Climate Change In The Argentinean Wheat Region: Temperature And Precipitation At Two Contrasting Sites


 Global climate change is shifting temperature and precipitation regimes, which is modifying the environments that define wheat yield and quality. The current work characterises the changes that have occurred in the thermal and hydric environment in two contrasting sites of the wheat growing region of Argentina, allowing comparison between sites for these changes and for how the changes are accelerating. Temperature and precipitation variables were analysed by regression and trend testing (Mann Kendall), and future projections were made based upon significant relationships. The two sites compared were in the zones around the cities of Azul in the Province of Buenos Aires and Marcos Juárez in the Province of Córdoba, located approximately 500 km apart. The climate data analysed covered the period 1931–2014 for Azul and 1952–2014 for Marcos Juárez. At both sites, temperatures increased significantly in mean and extreme values over these periods, where the rate of change accelerated during the first years of the 21st century. The changes observed were in general more pronounced in Marcos Juárez than in Azul. Furthermore, in Marcos Juárez, mean precipitation increased from September to December and there was a higher frequency of extremes of precipitation greater than 100 mm in September and October during the early 21st century. Evidence was found for temperature rise and the occurrence of extreme temperature and precipitation events occurring differently between sites, as well as for its acceleration rate in the early 21st century. The projected future changes made implied that wheat yield is expected to suffer losses over the coming century.

Numerous studies have attempted to quantify wheat yield loss due to climate change. In Argentina, Abbate and Lázaro (2010) reported falls of 1.15 mg in potential grain weight (approximately 4% for a mean grain weight of 30 mg) for each degree Celsius increase in mean temperature during 35 days postanthesis; while in Europe, Moore and Lobell (2015), reported yield losses of 2.5% for wheat, analysing long term trends of temperature and precipitation, and pointed to more detrimental effects in southern with respect to cooler regions (i.e. the United Kingdom and Ireland), which could not mitigate the impact with rainfall increase. Lobell et al. (2011), considering the whole growing season, reported a higher yield loss per degree increase, of 5.5%; furthermore, according to Zhao et al. (2017), without CO 2 fertilization, effective adaptation and genetic improvement, each degree-Celsius increase in global mean temperature would, on average, reduce global yields of wheat by 6.0 ± 2.9% (Zhao et al. 2017), in concurrence with Asseng et al (2015), who also reported a gure of 6%. Wheat yield losses have been observed in spite of the bene cial effects of CO 2 for C3 species (Lobell et al. 2011), and at low latitudes, the climate change effect will be more negative, specially at a high level of warming with nitrogen (N) stress, where there would be little true CO 2 compensation for C3 species (Rosenzweig et al. 2014). Schleussner et al. (2016) found substantial differences in impacts between 1.5 ºC and 2 ºC warming with local yield reduction at middle and low latitudes, especially for wheat and maize. Trends are particularly severe for temperature in wheat evaluated at regional and national level (Lobell et al. 2011), with 6 and 8% of yield losses for 1.5 ºC and 2 ºC respectively for wheat (Schleussner et al. 2016).
Many studies analysing the umbral and maximum temperature have found negative effects on wheat yield and/or quality, and there is a certain consensus that temperatures above 30°C lower yield and quality through reductions in the rate of starch deposition (Jenner 1994) and in dough strength (Randall and Mo 1990).
Added to the above, events between anthesis and maturity condition quality characteristics of the crop since during this stage the molecular weight of the storage proteins, which are major determinants of quality, is rising (Fraschina 2017). Water stress events after owering modify the rate of deposition of distinct grain components, negatively affecting yield and quality.
In Argentina, the long-term tendency for 1940-2007 of various agroclimatic indices showed heating, principally due to increments in the minimum temperature (Fernández Long et al. 2008), although for the most recent period , the tendency was weaker and in the opposite direction in some meteorological stations (Fernández Long et al. 2008). In the Pampas region, rises in annual minimum temperatures of 2°C/century were reported during 1959-1998 (Rusticucci and Barrucand 2004). Additionally across this region, except for a small number of locations, the frost period decreased for the period Regarding precipitations, increases have been reported across the planet throughout the 20th century (Easterling et al. 2000). For example, the frequency of abundant precipitations has risen in South-Central USA and sectors of South America during 1950-2005, by 2 days per decade for the number of days/year, and the number of consecutive days without rain has reduced since 1960 (Alexander et al. 2006).
In the Pampas region of Argentina, Magrin et al. (2005) reported that mean precipitation rose by over 150 mm during the last thirty years of the 20th century In Argentina, climate studies have been carried out in the wheat region for mean temperatures (Fernández Long et al. 2008Long et al. , 2012Magrin et al. 2005Magrin et al. , 2009), extreme temperatures in relation to frost (Fernández Long et al. 2012) and maximum and minimum temperatures (Rusticucci and Barrucand 2004), as well as on the impact of changes in temperature, radiation and precipitation on potential yields in wheat and other crops using simulators (Magrin et al. 2005(Magrin et al. , 2009).
Although these studies provide a base for the current investigation, more exhaustive studies are necessary that also analyse changes in extreme rainfall variables, as proposed by Easterling et al. (2000), and temperature in relation to the umbral that prejudices wheat yield and quality, as well as the situation over recent years and differences within the wheat region. This type of study could serve as a model for the elaboration of perspectives for each cultivar and zone in particular.
The aim of the current work was to characterise the trend in the change in temperature and rainfall variables in two contrasting sites representative of the wheat growing region of Argentina. In contrast to Magrin et al. (2009), who studied changes in the centre of this region, we chose two sites, one towards the north of the region and the other towards the south, in order to discern whether there were differential effects of latitude, as has been suggested by modelling studies (Barros et al 1996).

