Abstract
In recent decades European silage maize production has extended northwards, into Scandinavia, and the importance of maize in fodder production has increased substantially. For the northward expansion of maize production it is of interest to evaluate both the role of climate change that has occurred already, and scenarios for possible future climate change. The aim of this study was to assess for Swedish climatic conditions, the annual variation in silage maize yield and quality (dry weight and starch contents) of cultivars currently grown in Germany.
The MAISPROQ simulation model currently used in German maize production was applied to evaluate the effects of (i) cultivar differences (four cultivars; four sites; 2003–2009), (ii) intra-regional variation among ten sites representing three regions (two cultivars; 2003–2009), and (iii) climatic variability among two historical periods during 1961–2009 and three future periods during 2011–2100 using A2-emission climate scenarios and the Delta-method (two cultivars; four sites).
Forage quality assessments strongly influenced the assessments of harvest time and thereby the yield. Changes in simulated yield of the tested cultivars were high for the past climate, but relatively small under future climatic conditions due to earlier harvest caused by improved forage quality. By the end of the 21st century an appropriate fodder quality would be achieved every year in the south of Sweden, whereas in the middle of Sweden (60°N) about 30% of the years would not be successful, even for the earliest cultivar. In the east, increased water stress counteracted the positive effect of a prolonged growing season. It was concluded that adaptation of field experiments to model calibration requirements remains to be done, in order to enable extrapolation of observations from Swedish field trials to a changing future climate.
Introduction
Climate change will have a large impact on future land use and will require an adaptation of agronomic strategies (Easterling Citation1996, Moriondo et al. Citation2010), including the selection of new crop species and varieties. In Europe the area used for the production of silage maize has extended northwards into Scandinavia during the last decades (Fogelfors et al. Citation2009). In Sweden, the importance of maize in fodder production has substantially increased during the last decade, as indicated by an expansion of the maize acreage from 2000 ha in 2002 to 17000 ha in 2009 (Jordbruksverket Citation2010). This increase may be related to weather conditions becoming possibly more favourable for high yield and quality, as well as to progress in plant breeding, for instance in terms of chilling tolerance (Leipner and Stamp Citation2009). Temperature has increased (SOU 2007) allowing for faster development and achievement of the required quality at the time of harvest. Water availability has also changed, although the extent of its influence on crop yield is less clear (Eckersten et al. Citation2010). For the expansion of maize production in Sweden it is of interest to evaluate both the role of climate change that already has occurred and of scenarios for possible climate changes in the future. Quantitative evaluations of climate effects on the yield and quality are a basis for the choice of new maize cultivars as well as for evaluating the replacement of traditional fodder-production systems, such as leys, and for assessing future fertilizer needs and related environmental impacts.
A traditional approach to evaluate the role of climate in the introduction and choice of new species or cultivars is to perform field trials. However, the within-season variation in climate between years is high and very many experimental years are required for a representative sample of all combinations of weather conditions that may occur during the vegetation period within a ‘normal’ climate. In addition, field trials are usually located in areas that have favourable growing conditions, making the applied spatial scale a limiting factor for the evaluation of climate-change impact. Extrapolating field experimental results to future environmental conditions by using the ‘analogy’ model (e.g. analyses of variance) approach, has many theoretical and practical limitations, as only a few of the possible future conditions can be investigated in such an experiment. Instead, models based on growth processes are, theoretically, more appropriate for predicting the crop's yield performance in a changing environment and for scenario analyses of changing climatic conditions. The relationship between weather and yield has been investigated intensively for a range of crops and regions with models of different complexity (Lobell and Field, Citation2007, Tubiello et al. Citation2007, Challinor et al. Citation2009, Eckersten et al. Citation2010). In the case of maize, simple model approaches, such as thermal time indices based on accumulated temperature sums above certain threshold temperatures, have been used for maturity and harvest time predictions, for instance in France (AGPM Citation2000), the United States and Canada (Phipps et al. Citation1974, Gurr & Kimber Citation1983, Smith et al. 1982). These types of models have also been used for calculating the dry-matter production of silage maize (Phipps et al. Citation1974, Smith et al. Citation1982, Gurr and Kimber Citation1983, Bootsma et al. Citation2005), and are alternatives to more process-based model approaches that have been adapted and validated for various climatic regions (CERES Maize, Jones and Kiniry, Citation1986, MAIZE, Muchow et al. Citation1990, STICS, Brisson et al. Citation1998, WOFOST, Boogard et al. Citation1998, Yang et al. Citation2004, MAISPROQ, Herrmann et al. Citation2005b, SPN, Bleken et al. Citation2009). The more complex mechanistic crop growth models have the advantage of assessing the effects of non-linear combinations of changes in temperature and precipitation; however, many of them have the disadvantage of requiring a plethora of input data, which usually are not available from field trials. Thus these models are not testable for a wide range of environmental conditions (Zalud and Dubrovsky Citation2002). Moreover, they often lack cultivar-specific parameterization. In-between solutions, with less-complex models accounting for only some basic functioning of the crop, such as water availability, temperature and solar radiation response, and plant development, have been used successfully in prognoses of maize yield and DM content (Drury and Tau Citation1995, Bootsma et al. Citation2005, Pearson et al. Citation2008), of which the MAISPROQ model (Herrmann et al. Citation2005b) was used in the current study. The model has been parameterized and validated for several cultivars used in Germany. In 2005, it was implemented nationwide as a regional harvest-time prognosis tool for practical use (Herrmann et al. Citation2006).
