3.1 Probability of heat stress events
The probabilities of maximum daily temperature >30ºC (T>30ºC) and of 5ºC above the historical average (T+5ºC) were analysed for the evaluated 33 locations (Supplementary Table S1). Data of representative locations in each area are shown in Figure 2. Daily probabilities of both types of heat stress events differed in their trends in each locality from September 1st to December 31st. When heat events of T>30ºC were assessed, the daily probability of its occurrence was increasing after F in most of these locations (Fig. 2 left). Salto Grande showed the highest average probability (26%) of T>30ºC during the post-flowering period, followed by Reconquista, Tres Arroyos (25% in both locations) and Passo Fundo (11%). On the contrary, the lowest probability was recorded in Pichoy, where the probability was almost zero during the post-flowering period.
The heat stress event T+5ºC did not show a clear trend for the post-flowering period of wheat and canola. However, a slight decline towards PM was found in low latitude locations such as Passo Fundo, Reconquista and Salto Grande. In these locations, the daily probability of T+5ºC event was close to or lower than 20% (Fig. 2, central panels). In other words, maximum daily temperature was over the historical daily maximum temperature by 5ºC every ~2 out of 10 years. This daily probability was similar along the post-flowering period. Across the evaluated locations, the average probability of T+5ºC during the post-flowering period ranged from 10 to 16%.
Additionally, the typical duration of the post-flowering period for each of the 33 locations and the probabilities of occurrence of T>30ºC and T +5ºC events were represented graphically (Supplementary Fig S1).
3.2 Probability of cloudiness events
Cloudiness events decreasing global solar radiation by 50% relative to historical data (R50%) from F to MF showed higher daily probability at the beginning of the post-flowering period (Fig. 2, right panels). Across areas and locations of the Southern Cone, clear differences were found in cloudiness patterns. In the northern and central areas, the daily probability ranged from 10 to 60% throughout the post-flowering period. A lower daily probability was observed in the southern area, with daily probability ranging 3% to 30%. When comparing the average probability for post-flowering, Passo Fundo showed the highest value (38%) followed by Reconquista, Salto Grande and Tres Arroyos (23%), while the lowest was found in Pichoy (10%) (Fig. 2, right panels).
Similar to T>30ºC and T+5ºC events, the average probability of daily occurrence of R50% for each location during the typical duration of the post-flowering period is shown in Supplementary Figure S1.
3.3 Time-course of high temperature and cloudiness events
The time-course of the probability of T>30ºC and T+5ºC events has been increasing in some locations and has remained stable in others over the last 28 years (Supplementary Figs. S2 to S7). Example of locations with a tendency to increase the probability of T>30ºC events over the years were Ceres, Salta, Laboulaye, Rio Cuarto, Trenque Lauquen, Santa Rosa, General Pico and Bolivar (Figs. S2 to S4). Other locations as Santiago del Estero, Sauce Grande and Bahía Blanca showed stability for this event, while it did not occur in Pichoy and Chillán. For the T+5ºC event, locations such as Salta, Bahía Blanca, General Pico, Coronel Suárez, Mar del Plata, Trenque Lauquen, Santa Rosa, Bolivar and Tres Arroyos increased the probability of occurrence of high temperatures since 2001, while Guarapuava, Gualeguaychú and Pichoy remained stable throughout the last 28 years (Figs. S5 to S7).
The probability of R50% event has been variable according to each area (Supplementary Figs. S8 to S10). In the northern area, it is worth noting the increase in the incidence of R50% in Guarapuava and Santiago del Estero since approximately 2010. Also, since this year, certain increases in the probability of occurrence of R50% have also been observed in some locations of the central area, such as Gualeguaychú, Venado Tuerto, Sauce Viejo, Rosario, and Salto Grande. Regarding the southern area, locations such as Olavarría, Bolivar, Coronel Suárez, Junín, and Santa Rosa tended to lower the probability of R50% until 2010 and then it seems to increase slightly. A separate case in this area was Trenque Lauquen, which increased its probability until 2001 and subsequently decreased (Supplementary Fig. S10).
- Probabilities of occurrence of combined events of high temperature and cloudiness
For T>30ºC plus R50% combined event (Fig. 3), the monthly probability of occurrence varied in both temporal and spatial scale. In the sites located in the north-northwest of the Southern Cone of South America the T>30ºC plus R50% event was already evident in September, and in the central and southern sites, it was only clear from October onward. The monthly probability of this event increased from September to December across locations, largely explained by the increase in maximum temperature from the end of spring.
Within each area (north, central, south) the monthly probability of occurrence of the event presented variability among locations. In the northern area, during the post-flowering period of wheat and canola (September-October), monthly probabilities of combined stress (T>30ºC plus R50) were from <10% in those locations of greater elevation (e.g., Guarapuava, 1120 m of altitude) to probabilities ranging 61 to 70% at lower altitudes (e.g., Londrina, 650 m and Santiago del Estero, 200 m).