Materials And Methods
Sites studied Climatic data were analysed from the meteorological stations of Azul in the Province of Buenos Aires and Marcos Juárez (abbreviated as MJ) in the Province of Córdoba, located approximately 500 km apart (Fig. 1) Azul and MJ are found in different wheat growing subregions, namely IV and II North, respectively.
The climatic conditions in wheat growing region IV are adequate for crop development, with a cool temperate climate of long winters and annual rainfall of around 800-900 mm, resulting in yields generally above the national average. Within this region, Azul itself has a temperate climate associated with the Pampas, with a mean annual temperature of 15°C and annual rainfall exceeding 900 mm. Subregion II North gives yields similar or slightly above the national average, where MJ itself has a humid temperate climate with maximum mean annual temperatures of 24°C and minimum mean annual temperatures of 11°C, although minimum temperatures can fall below − 10°C and maximum temperatures above 41°C. Annual rainfall is around 800 mm.
The main sowing, anthesis and harvest dates are early June, October and mid-December for MJ and early July, towards the end of October / into November, and the end of December / early January for Azul, respectively. Data sets and variables evaluated Data of maximum (Tmax), minimum (Tmin), mean (Tmean) temperatures and rainfall were provided by the National Meteorological Service (Servicio Meteorológico Nacional, SMN).
The original weather station in Azul (A, 36°45´S 59°50´O 132 m) was disactivated in 1994 and a new station was immediately activated 10 km away (B, 36º   50 O 147 m). The data from these (original station: 1/1/1931-15/12/1994 and new station: 16/12/1994-28/2/2015) were cross-checked with a third station (Tandil aero, 37º14´S 59º20´O), in order to verify that the 10 km distance had not signi cantly affected the measurements; this procedure showed that the original and new stations could be considered as one and the same.
The study of the monthly temperature series included Tmax, Tmin ,Tmean and daily thermal range (DTR). The duration of the period in each year with minimum temperatures equal to or below 0°C (Du T ≤ 0 ºC) was also determined. Extreme monthly temperature values included maximum (Vmax) and minimum (Vmin) temperatures, number of days with temperatures equal to or above 30°C (*) (ND T ≥ 30 ºC) and accumulated degrees above 30°C (*) (ºC T 30 ºC). The rainfall monthly variables (which included rainfall of less than 1 mm) were number of days with precipitation (ND PP) and total precipitation (mmt PP). Extreme rainfall variables as number of days with precipitation equal to or below 5 mm (*) (ND PP ≤ 5 mm), equal to or above 50.8 mm (*) (ND PP ≥ 50.8 mm) and equal to or above 101.6 mm (*) (ND PP ≥ 101.6 mm) were included according to Easterling et al. (2000) (2 and 4 inches equal to 50.8 and 101.6 mm, respectively). Accumulated precipitation from September to December each year (mm PP4 months) were also included. Variables above marked with with * included many values equal to 0 due to the criteria used for their construction. Additionally, the linear regression gradients from the whole period for each site for those cases where the value of p from MK was signi cant were compared by t-test (see Eq. 1).