The aim of this study was to assess maize yield and forage quality (dry matter and starch contents) of four cultivars, which are grown in Germany, for Swedish climatic conditions. This includes the quantification of variability in yield and quality performance as influenced by cultivars under current local climatic conditions, as well as by climate change in the past and in the future. Since there are not enough data from Swedish field trials available for model calibration, the model application to Swedish climatic conditions may be regarded as an assessment approach of how cultivars used in Germany might perform in Sweden. The expected uncertainties related to the extrapolation from the German to the Swedish climate, and from climate of a specific calibration period to future climate change, are discussed.
Materials and methods
Modelling strategy
The MAISPROQ model was applied to conduct three simulation studies. First, the impact of cultivar on yield and forage quality was evaluated for four cultivars grown under current climatic conditions (2003–2009) at four sites (Kristianstad, Visby, Skara, Uppsala), representing three different regions of southern and central Sweden. The selected varieties () cover a range of maturity classes from early to mid-late, including different maturity types (normal, dry-down and stay-green) relevant for Sweden. The earliness of the varieties is characterized by the German maturity-rating system, which includes a separate evaluation of the maturation behaviour for the whole crop (S) and for grain (K). The S and K values are estimated under German conditions, and low values characterize early maturity and high values late maturity. For details concerning the locations see . The second simulation study focused on the intra-regional variation of climatic conditions and its impact on crop performance. The simulations were conducted for the early cultivar ‘Janna’ grown under current climatic conditions (2003–2009) at ten sites representing three regions. The third simulation study included historical weather data (1960–2009) and climate scenarios (2011–2100) to quantify the effect of past and future climate change on performance of an early cultivar grown at four locations, i.e. Lund, Visby, Skara and Uppsala. Results for a later maturing cultivar are given in the appendix for comparison.
Table I. Characterization of the simulated German silage maize cultivars. For explanation of S and K, see the text.
Table II. Positions of locations (weather stations) used for the simulations, and the simulation periods and applications in which the locations are used. Locations are sorted into three regions: Southern Götaland, Northern Götaland and Svealand, respectively.
Model description
Simulations of silage-maize yield and forage quality were conducted using the MAISPROQ model (Herrmann et al. Citation2005b, Kruse et al. Citation2008) which consists of two interacting modules, one for simulating growth and harvestable biomass and the other for calculating forage quality properties for dry matter and starch. The model was originally developed for leys and applied to different ley types in several studies (Angus et al. Citation1980, Torssell et al. Citation1982, Kornher and Torssell Citation1983, Fagerberg and Torssell Citation1995), but later modified to simulate silage maize. Since 2006, the MAISPROQ model has been used for predicting regional harvest time of silage maize for the whole of Germany (Herrmann et al. Citation2006).
Maize dry-matter accumulation was assumed to start when the accumulated sum of daily temperatures above 6 °C since 1 January has achieved 153 d °C (Herrmann et al. Citation2005b). The daily growth (dW t /dt) of above-ground biomass (W t−1) is basically proportional to the maximum relative growth rate (rs), reduced by an age factor (AGE) and a weather-dependent growth index (GI):
The weather-dependent growth index (GI) is the product of indices for temperature, incident solar radiation and the plant-available soil water. The plant-available water is simulated by means of a single soil layer budget increased by precipitation P (mm), and decreased by actual evapotranspiration. The water index is obtained as the ratio of actual to potential evapotranspiration.
The quality sub-model for predicting the contents (%) of dry matter and starch is based on the assumption that these forage-quality parameters (Q) increase gradually on a daily time-step basis during the growth period, from a predefined initial level (QMin) towards a predefined maximum level (QMax) as influenced by environmental conditions (fQt ). For a given day t the quality trait is obtained from:
Model application
Parameter values (App. a) of the four maize cultivars were those obtained by Herrmann et al. (Citation2005b) when calibrating the model to German field trials fertilized at an optimal level in terms of crop demand. In spring, at start of simulations, the soil water content in most agricultural soils is high and roughly corresponds to the soil-water storage capacity in the uppermost 0–100 cm layer of the soil profile. Based on a conversion of values given by Wösten et al. (Citation1998) to the Swedish soil types of interest for this study (primarily ‘mellanlera’ with a clay content of 25–40%; Wiklert et al. Citation1983; computations not shown), we chose an initial plant-available soil water of 180 mm for all simulations. The initial amount of biomass was taken from Herrmann et al. (Citation2005b); App. a).
The model driving variables, including daily weather data on air temperature, precipitation, solar radiation and potential evapotranspiration, were obtained from the national network of weather stations (both observed and interpolated values; ) as provided by the Swedish Meteorological and Hydrological Institute (SMHI; FFE 2009), and for the Uppsala site from the Ultuna weather station (SLU Citation2009). In the case of missing values, for a period of a few days, interpolated values were used. Average climate data are presented in App. a. The excess rain of Lund compared with the other sites was mainly due to high precipitation during winter, and the lower annual precipitation at Visby was mainly due to drier summers. The monthly average temperatures were basically the same for all locations during summer, whereas Uppsala was considerably colder during winter, and Lund was warmer during spring (App. a).
