Retrospective analysis

In the estimated hourly data validation of an automatic weather station and their respective real hourly data, they were adjusted to linear regression (R² = 90 %, intercept equal to zero and slope value equal to one) (Figure 2) (^{Fritz et al., 1997}).

Figure 2 Model 1:1 with hourly data of observed and estimated temperature. 

With the hourly data, it was estimated the number of hours in which the temperature was found inside and outside the optimal ranges of cardinal temperatures for the development of cactus pear in the annual crop cycle. The historical trend analysis for the years 1981 to 2010 (considering the 244 stations) is presented in Table 2.

The number of hours within the optimal temperature range for the development of cactus pear (14-19 °C), did not show a significantly defined trend, that is, the number of hours with appropriate temperature for this crop remained without statistically significant changes in the study period. The lower marginal temperature (< 13 °C), presented a tendency to decrease, but not significantly; while the upper marginal temperature (> 21 °C) presented a tendency to increase significantly (Table 2). This suggested that only the upper marginal temperature could negatively affect the development of this crop when it exceeds 23 °C (^{Inglese et al., 2017}).

Concerning the average temperature in January, which is important for cactus pear, since in regions with severe frost, it negatively affects this crop (^{Nobel, 1996}; ^{Nobel & Bobich, 2002}), the analysis revealed that the number of hours in the optimal temperature range for the month of January (10-14 °C), those below (< 10 °C) and above (> 14 °C) to the optimal range in the month of January, independently of the value of the slope, resulted without significant tendency in the period from 1981 to 2010 (Table 2). These results indicated that the number of hours with average temperatures and temperatures below or above the optimal range in January appropriate for cactus pear, until now, have not been affected by climate change in the region producing cactus pear for the study period. This last analysis could be encouraging for cactus pear, however, the risk of a drop in temperature ≤ -5 °C is latent during the winter season, which, if it occurs, would minimally affect the yield and cosmetic quality of the fruit (^{Granata et al., 2017}).

Prospective analysis

For the analysis of future climate scenarios, the number of hours of each temperature range for the annual crop cycle and the month of January were related to their respective average maximum and minimum temperatures, this for 61 stations distributed nationwide. The generated models were significant (p ≤ 0.0001) with values of determination coefficients (R²) between 86.4 and 99.4 % (Table 3).

With the models outlined above, maps were generated of the number of hours of each of the ranges or threshold temperatures for the development of cactus pear (Table 1) in the current climate scenario (1981-2010) and for 2030, 2050 and 2070 climatologies, for RCP 4.5 and 8.5. As an example, the maps of temperature ranges 14 to 19 °C in the annual crop cycle and 10 to 14 °C average temperatures in the month of January are presented (Figures 3 and 4).

Figure 3 Percentage of the number of hours with temperature from 14 to 19 °C (optimum temperature for the development of cactus pear) in the annual cultivation cycle, in the current scenario (A) and three future climate scenarios of RCP 4.5 (B, C and D) for the cactus pear producing region in Mexico. 

Figure 4 Percentage of the number of hours with a temperature of 10 - 14 °C in January (optimal temperature in January for the development of cactus pear), in the current scenario (A) and in three future climate scenarios of RCP 4.5 (B, C and D) in the cactus pear producing region in Mexico. 

Optimal average annual temperature (14 to 19 °C)

Figure 3 shows the maps with the percentage of the number of hours with temperature in the optimal range for cactus pear development (14 to 19 °C) in the annual crop cycle in the north-central region. In reference to Figure 3A, it was observed that 100 % of the area planted with cactus pear in the producing municipalities (current scenario), is in the range of 20 to 30 % of the total number of hours in the year (Table 4). By the year 2030, a slight decrease in the cultivated area with percentage of hours from 20 to 30 % (orange color) would be expected, going from 100 to 99.6 % (Figure 3B and Table 4), which means that it would be practically the same area with temperature within the optimal range for cactus pear development. In the climate scenarios 2050 and 2070, it would also slightly decrease the surface area in the range 20-30 % of hours between 14 and 19 °C, but only 2.9 and 4.0 % in the 2050 and 2070 scenarios of RCP 4.5, respectively (Table 4). This indicates that the number of hours with optimal temperature for the development of this crop would only be slightly affected by the effect of climate change and would remain at least until 2070 in 96 % of the surface area of the producing municipalities. Considering RCP 8.5, the surface area in the range 20-30 % of hours with optimal temperature for cactus pear would have a decrease of up to 15.6 % in the 2070 scenario. Considering the pressure on resources (e.g., water, soil, etc.) due to population increase and climate change, food security would not be guaranteed by that time (^{Rosenzweig et al., 1994}). However, the above scenarios, based only on temperature, offer the possibility to continue growing this species in the present and immediate future. The latter would be adverse for other crops (e.g., wheat, maize and barley) (^{Lobell & Field, 2007}).

Lower marginal average annual temperature (< 13 °C)

Table 5 shows four ranges with the percentage of the number of hours with lower marginal temperature for the development of cactus pear (< 13 °C) in the annual crop cycle in the north-central region of Mexico. In the current scenario, 89.7 % of the surface in the municipalities that produce cactus pears has between 2030 and 30-40 % of the total number of hours in the year < 13 °C. By the year 2030 with an RCP 4.5, the surface with percentage of 20 to 30 % would have an increase from 31.5 to 67.6 %, this surface would continue to increase in the climate scenarios 2050 and 2070. In contrast, the surface between 30 and 40 % of hours with temperature < 13 °C would decrease from 58.2 to 6.1 % in 2070 with an RCP 4.5. The latter scenario would be more negative by 2070 with an RCP 8.5; that is, there would be fewer hours with temperature below 13 °C (Table 5).