Equation 1:
Where X1 and X2 are the gradients from sites 1 and 2, respectively. Signi cance level 5%. Impact of the last years on the general tendency for the constructed variables  Where p1 is the proportion of the rst sample (Azul); p2 is the proportion of the second sample (MJ); N1 is the size of the rst sample (Azul); N2 is the size of the second sample (MJ); p = (p1*N1 + p2* N2)/(N1 + N2); q = 1-p and the degrees of freedom are calculated as (N1 + N2)-2. 10, 50 and 100 years future projections for both sites Using the regression equations for the complete period of each site, estimations were made of future projections for the next 10, 50 and 100 years for the variables with signi cant trends. Projections were also made of future wheat yield losses due to global heating, based upon an estimated 6% loss in yield per°C

Analysis of temperature
The monthly mean temperature variables increased signi cantly over the years in certain months at both sites (Table 1) for October in MJ were approximately three-fold higher than those in Azul (Tmin non-signi cant in Azul). These observations imply that, in general for the temperature variables, the rate of climate change in MJ was higher than that in Azul Table 1 Regression coe cients and Mann Kendall tests for the complete period in Azul and Marcos Juárez for the mean and extreme temperature and rainfall variables, and comparison between regression coe cients where these were signi cant at both sites. b -regression coe cient. * and ** indicate signi cant differences at 0.05 and 0.01, respectively. N -no days observed with daily rainfall ≥ 101.6 mm for these months (September in Azul and  In contrast, the daily thermal range monthly mean (DTR) and the duration of the period with minimum temperatures equal to or below 0°C (Du T ≤ 0 ºC) showed no signi cant changes at either site (Table 1).
From these results it appears that the variables in general are by no means showing similar changes over the study period; further evidence for this is provided by the analysis of variables focused upon more extreme temperature phenomena: the maximum monthly temperature value (Vmax), the minimum monthly temperature value (Vmin), the number of days per month with temperatures equal to or above 30°C (ND T ≥ 30 ºC) and the accumulated degrees per month above 30°C (ºC T 30 ºC) ( Table 1). While no signi cant regressions were observed for Vmax, Vmin did show such regressions, although in Azul only in October (albeit with a regression coe cient higher than those observed for the temperature variables mentioned above); in contrast, in MJ, the regressions were signi cant in October, November and December, with gradients higher than those observed in Azul. Just as in the case of the variables Tmax, Tmin and Tmean mentioned above, this again implies that the rate of climate change in MJ is generally more pronounced that that in Azul.
Further evidence for differences between sites was provided by ND T ≥ 30 ºC and ºC T 30 ºC, for which signi cant regressions were only observed in MJ, in October and November for the former variable and in October for the latter, with regression coe cients amongst the highest observed for all the variables analysed. Hence there seems to be no appreciable climate change for these characters in Azul, in contrast to MJ, where the rate of change was considerable.
In conclusion, of all the thirty-three combinations of comparisons carried out (eight variables for which four months were analysed per variable, plus one variable (Du T ≤ 0 ºC) analysed over the four months as a whole) ( Regarding variables associated with extreme rainfall phenomena, i.e. the number of days with precipitation equal to or below 5 mm per month (ND PP ≤ 5 mm), the number of days with precipitation equal to or above 50.8 mm per month (ND PP ≥ 50.8 mm) and the number of days with precipitation equal to or above 101.6 mm (ND PP ≥ 101.6 mm) (NB: the distribution of the residual errors was non-random for the latter two variables), no signi cant differences were found over the months analysed for either site, again supporting the idea that there was relatively little climate change observed for rainfall (Table 1).