Daily weather data for future climate-change impact assessments were obtained applying the Delta method (Quilbé et al. Citation2008). In a first step, the predicted monthly changes of the chosen scenarios were added (temperature) or multiplied (precipitation) to the observed daily climate during 1961–1990. Daily changes were then derived by linear interpolation between the monthly changes for the scenarios, assumed to occur on the 15th day of each month (App. a). For potential evapotranspiration (Evt) an empirical relation was derived between daily Evt and air temperature for SMHI data in Uppsala 1961–1987 (Evt=0.003 T2+0.022 T − 0.0354; T > 0; R2=0.6). The relative change of Evt(T) due to the temperature change of the scenario was applied to the daily values of Evt 1961–1990. The solar radiation was assumed to be unchanged relative to that for 1961–1990. The climate-change scenarios for differences compared with 1961–1990, for the socioeconomic scenario A2 (IPCC Citation2000) and climate models Echam4/RCA3, were taken from maps produced by the Rossby-Centre at SMHI (2007) for periods 2011–2040, 2041–2070 and 2071–2100, respectively (SOU Citation2007). The statistical analyses were made with the JMP program package (SAS Institute 2007). The simulation studies were conducted using the simulation routine FOSIM, which runs MAISPROQ with a defined harvest strategy, i.e. a target dry-matter content of 340 g kg−1 (34%). Latest admissible harvest date was fixed to 31 October.
Results
Cultivar impact
Average simulated yield during 2003–2009 varied between 10.8 and 13.3 t DM ha−1 among cultivars and locations included in the first simulation study (). Later-maturing cultivars usually achieved higher yield compared to earlier ones under favourable environmental conditions. In the current simulation study, however, the early dry-down cultivar Justina was, in almost all cases, among the highest yielding, indicating that later-maturing Noveta and Calimera could not realize their full yield potential. The variation in DM content among years was smaller for the Kristianstad site than for the locations representing the other regions, as indicated by a higher proportion of years with DM content achieving 34% (340 g kg−1) before 31 October. The 34% level was achieved in about half of the years for Janna and Justina, and never for Calimera. For the other sites the target value was not achieved. A dry matter content of 28%, which may be acceptable as the lower limit for ensuring successful ensiling, was attained for almost all cultivars and all years in Kristianstad, and frequently for Janna and Justina in Visby and Uppsala. Starch content was always highest for Janna and lowest for Justina and Calimera, and the relative variation in starch content among cultivars (−45 to +70%) was substantially higher than for yield and DM content.
Table III. Simulated maize harvest date, dry-matter yield, dry-matter content and starch content for different cultivars at selected locations during the 2003–2009 period. Values are provided as absolute values and expressed as the relative value to the mean of all cultivars. A year is classified as successful when a dry-matter content of 340 g kg−1 is achieved; [ ] refers to 280 g kg−1.
Local variations
The variation in yield within Southern Götaland was small in comparison with the corresponding variation in the number of successful years and in starch content (). An exception was Halmstad, which had the highest yield. The yield for Lund was lower than for Halmstad due to earlier harvest. The other regions were characterized by lower yields, and the target DM content of 34% was not achieved in any single year. For the mid-early cultivar Calimera the same ranking of locations was observed as for Janna, but a dry matter content of 34% was only achieved for Lund (App. a).
Table IV. Simulated maize sowing and harvest dates, dry-matter yield, dry-matter content and starch content for different locations for the early cultivar Janna. Average for 2003–2009. A year is classified as successful when a dry-matter content of 340 g kg−1 is achieved before 31 October.
Climate variability assessments
Substantial changes in climatic conditions have been observed in Sweden during the last century (SOU Citation2007), which are reflected in the long-term simulations over the 1961–2009 period, as exemplified for the southernmost location of Lund and northernmost of Uppsala (). Yield and contents of DM and starch were generally higher at Lund. For both locations yield and quality traits increased significantly over time, but yield increase was larger for Uppsala; the slope of the regression line was about 0.1 t DM ha−1 year−1 (A). The effects on harvest date were somewhat different. Whereas the harvest at Uppsala never took place before 31 October, maize achieved silage maturity much earlier at Lund (D). In summary, climatic changes caused an improvement of maize performance for both locations. Yet, variations of yield and quality among years were as high as the average yield and quality increase that had been progressively obtained for the 49-year period.
Figure 1. Simulated maize production and harvest date for cultivar Janna from 1961 to 2009 for Lund (circles and upper regression line) and Uppsala (crosses and lower regression line) versus year. The probability of slope (pSlope) of the regression lines being different from zero is larger than 99.9% unless specified. Year in the equation is the number of year since 1960.
In general, the long-term assessments for future climate-change scenarios basically show a positive impact on yield from 2011 and onwards (). For the most southern and eastern locations, Lund and Visby, however, the predicted average yield in the 2071–2100 period is about 1 t DM ha−1 lower than for the 2041–2070 period. For Skara and Uppsala a slight yield decline is predicted during the next 30 years, after which yield then increases again. For all locations, the risk of not achieving the target DM content of 34% is reduced over time. In the Lund region the DM content currently reaches the target in 70% of the years, and will remain at that level during the coming 30 years, but thereafter increases towards 100% by the end of the 21st century. At Visby the proportion of successful years is currently much lower, but is predicted to increase substantially for the coming 30 years, while at Skara and Uppsala silage maize production from the cultivar Janna will still carry a high risk at the end of the 21st century. Consequently, by the end of this century harvest date will be brought forward by about 2 months for southern locations, but only by 1 month for northern locations. Furthermore, starch contents will increase, especially for the more northern locations of Skara and Uppsala, whereas at Lund and Visby, where starch content is already relatively high, future increases will be less pronounced. The effect of 5 years of missing values during the period 1991–2009 was a small underestimation of maize fodder quality for Uppsala, indicating that the missing years of Lund, Visby and Skara had a minor influence on the results (). A dry matter content of 28%, instead of 34%, will be achieved in all years in Lund, almost all years in Visby, and in about 50% of the years in Skara and Uppsala, during the coming 30 years. The response of the mid-early cultivar Calimera is similar to that of Janna. For southern locations, silage maturity will be achieved in all years at the end of this century. For northern locations, however, harvest date occurs only 2–3 weeks earlier, due to higher temperature demand compared with the earlier cultivar Janna (App. a).