In the central area, the post-flowering period took place from October to November. In that period, monthly probabilities of occurrence of T>30ºC plus R50% events also varied among locations, from <10% in La Estanzuela to the highest values in Venado Tuerto, ranging 41-50%, while in the rest of locations probability ranged between 21 and 40% (Fig. 3).
For the southern area, the highest monthly probability of T>30ºC plus R50% events during the post-flowering period was found from November to December in Trenque Lauquen, with maximum monthly probabilities from 71 to 80%. In Santa Rosa, Tandil, Coronel Suárez, and Tres Arroyos, probability ranged between 21 and 30%, while the lowest probability for combined stress T>30ºC plus R50% was observed in the Chilean locations of Chillán and Pichoy (<10%).
The T+5ºC plus R50% combined events presented the highest monthly probability in the south-southeastern locations of the region, especially during September, which is prior to the occurrence of post-flowering in those locations (Fig. 4). Towards October and November, the monthly probability of the events declined from the north to the south areas, while in December the probabilities of occurrence decreased in the great majority of the locations of the Southern Cone, with the exception of General Pico, Olavarría and Mar del Plata (in the southern area). Despite the general trend described above, the magnitude of the monthly probability of occurrence of T+5ºC plus R50% varied between locations of each area. Salta and Passo Fundo (both from the north area) as well as Río Cuarto and Tacuarembó (both from the central area) exhibited the highest monthly probabilities of the T+5ºC plus R50% combined events during post-flowering, reaching values close to 40%. For the rest of locations, such probability was below 30% (Fig. 4). For the southern area, Mar del Plata and General Pico stood out from the rest, with a mean monthly probability of 56% and 46%, respectively.
Regarding the number of days per month under a combined event (Figs. 3 and 4), an association was observed between the magnitude of the calculated probability values of T>30ºC plus R50% and T+5ºC plus R50% events for each month (colour scale) and the number of days when it took place within each month (point size). In other words, probabilities below 30% in the months of September to December were associated with very low frequencies of days of occurrence per month (less than 0.5 days in a month); probabilities between 31 and 50% were related to frequencies of occurrence of 0.5 to 1 day per month and the highest probabilities, between 51 to 80%, with the highest frequencies of occurrence of days per month, with values from 1 to 4 days per month. When the probability of occurrence was related to the number of days of stress per month (Figs. 5 and 6), an asymptotic curve was observed for both combined stress events but differing among monthly periods. Thus, for the T>30ºC plus R50 % event the relationship was more noticeable in November and December with no different trends between locations per area (north, central, south). In the T+5ºC plus R50% event the relationship was stronger in September where it was observed that the locations of the southern area presented the highest probabilities and the largest number of days.
3.4 Cumulative probability during post-flowering period
Cumulative probabilities of occurrence of both combined stress events, throughout the post-flowering period typical for each area were calculated for all locations (Table 2). In addition, cumulative probability (cumulated from the 1st September) of occurrence of both types of combined events T>30ºC plus R50% and T+5ºC plus R50%, are compared for selected locations from each area: Passo Fundo and Reconquista for the northern area, Salto Grande for the central area, and Tres Arroyos for the southern area (Fig. 7).
In Passo Fundo, the T>30ºC plus R50% event presented a high cumulative probability (from the 1st September) during the period between F and PM, with 19% in F and 48% in PM, which indicated a cumulative probability of 29% for the post-flowering period (Fig. 7 left and Table 2). Thus, in ~3 out of 10 years, the T>30ºC plus R50 combined event occurred during the typical post-flowering period in this northern location. A slightly higher value of cumulative probability for the T+5ºC plus R50% event during post-flowering, around 38% was found in Passo Fundo, which was the highest cumulated value obtained compared to the rest of the sites in the northern area (Table 2). By contrast, in Reconquista the cumulated probability during the post-flowering period was higher for T>30ºC plus R50% (43%) compared to T+5ºC plus R50% (26%; Table 2).
Finally, Salto Grande and Tres Arroyos showed similar cumulated values in PM, with 44-40% for T>30ºC plus R50%, and 72-74% for T+5ºC plus R50%, respectively (Fig. 7). During the post-flowering period, the cumulated probability was similar at both sites, with around 40% for T>30ºC plus R50%, and around 20% for T+5°C plus R50% event (Table 2).
Comparing (across locations) the average cumulative probability during post-flowering period was 42% in the T>30ºC plus R50% event vs. 15% in the T+5ºC plus R50% event for the northern area; whereas it was 26% vs 19% in the central area, and 28% vs 1% in the southern area, respectively (Table 2). It is worth noting that only for Pichoy and Chillán, in the southernmost area at the west of the Andes, there were no increases of cumulative probability of any combined stress events during the post-flowering period (Table 2).