Multivariate comparison between sites
Having provided evidence that for individual variables the sites showed notable differences in their rates of climate change, we applied principal components (PC) analysis to see to what extent the sites differed when the variables were considered as a whole across the whole period (1952-2014) (Fig. 2).
The sites clearly differed for PC 1 since all data points had higher values for PC 1 than do those in Azul, component positively correlated with the temperature variables Tmin, Tmean and Tmax, together covering the four months under study, and negatively correlated with the duration of the period with temperatures equal to or below 0°C (Du T ≤ 0 ºC), albeit that the correlations were low in magnitude (Table 2). For the period 2000-2014, PC 1 was related to temperature variables and Du T ≤ 0 ºC in a similar way to that observed for the whole period, as well as to variables associated with extreme events (Vmax and ND T > 30 ºC, and there was also a slight tendency to be related to rainfall variables ( Table 2).
In general, the sites were not discriminated by PC 2 (Fig. 2) or any other PC (results not shown).
Taken as a whole, together with the results for individual variables described earlier, the data con rm that MJ is clearly the warmer of the two sites (for the variables correlated with PC 1).
Additionally, the points for the years 2000-2014 for MJ are slightly displaced towards higher values of PC 1 than the remaining years at this site, whereas this is not so for Azul, implying that the more recent years at MJ are warmer than previous years and indicating a possible acceleration in climate change at this site for the variables correlated with PC 1.

Impact of the period 2000 -2014
Following this lead, analysis of the frequency of extreme temperature and rainfall events for the period 2000-2014 compared with that for 1952-2014 (Table 3) showed that in the more recent period it was found that, at both sites, there were increases in variables associated with the 30°C limit (ND T ≥ 30°C and °C T > 30°C) in October. Additionally, all events of rainfall of 101.6 mm or more (ND PP ≥ 101.6 mm) have been recorded in MJ in September and October during more recent years (Table 3). The analysis of differences between proportions (Eq. 2) of the rst fteen years of the century compared to the whole period showed that the temperature extremes (ND T ≥ 30°C and °C ≥ 30°C) had become more frequent at both sites in September, differentially so (P value of 0.03 and 0.02, respectively).
Regarding rainfall extremes, this was also observed for ND PP ≥ 50.8 mm for September (P value of 0.01).
Hence climate change appears to have become more marked in recent years compared to previous ones, seemingly more so in MJ than in Azul.

Future projections
From the above results, twenty-four of the variables analysed were found to be changing signi cantly and we made projections over the next 10, 50 and 100 years based upon our regression coe cients over the whole period ( Table 4). Given that the changes appear to have accelerated over the rst fteen years of the century, these will likely be underestimates of the expected changes, which in any case will, of course, depend upon future greenhouse gas emission scenarios and a host of other factors.  (Table 4). Furthermore, within the next fty years, the number of days with temperatures equal to or above this temperature could rise by 27% in October and 23% in November, accumulating 6.68°C above the current value for October (Table 4).

Discussion
The analysis of historical climate data from Azul, Province of Buenos Aires, and Marcos Juárez, Province of Córdoba, sites in two contrasting zones representing the wheat growing region of Argentina, allowed the characterisation of past and possible future changes in temperature and rainfall variables during the months in which wheat is grown in this country. Changes in mean and extreme temperature and rainfall variables observed were largely consistent with those reported in the relevant literature (Alexander et al. 2006, Nakicenovic et al. 2007, IPCC 2007b, Trenberth et al. 2007, Fernández Long et al. 2008, Stocker et al. 2013).