Table V. Simulated maize dry-matter yield, dry-matter content and starch content for different years at selected locations for the early cultivar Janna, for different periods between 1961 and 2100. A year is classified as successful when DM content becomes 340 g kg−1 before 31 October; [ ] refers to 280 g kg−1. For information concerning missing years see Table II.
The simulation studies conducted so far have assumed sowing date to be dependent only on temperature sum. Accordingly, we estimated advanced sowing dates for the historic weather data as well as for future scenarios. Precipitation is predicted to have a different pattern in the future, with a shift towards more precipitation in winter and March, and also later in April (App. a; SOU Citation2007), and this may hamper earlier sowing. Therefore, an additional simulation study was conducted, in which sowing dates were assumed to occur on average about 2 weeks later. A comparison of the ‘sowing-delayed’ () with the ‘early-sowing’ () simulations, for the early variety Janna, revealed a postponement of harvest date by about 2 weeks. For DM yield, only slight differences were detected, whereas there was a considerably higher risk of not achieving silage maturity at the Skara and Uppsala locations, suggesting that early sowing date is an important factor for harvesting silage maize with acceptable quality under conditions predicted for climate change.
Table VI. Maize dry-matter yield, dry-matter content and starch content for the early cultivar Janna grown at four locations simulated for different periods between 2011 and 2100. A year is classified as successful when a dry-matter content of 340 g kg−1 is achieved before 31 October; [ ] refers to 280 g kg−1. Simulations are similar to Table V except that sowing date is assumed to occur on average about 2 weeks later. For information concerning missing years see Table II.
The corresponding delay in sowing date for the mid-early cultivar Calimera revealed an even higher risk of harvesting at an immature developmental stage. Again, this effect was more pronounced for the northern locations, where, by the end of this century, even the lower DM content of 28% can be achieved in only one year in three (App. a).
Discussion
Effects of climate and weather variability
The simulation studies document a considerable change in silage maize yield and forage quality in southern Sweden caused by the past climate change, and that trend is likely to continue under conditions of future climate change, although at a lower rate for the current cultivars. The major influence of a warmer climate is to decrease the annual variation in quality and the risk of not achieving the forage quality required. The simulations resulted in a considerable yield increase over the 1961 to 2100 period, at most being about 0.1 t DM ha−1 year−1 during a 30-year period. This is consistent with climate-change scenario simulations conducted for Europe and the United States for grain maize (Olesen et al. Citation2007, Torriani et al. Citation2007, Gonzalez-Camacho et al. Citation2008, Kucharik and Serbin Citation2008), and silage maize (Wolf and Van Diepen Citation1994, Davies et al. Citation1996, Holden and Brereton Citation2003). For silage maize quality, however, no comparable studies were available. In our study the quality predictions strongly influenced the yield predictions.
The spatial and temporal variation of climate-change patterns and the crops’ complex response to weather (Lobell and Field Citation2007) explain the simulated variability in silage maize performance detected among different Swedish locations. Yield mostly increases over time but also declines, especially in Visby by the end of the 21st century (), due to simulated increase of water shortage. Precipitation in Visby was on average only about 1 mm per day from April to August, and in total for these months about 100 mm (38%) lower than in Lund (App. a). Our findings are thus only partly similar to the predictions by Wolf and van Diepen (Citation1994), who reported potential production of silage maize to increase in northern Europe.
Various agronomic adaptation strategies have been suggested to reduce negative effects or to exploit potential positive effects, one of them being the change of sowing date (Meza and Silva Citation2009). A shift towards earlier sowing of maize has been observed across many regions over the past decades (Chmielewski et al. Citation2004, IPCC Citation2007). For the future, our simulations estimated that the effect of temperature change on sowing date will lead to an earlier harvest. This reflects uncertainty in the predictions related to shifts in rainfall pattern from summer to winter, which might reduce soil workability and trafficability in spring and thus counteract sowing date advancement (Cooper et al. Citation1997), but which are not considered by the MAISPROQ model.
Apart from earlier sowing date, climate change is assumed to prolong growth in autumn. When allowing maize growth to continue until November instead of harvesting at 34% DM content on 31 October at the latest, our simulations gave yield improvements up to 3.9 t ha−1 for cultivar Calimera in Visby. Growing degree days (GDD) above 8 °C increase by on average 340 GDD from average harvest time, determined by the quality (7 September; a), to end of November, of which 330 GDD occur already before end of October (data not shown). Thus, the advancing harvest date due to advancing quality establishment accounts for almost 40% decrease in potential yield in the east, 25% in the south, and only about 5% in the north. Choosing other high-yielding cultivars might potentially give stronger yield response to climate change; however, this has not yet been examined for the Swedish climate.
