase in the nitrogen absorption by the plants due to the stimulation of mineralisation* by the temperature (fig. 6), one can expect an increase in the protein content of organic wheat. Figure 6: change in the absorption of nitrogen of organic wheat at Toulouse and Versailles for soil 1 and the WT downscaling method. This increase is indicated more regularly at the northern sites than at those in the south (fig. 7). In fact in the south (Toulouse), the risks of low N contents increase because of the drought which hinders nitrogen absorption. 240 Green Book The crops Organic farming Nadine Brisson

103 C 9 Organic farming Figure 7: change in protein content of organic wheat for various sites, soil 1 and WT downscaling method. E Organic farming and the environment The presence of legumes in the rotation and the massive application of slurry favour the leaching of nitrates: this is one problem with organic systems which, from this point of view, are fully comparable with conventional intensive rotation farming (MWRW) as the maps in figure 8 show. However the reduction in rainfall reduces the severity of this problem in the future, particularly at the sites in the west (Lusignan, Rennes, Bordeaux). Standard deviation (kg N-NO3 ha-1) < > Rennes Bordeaux Lusignan Toulouse Mons Versailles St-Étienne 8 92 Mirecourt Clermont 32 Dijon Avignon 38 Colmar Standard deviation (kg N-NO3 ha-1) < > Rennes Bordeaux Lusignan Mons Versailles Mirecourt Toulouse St-Étienne Clermont 37 Dijon Avignon 40 Colmar Lixiviation (kg N-NO3 ha-1) Lixiviation (kg N-NO3 ha-1) Figure 8: comparison of the change in nitrate leaching under the organic rotation (left) and the conventional rotation (right) (soil 1 and WT downscaling method). Regarding carbon storage, even when exporting part of the livestock excreta often associated with organic systems, the system remains very worthwhile (cf. ORGANIC MATTER section). Green Book The crops Organic farming Nadine Brisson 241

104 C 9 Organic farming F Regional differences and uncertainties NF-RP DF-RP Avignon Bordeaux Clermont Colmar Dijon Lusignan Mirecourt Mons Rennes Saint-Étienne Toulouse Versailles All sites Table 1: change in yields of organic wheat, t ha -1 (soil1, climate WT). The significance of the change in comparison with the year-to-year variability is denoted as follows: Bold (p < 0.01), Italics (p < 0.05), struck through (non significant). Although the change in organic wheat yields is on average positive everywhere, table 1 shows that it is only rarely significant in view of the year-to-year variability, even at sites, like Versailles or Lusignan, where the average increase is more than 1t ha -1. This is due to the great sensitivity of the system to weather variability from one year to another. We see that certain cold-climate sites (Colmar, Dijon, Clermont) are an exception and show a significant increase in organic wheat yields. It is also of interest to point out that the trend into the DF continues that into the NF without change. 242 Green Book The crops Organic farming Nadine Brisson

105 C 9 Organic farming Figure 9: influence of the climate downscaling method combined with the soil type on the results for organic wheat yields at Mons. If we focus on the Mons site (fig. 9), it appears that this increase is not confirmed by all the downscaling methods. Thus the QQ method, which by its nature produces more variable weather scenarios, simulates much more risky organic wheat production than the WT method. When one assumes a poorer soil, the responses are in better agreement for the WT and QQ methods and indicate a non-significant change, whereas the method of anomalies, which reproduces for the future the variability of the past, gives a similar result as for the deep silty soil. Green Book The crops Organic farming Nadine Brisson 243