Considerations for each site
Mean temperature values The maximum temperatures (Tmax) in September and October in Azul and MJ have increased signi cantly over the years (Table 1) Regarding the minimum temperatures (Tmin), these increased signi cantly over the years in November at both sites (Table 1) Table 1). The increments in this variable in October and November in MJ (Table 4) are notably greater than those projected by Rusticucci and Barrucand (2004) of 2°C per 100 years for the Pampas region. Furthermore, increments of 0.18°C per decade are projected for Azul.
The mean temperatures (Tmean) have also risen during the months analysed at both sites over the years, except for December in MJ. These rises are consistent with those reported by Stocker et al. (2013) for the 20th Century. The rises were more marked in MJ than in Azul (Table 1). If these rates of rise Regarding the duration of the period with minimum temperatures equal to or less than 0°C (Du T ≤ 0 ºC), no signi cant differences were observed ( In terms of industrial wheat quality, temperatures above 30°C break the positive relationship between grain protein content and dough strength, resulting in possible negative effects on quality, with an increase in the gliadin to glutenin ratio. Mean and extreme rainfall values Accumulated rainfall for September to December in MJ increased signi cantly (Table 1), which is consistent with Alexander et al. (2006) at the global level, Barros et al. (1996) and Magrin et al. (2009) in Argentina and Easterling et al. (2000) in the USA. Additionally, all rainfall events equal to or greater than 101.6 mm recorded in MJ in September and October occurred between 2000 and 2014, with none before that (Table 3). Gelmi and Seoane (2013) found a greater frequency of occurrence of extreme precipitation events over years in the zone including Azul between 1971 to 1999 compared to 1951 to 1970, implying long term changes appear to be underway at both sites.
Regarding the consequences of these changes, the increase in abundant rainfall events could provoke N leaching and therefore changes in agronomic practices that could mean that regions of the country hitherto not apt for wheat growing could become so, and vice versa. Future changes could be even more dramatic than the projected changes given in Table 4, since these are likely to be underestimates, given the apparent acceleration of change signalled by the results from 2000 to 2014. Nonetheless, it ought to be remembered that regional precipitation changes can be projected with less certainty than temperature (Zhao et al. 2017).

Differences between sites
As seen from the results given in Table 4 for the whole period, the variables that changed signi cantly at both sites did so more pronouncedly in MJ than in Azul, implying more extreme future conditions at the former site. In contrast, Zhao et al. (2017) found similar impacts at site scale due to temperature increase.
Furthermore, changes in extreme temperature phenomena related to the umbral temperature were only found in MJ, not in Azul. If, besides, we consider that MJ is, according to its mean and maximum temperatures for each of the months analysed, approximately 5°C warmer than Azul, the consequences for yield and quality in MJ could be more severe than in Azul. Naturally future change is related to future gas emission scenarios, as stated by the fourth report of the IPCC, but in general the indications are that climate change will continue and accelerate unless dramatic action is taken to counter it.
The sites were strikingly different in their rates of change for maximum and median temperatures in October across the whole period (Table 1). In more recent years, the sites are beginning to differ for the number of days equal to or above 30°C (ND T ≥ 30°C) and the accumulated degrees above 30°C (°C T > 30°C) in September. These changes can increase variability in yield (Hawkins et al., 2013) and quality. In general, more variables have been subject to change, and more intensely, in MJ than in Azul. Nevertheless, adaptation also signi cantly in uenced yield, with adapted crops yielding on average 7-15% greater than non-adapted (Challinor et al. 2014). The different impact of climate change in these two cities should encourage differential agricultural practices and more studies of climate change impact at the local level. In addition, more extreme measures will probably need to be adopted to ameliorate these changes for yield and quality at these sites, involving, for example, earlier sowing to escape the more extreme months for certain critical stages of plant development.