In northern Europe, temperature is the most limiting growth factor, as shown for instance for winter wheat yields in Uppsala, which correlated with winter temperature during 1968–1996 (Eckersten et al. Citation2010). For maize, the length of the winter season is the main limiting factor. The predicted increased temperature under climate change is, however, expected to be associated with reduced water availability, which in our study significantly limits the expected increase in maize yield, especially in Visby in eastern Sweden. This negative influence of dry summers in Visby has also been found for winter wheat yield that was negatively correlated with high June temperatures and simulated low water availability for the 1965–1996 period (Eckersten et al. Citation2010).
The uncertainty of these projections is influenced by the uncertainty of the climate-change inputs used, and especially precipitation scenarios are of interest to evaluate for alternative greenhouse gaseous emissions scenarios and climate models. Differences between the high and low IPCC emission scenarios, A2 and B2, start to become systematic after c. 2050. For the 2071–2100 period the accumulated decrease in precipitation during June–August is about 40 mm for the A2 scenario used for southern Sweden (App. a), whereas it is about 15 mm for the B2 scenario (Eckersten et al. Citation2008), which would stimulate yield and hasten quality development compared with A2. However, this effect would be counteracted by the lower temperature increase projected in B2. The main difference in precipitation scenarios between the Echam4 climate model (used here) and the HADAM3H climate model are that the latter projects dryer conditions from September to November (Eckersten et al. Citation2008). Using the scenarios of the HADAM3H climate model would then delay the simulated harvest time for the more northern locations, whereas for the southern locations the influence would be less pronounced since harvest was predicted to occur already in September by the end of the 21st century ().
Model validity
A basic requisite for relying on model predictions is that the model's predictability has been tested. Unfortunately, data availability did not allow for thoroughly validating model predictions for German cultivars under Swedish conditions. Available yield and quality data collected in several Swedish field experiments (FFE Citation2009, Hushållningssällskapen Citation2009) refer to different cultivars and do not meet the model requirements. Nevertheless, a preliminary comparison between simulations and field data, where the observed data were recorded for cultivars different from those used in the simulations, revealed a better agreement for yield than for quality, although not statistically confirmed (data not shown; observed data varied between 11–17 t yield ha−1, 27–43% DM content, and 23–36% starch content during 2006–2009).
It is generally assumed that applying a model that has been developed and calibrated for a given region, to another region characterized by different climatic conditions, may carry a risk of biased yield estimations. A major part of this problem is often related to changed soil conditions, making the extrapolation in space more uncertain for sub-optimally fertilized crops (Eckersten et al. Citation2007). However, the current study concerns fertilized crops and the MAISPROQ model has successfully been applied to the whole of Germany (Herrmann et al. Citation2006). Hence, the extrapolation to Swedish climatic conditions would be reasonable, at least for the southern locations (c. 55.5°N) and possibly also for the northern locations (c. 60°N).
The extrapolation is, to an unknown degree, justified by the semi-mechanistic character of the model. However, some environmental factors, which are not accounted for in the model since they are of minor importance when applying the model for German conditions, may become relevant for Swedish conditions, such as photoperiod and the occurrence of autumn frost. Day length at midsummer ranges from c. 18.4 hours in Uppsala to 17.2 hours in Lund, compared with 17.0 hours in northern and 15.8 hours in southern Germany. Most modern temperate maize cultivars are assumed to be day-length neutral. Therefore our model approach as concerns the photoperiod, would be reasonable. If varieties were photoperiod-sensitive a delayed maturation may be expected (Bonhomme et al. Citation1994, Birch et al. Citation1998, Miller et al. Citation2008). For such varieties our model simulations would overestimate the advance of harvest date for the northern locations (Uppsala) but only marginally in Lund. Unfortunately, there is no scientifically based information available on the photoperiodic sensitivity of the German cultivars used in this study. However, the current model approach has been proved valid for the whole of Germany, i.e. from 48 to 54.5°N (Herrmann et al. Citation2006).
The effects of frost in autumn are not considered in the model but the risk of low temperatures during August, September and October was estimated to decrease considerably in the future. In Visby during the reference period 1961–1990, the fraction of years during August to October with mean temperatures below 5 °C was 87%, and the corresponding values for 4, 3, 2, 1, and 0 °C were 60, 43, 30, 17 and 10%, respectively. The corresponding values for 2011–2040, 2041–2070 and 2071–2100 periods are 47, 27, 13, 7, 0, 0, and 33, 17, 10, 3, 0, and 13, 7, 0, 0, 0%, respectively. Hence, the risk of low-temperature days before 31 October in Visby by the end of the 21st century will be small, and in Lund even smaller (3% of the year's temperature is lower than 5 °C), and would not have influenced the simulation results significantly. In Skara and Uppsala in 2071–2100 the risks are essentially higher, about equal to that of Lund during the reference period 1961–1990 (70–73 and 13–17% of the year's temperature is lower than 5 °C and 2 °C, respectively), and would to some degree be added to the high risk of crop failure predicted by our simulations (). Also the risk of low temperatures in May decreases in the future. The fraction of years with mean temperature below 5 °C in Visby decreases from 80% in 1961–1990 to 27, 10 and 0% in 2011–2040, 2041–2070 and 2071–2100, respectively. The corresponding values for Lund, Skara and Uppsala in 1961–1990 are 27, 57 and 73%, respectively, and for 2071–2100 0, 13 and 20%, respectively. The frost risk in spring is to some extent considered in the simulations by means of a temperature-dependent sowing date ().