Consequences of these ndings for wheat cultivation
Our ndings suggest climate change is expected in the wheat growing region under study, the magnitude of which will depend upon the particular site involved. A special report of the IPCC (IPCC 2018) shows the consequences for crop production in general, and wheat production in particular, of these sorts of changes. It states that limiting global warming to 1.5°C, compared with 2°C, is projected to result in smaller net reductions in yields than would otherwise occur in maize, rice, wheat and potentially other cereal crops, particularly in sub-Saharan Africa, Southeast Asia, and Central and South America, as well as in smaller reductions in the CO 2 -dependent nutritional quality of rice and wheat. Temperature and precipitation trends have reduced crop production and yields, with the most negative impacts being on wheat and maize, and climate variability has been found to explain more than 60% of maize, rice, wheat and soybean yield variation in the main global breadbasket areas, with the percentage varying according to crop type and scale. Additionally, the report provides evidence that higher atmospheric CO 2 concentration will not compensate for increased temperatures, stating that observations of trends in actual crop yields indicate that reductions as a result of climate change remain more common than crop yield increases, despite increased atmospheric CO 2 concentrations.
Furthermore, production stimulation at increased atmospheric CO 2 concentrations was mostly driven by differences in climate and crop species, whilst yield variability due to elevated CO 2 was only about 50-70% of the variability due to climate. A signi cant reduction has been projected for global wheat production of 6.0 +/-2.9% for each degree Celsius increase in global mean temperature, and it should be noted that crop production is also negatively affected by the As well as the consequences for yield, the faster growth rates induced by elevated CO 2 have been found to coincide with lower protein content in several important C3 cereal grains (Myers et al. 2014), consistent with the reduced grain protein content and hence nutritional quality observed by Taub et al. (2008) and Pleijel and Uddling (2012).
Hence the changes we are projecting in the study have potentially serious consequences for wheat production in Argentina. Assuming the aforementioned 6% fall in yields per degree Celsius increase in global mean temperature (Asseng et al. 2015, Zhao et al. 2017, IPCC 2018), our mean projections suggest that yields in Azul and Marcos Juaréz could fall by as much as approximately 8% and 16%, respectively, over the coming one hundred years; the true gures will depend upon many factors, including the coincidence of the temperature changes with the critical yield determining periods in the wheat plant cycle. Although the increase in CO 2 could to some extent counteract the negative effects of warming on yield, this would be expected to reach a plateau due to possible feedback mechanisms (Long et al. 2006), whereas the temperatures would be expected to continue to exert deleterious effects in an increasing way over time.
This in turn suggests that considerable countrywide changes in wheat production practices might be needed in the future. The changes in precipitation mentioned by Lovino et al (2011) (see Introduction) have already led to land use changes. It may be the case, as previously mentioned, that some current wheat growing areas will cease to be apt for this purpose, representing enormous challenges that might only be offset by the conditions in some currently nonwheat growing areas changing to allow wheat to be grown there. This pattern would be expected to occur in wheat growing regions across the globe, unless large-scale remedial actions are implemented.
In this shifting regional and global scenario, what seems certain is that uncertainty lies ahead. Juárez than for Azul, implying greater problems for yield and quality, unless measures can be found that ameliorate the changes; this in spite of the fact that the two sites are only separated by approximately 500 km. As far as we are aware, extreme temperature and rainfall variables had not been previously analysed in this way for these sites.

Conclusions
The rst fteen years of the current century gave greater rates of change than the complete period, implying that future conditions could be even more acute than the simple projections imply, suggesting that new and severe challenges for agronomic production will have to be faced in the future. And, as previously mentioned, warming yield detrimental effects not will be totally countered by CO 2 increases, since these are expected to reach a plateau (Long et al. 2006), while temperature increases are expected to continue to rise. This type of study could help to rede ne the current "core" regions of different crops in the future.
We consider that the root causes of climate change need to be tackled and where this proves inadequate, alternative remedial action needs to be taken to avert prejudicing crop production in terms of both yield and quality, which would add serious uncertainty to a world already facing enormous challenges in feeding its burgeoning population.

CONFLICTS OF INTEREST
The authors declare no con ict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.