The atmospheric CO2 concentration is another factor not explicitly accounted for in the model. For C4 plants, such as maize, an increased CO2 concentration is regarded as having an essentially lower influence on yield than for C3 plants such as winter wheat (Lawlor and Mitchell Citation1991, Leakey et al. Citation2006). Hence the influence would be essentially lower than, for instance, +30%, which is the relative increase assumed in a study for winter wheat by the end of the 21st century for the A2 emission-scenario (Ewert et al. Citation2005). We therefore regard our maize yield projections as not substantially biased when not considering the expected CO2 increase. A possible exception is that the predicted yield might underestimate the real yield at Visby (where simulated growth becomes increasingly limited by drought), when not considering that increased CO2 concentration would increase water-use efficiency (Zalud and Dubrovsky 2002). An influence of CO2 concentration on maize phenology, i.e. flowering time, can probably be excluded (Leakey et al. Citation2006).
In conclusion, the simulation study revealed potential for future expansion of the silage maize acreage in Sweden. Changes in silage maize yield for the tested cultivars were relatively small under future climatic conditions in Sweden. However, forage quality would improve substantially, especially for northern locations, due to a prolonging of the growing season, early sowing dates and warmer summers. Nevertheless, by the end of the 21st century a high forage quality would not be achieved in about 30% of the years in the north. The reason for the small yield changes in southern Sweden is mainly attributed to a predicted earlier harvest due to hastening quality development, while in the east increased water stress counteracts the positive effect of prolonged growing season. The model may provide important predictions on regional variation of the yield and quality performance of silage maize, and thus become a useful planning tool for future land-use systems. However, there is a need to adapt field experiments to the requirements of model calibrations in order that observations from Swedish field trials may be reliably extrapolated into a changing future climate.
Related Research Data
References
- AGPM 2000 . Besoins en degrés-jours des variétés de mais. Reactualisation 2000. Info No. 266 . Montardon , , France : Association Générale des Producteurs de Mais .
- Angus , J. F. , Kornher , A. , & Torssell , B. W. R. 1980 . A systems approach to estimation of Swedish ley production . XIV International Grassland Congress . Lexington Kentucky , , USA .
- Birch , C. J. , Hammer , G. L. and Rickert , K. G. 1998 . Temperature and photoperiod sensitivity of development in five cultivars of maize (Zea mays L.) from emergence to tassel initiation . Field Crops Research , 55 : 93 – 107 .
- Bleken , M. A. , Herrmann , A. , Haugen , L. E. , Taube , F. and Bakken , L. 2009 . SPN: A model for the study of soil-plant nitrogen fluxes in silage maize cultivation . European Journal of Agronomy , 30 : 283 – 295 .
- Bonhomme , R. , Derieux , M. and Edmeades , G. O. 1994 . Flowering of diverse maize cultivars in relation to temperature and photoperiod in multilocation field trials . Crop Science , 34 : 156 – 164 .
- Boogard , H. L. , Van Diepen , C. A. , Rötter , R. P. , Cabrera , J. C. M. A. and Van Laar , H. H. 1998 . WOFOST 7.1. user guide for the WOFOST 7.1. crop growth simulation model and WOFOST control centre 5.1. Tech. Doc. 52 , Wageningen : Alterra, WUR .
- Bootsma , A. , Gameda , S. and McKenney , D. W. 2005 . Potential impacts of climate change on corn, soybeans and barley yields in Atlantic Canada . Canadian Journal of Soil Science , 85 : 345 – 357 .
- 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 , G. , Recous , S. , Tayot , X. , Plenet , D. , Cellier , P. , Machet , J. M. , Meynard , J. M. and Delécolle , R. 1998 . STICS: a generic model for the simulation of crops and their water and nitrogen balance. I. Theory and parameterization applied to wheat and corn . Agronomie , 18 : 311 – 346 .
- Challinor , A. J. , Ewert , F. , Arnold , S. , Simelton , E. and Fraser , E. 2009 . Crops and climate change: progress, trends, and challenges in simulating impacts and informing adaptation . Journal of Experimental Botany , 60 : 2775 – 2789 .
- Chmielewski , F.-M. , Müller , A. and Bruns , E. 2004 . Climate change and trends in phenology of fruit trees and field crops in Germany, 1961–2000 . Agricultural and Forest Meteorology , 121 : 69 – 78 .
- Cooper , G. , McGechan , M. B. and Vinten , A. J. A. 1997 . The influence of a changed climate on soil workability and available workdays in Scotland . Journal of Agricultural Engineering Research , 68 : 253 – 269 .
- Davies , A. , Shao , J. , Brignall , P. , Bardgett , R. D. , Parry , M. L. and Pollock , C. J. 1996 . Specification of climatic sensitivity of forage maize to climate change . Grass and Forage Science , 51 : 306 – 317 .
- Drury , C. F. and Tau , C. S. 1995 . Long-term (35 years) effects of fertilization, rotation and weather on corn yields . Canadian Journal of Plant Science , 75 : 355 – 362 .
- Easterling , W. E. 1996 . Adapting North American agriculture to climate change in review . Agricultural and Forest Meteorology , 80 : 1 – 53 .
- Eckersten , H. , Andersson , L. , Holstein , F. , Mannerstedt Fogelfors , B. , Lewan , E. , Sigvald , R. , Torssell , B. , & Karlsson , S. 2008 . Bedömningar av klimatförändringars effekter på produktion inom jordbruket i Sverige (An evaluation of climate change effects on crop production in Sweden) . Report No. 6, Department of Crop Production Ecology, Swedish University of Agricultural Sciences, Uppsala, Sweden. 145 pp. http://www2.vpe.slu.se/VPEreport/VPE%20report%20No%206.pdf. (In Swedish with English summary)
- Eckersten , H. , Kornher , A. , Bergkvist , G. , Forkman , J. , Sindhoj , E. , Torssell , B. and Nyman , P. 2010 . Crop yield trends in relation to temperature indices and a growth model . Climate Research , 42 : 119 – 131 .
- Eckersten , H. , Torssell , B. , Kornher , A. and Boström , U. 2007 . Modelling biomass, water and nitrogen in grass ley: estimation of N uptake parameters . European Journal of Agronomy , 27 : 89 – 101 .
- Ewert , F. , Rounsevell , M. D. A. , Reginster , I. , Metzger , M. J. and Leemans , R. 2005 . Future scenarios of European agricultural land use. I. Estimating changes in crop productivity. Agriculture . Ecosystems and Environment , 107 : 101 – 116 .
- Fagerberg , B. and Nyman , P. 1994 . Modeling weather effects on nutritional value of grassland leys. Estimation and validation of parameters in a model for changes in metabolizable energy content . Swedish Journal of Agricultural Research , 24 : 147 – 156 .
- Fagerberg , B. and Torssell , B. 1995 . Estimation of variation in growth potential of leys with different red clover contents, nitrogen applications and cutting regimes . Das Wirtschaftseigene Futter , 41 : 145 – 162 .
- FFE 2009 . Fältforskningsenheten, Swedish University of Agricultural Sciences. www.ffe.slu.se.
- Fogelfors H. , Wivstad M. , Eckersten H. , Holstein F. , Johansson S. , & Verwijst T. Strategic analysis of Swedish agriculture – production systems and agricultural landscapes in a time of change . Report 10. Department of Crop Production Ecology, Swedish University of Agricultural Sciences Uppsala 2009
- Gonzalez-Camacho , J. M. , Maillhol , J. C. and Ruget , F. 2008 . Local impact of increasing CO2 in the atmosphere on maize crop water productivity in the Drome valley, France . Irrigation Science , 57 : 229 – 243 .
- Gurr , G. M. and Kimber , D. S. 1983 . Ontario units and maize growth in the UK . Journal of the National Institute of Agricultural Botany , 16 : 257 – 265 .
- Herrmann , A. , Kelm , M. , Kornher , A. and Taube , F. 2005a . Performance of grassland under different cutting regimes as affected by sward composition, nitrogen input, soil conditions and weather – a simulation study . European Journal of Agronomy , 22 : 141 – 158 .
- Herrmann , A. , Kornher , A. and Taube , F. 2005b . A new harvest time prognosis tool for forage maize production in Germany . Agricultural and Forest Meteorology , 130 : 95 – 111 .
- Herrmann , A. , Kornher , A. , Höppner , F. , Greef , J. M. and Taube , F. 2006 . Harvest time prediction, validation, and implementation in practical agriculture . Grassland Science in Europe , 11 : 559 – 651 .
- Holden , N. M. and Brereton , A. J. 2003 . “ The impact of climate change on Irish agriculture ” . In Climate change scenarios and impacts for Ireland , Edited by: Sweeney , J. 33 – 79 . Johnstown Castle , Wexford : ERTDI Report Series No. 15. Environmental Protection Agency .
- Hushållningssällskapen 2009 Agricultural Extension Services Organisation . http://www.skaneforsoken.nu/. (in Swedish)
- IPCC 2000 IPCC Special Report on Emissions Scenarios . http://www.ipcc.ch/ipccreports/assessments-reports.htm
- IPCC 2007 . , in K. L. Parry , O. F. Canziani , J. P. Palutikof , P. J. van der Linden , & C. E. Hanson Climate change 2007: Impacts, adaptation and vulnerability , Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change . Cambridge : Cambridge University Press .
- Jones , C. A. and Kiniry , J. R. 1986 . CERES-Maize. A simulation model of growth and development , College Station , TX : Texas A&M University Press .
- Jordbruksverket 2010 . Swedish Board of Agriculture http://www.jordbruksverket.se/amnesomraden/odling/andrajordbruksgrodor/majs/skordestatistik.html. (in Swedish)
- Kornher , A. and Torssell , B. W. R. 1983 . Estimation of parameters in a yield prediction model for temporary grasslands using regional experimental data . Swedish Journal of Agricultural Research , 13 : 137 – 144 .
- Kruse , S. , Herrmann , A. , Kornher , A. and Taube , F. 2008 . Evaluation of genotype and environmental variation in fibre content of silage maize using a model-assisted approach . European Journal of Agronomy , 28 : 210 – 223 .
- Kucharik , C. J. , & Serbin S. P. 2008 . Impacts of recent climate change on Wisconsin corn and soybean yield trends . Environmental Research Letters , 3 034003 .
- Lawlor , D. W. and Mitchell , R. A. C. 1991 . The effects of increasing CO2 on crop photosynthesis and productivity: a review of field studies . Plant, Cell and Environment , 14 : 807 – 818 .
- Leakey , A. D. , Uribelarrea , M. , Ainsworth , E. A. , Naidu , S. L. , Rogers , A. , Ort , D. R. and Long , S. P. 2006 . Photosynthesis, productivity, and yield of maize are not affected by open-air elevation of CO2 concentration in the absence of drought . Plant Physiology , 140 : 779 – 790 .
- Leipner , J. and Stamp , P. 2009 . “ Chilling stress in maize seedlings ” . In Handbook of maize. Its biology , Edited by: Bennetzen , J. L. and Hake , S. C. 291 – 310 . New York : Springer Sci.+Business Media .
- Lobell , D. B. and Field , C. B. 2007 . Global scale climate-crop yield relationship and the impacts of recent warming . Environmental Research Letters , 2 : 014002
- Meza , F. J. and Silva , D. 2009 . Dynamic adaptation of maize and wheat production to climate change . Climatic Change , 94 : 143 – 156 .
- Miller , T. A. , Muslin , E. H. , Muslin , E. H. and Dorweiler , J. E. 2008 . A maize CONSTANS-like gene, conz1, exhibits distinct diurnal expression patterns in varied photoperiods . Planta , 227 : 1377 – 1388 .
- Moriondo , M. , Bindi , M. , Kundzewicz , Z. W. , Szwed , M. , Chorynski , A. , Matczak , P. , Radziejewski , M. , McEvoy , D. and Wreford , A. 2010 . Impact and adaptation opportunities for European agriculture in response to climatic change and variability . Mitigation and Adaptation Strategies for Global Change , 15 : 657 – 679 .
- Muchow , R. C. , Sinclair , Y. R. and Bennet , J. M. 1990 . Temperature and solar radiation effects on potential maize yield across location . Agronomy Journal , 82 : 338 – 343 .
- Olesen , J. E. , Carter , T. R. , Diaz-Ambrona , C. H. , Fronzek , S. , Heidmann , T. , Hickler , T. , Holt , T. , Minguez , M. I. , Morales , P. , Palutikof , J. P. , Quemada , M. , Ruiz-Ramos , M. , Rubaek , G. H. , Sau , F. , Smith , B. and Sykes , M. T. 2007 . Uncertainties in projected impacts of climate change on European agriculture and terrestrial ecosystems based on scenarios from regional climate models . Climatic Change , 81 : 123 – 143 .
- Pearson , C. J. , Bucknell , D. and Laughlin , G. P. 2008 . Modeling crop productivity and variability for policy and impacts of climate change in eastern Canada . Environmental Modeling and Software , 23 : 1345 – 1355 .
- Phipps , R. H. , Fulford , R. J. and Crofts , F. C. 1974 . Relationships between the production of forage maize and accumulated temperature, Ontario heat units and solar radiation . Agricultural Meteorology , 14 : 385 – 397 .
- Quilbé , R. , Rousseau , A. N. , Moquet , J.-S. , Savary , S. , Ricard , S. and Garbouj , M. S. 2008 . Hydrological responses of a watershed to historical land use evolution and future land use scenarios under climate change conditions . Hydrology and Earth System Sciences , 12 : 101 – 110 .
- SAS Institute 2007 . JMP User Guide, Release 7 SAS Institute Inc Cary , N.C
- SLU 2009 . Ultuna climate station . Swedish University of Agricultural Sciences. http://grodden.evp.slu.se/slu_klimat/index.html
- SMHI 2009 . Swedish Meteorological and Hydrological Institute . http://www.smhi.se/en.
- Smith , P. J. , Bootsma , A. and Gates , A. D. 1982 . Heat units in relation to corn maturity in the Atlantic region of Canada . Agricultural Meteorology , 26 : 201 – 213 .
- SOU 2007 . Sverige inför klimatförändringarna – hot och möjligheter (Sweden and climate change – threats and opportunities . http://www.regeringen.se/sb/d/8704/a/89334.
- Torriani , D. S. , Calanca , P. , Schmid , S. , Beniston , M. and Fuhrer , J. 2007 . Potential effects of changes in mean climate and climate variability on the yield of winter and spring crops in Switzerland . Climate Research , 34 : 59 – 69 .
- Torssell , B. W. R. Kornher , A. & Svensson , A. 1982 Optimization of parameters in a yield prediction model for temporary grasslands . Report 112 . Department of Crop Production, Swedish University of Agricultural Sciences. Uppsala
- 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 USA , 104 : 19686 – 19690 .
- Wiklert , P. Andersson , S. Weidow , B. Karlsson , I. & Håkansson , A. 1983 Studier av markprofiler i svenska åkerjordar. En faktasammanställning . Reports from the Department of Soil Sciences, Division of Agricultural Hydrotechnics, SLU.
- Wolf , J. and Van Diepen , C. A. 1994 . Effects of climate change on silage production potential in the European Community . Agricultural and Forest Meteorology , 71 : 33 – 60 .
- Wösten , J. H. M. Lilly , A. Nemes , A. & Le Bas , C. 1998 . Using existing soil data to derive hydraulic parameters for simulation models in environmental studies and in land use planning . Report 157 DLO Winand Staring Centre. Wageningen
- Yang , H. S. , Dobermann , A. , Lindquist , J. L. , Walters , D. T. , Arkebauer , T. J. and Cassman , K. G. 2004 . Hybrid-maize – a maize simulation model that combines two crop modelling approaches . Field Crops Research , 87 : 131 – 154 .
- Zalud , Z. and Dubrovsky , M. 2002 . Modelling climate change impacts on maize growth and development in the Czech Republic . Theoretical and Applied Climatology , 72 : 85 – 102 .