The decrease in the number of hours in the range of 30 to 40 % with mean annual temperature < 13 °C, could be beneficial for the cultivation of cactus pear, since daily temperatures < 15 °C delay fruit grown and maturation, but adversely, this would cause fruit with a higher proportion of peel than pulp, lower content of soluble solids and deficient pigmentation of the peel (^{Inglese et al., 1995}; ^{Liguori et al., 2006}) cited by ^{Inglese et al. (2017)}.

Average annual marginal higher temperature (> 21 °C)

88.3 % of the municipalities surface that produce cactus pear (current scenario) is in the ranges of 20-30 and 30-40 % of the total number of hours with higher marginal temperature (> 21 °C) for the annual development of cactus pear in the north-central region (Table 6). The 2030 year with an RCP 4.5 projection would consider a decrease of the surface with a percentage of 20-30 % going from 38.5 to 14.2 %, this tendency would be maintained in the climate scenarios 2050 and 2070. Similarly, the surface area in the range of 30 to 40 % of hours above 21 °C would decrease from 49.8 to 27.1 % in 2070 with an RCP 4.5; while the range of 40 to 50 % would increase, it means, there would be more hours with an annual average temperature above 21 °C. A similar prognosis would occur with an RCP 8.5, that is, the surface of the 40 to 50 % range of hours with marginal higher temperature for cactus pear would increase to 59.7 % in the 2070 scenario. The above indicates that the superior marginal temperature for the development of this crop would be presented in greater proportion by the effect of climate change, which would be negative for the cultivation of cactus pear, since, although it is a species that can support high temperatures, when these are > 25 °C they reduce the net CO₂ assimilation in approximately 28 % (^{Nobel & Hartsock, 1984}). When temperatures above 30 °C occur during the initial phase of fruit growth, the third phase of growth can be shortened, when the greatest cell elongation occurs in the edible portion, leading to early ripening and reduction in fruit size, firmness and sugar content (^{Inglese et al., 2009}). Besides, high temperatures during fruit growth reduce postharvest storage and shelf-life of cactus pear fruits, as this fruit is often more sensitive to low temperatures (< 8 °C) during storage (Inglese et al., 2002). Therefore, instead of the increase in atmospheric CO₂ (^{Nobel, 1991}), the increase in the upper marginal temperature estimated in this work would suggest, in the future, a greater limitation for the cultivation of this species (^{Nobel & Hartsock, 1984}).

Optimum average temperature for the month of January (10 to 14 °C)

Figure 4 shows the maps with the percentage of the number of hours with an average temperature of January optimal for the development of cactus pear (10 to 14 °C) in the northcentral region. In the current scenario (Figure 4A), it was observed that 82.8 % of the surface area of the producing municipalities, accumulated between 20 and 30 % of the total number of hours in January (Table 7). By 2030, a slight increase in the area with a percentage of hours of 20 to 30 % (green color) would be expected, going from 82.8 to 85.9 % (Figure 4B and Table 7). The latter means that it would be practically the same surface with temperature within the optimal range in the month of January for the development of cactus pear. In the climate scenarios 2050 and 2070, the surface would continue to increase slightly in the range of 20-30 % of hours between 10 and 14 °C in January until reaching 89.1 and 92.7 % in RCP 4.5. This indicates that the optimal temperature in January for the development of this crop would be positive and slightly modified as a result of climate change at least until 2070. Considering the RCP 8.5, the surface with a percentage of 20 to 30 % of hours with optimal temperature in January for the development of this crop would have a similar behavior until reaching 96.0 % of the surface by 2070. This last result, suggests to support the theory that global warming would favor the expansion of the surface cultivated with O. ficus-indica (^{Nobel & Bobich, 2002}; ^{García-Moya et al., 2012}; ^{Lobell & Gourdji, 2012}) because this species is more susceptible to low than high temperatures (^{Nobel, 1996}; ^{Nobel & Bobich, 2002}); but in addition, anatomically and physiologically it would adapt better to scenarios with high concentrations of atmospheric CO₂ (^{Nobel, 1991}) where other plant species of agronomic interest would not prosper (^{Lobell and Field, 2007}). However, research on the reproductive biology and physiology of this species should be considered (^{Nobel, 2002}; ^{Pimienta-Barrios & del Castillo, 2002}) under possible climate change scenarios (^{Inglese et al., 2017}).

Average temperature of the month of January sub-optimal lower (< 10 °C)

Table 8 shows five ranges with the percentage of the number of hours with an average temperature of January sub-optimum lower for the development of cactus pear (< 10 °C) in the north-central region. In the current scenario, 85.0 % of the surface of the producing municipalities has between 30 and 50 % of the total number of hours in the month of January. With reference to this last scenario, the forecast for the years 2030, 2050 and 2070 with an RCP 4.5 would indicate an increase of this area in 17.3, 30.5 and 32.9 %, respectively. In contrast, the area in the range 40 - 50 % would decrease by 15.0, 33.8 and 43.6 % for 2030, 2050 and 2070 years, respectively. Thus, the decrease of the surface in the range 40-50 % would contribute with more surface to the range 30-40 %. The above indicates that the lower sub-optimal temperature of January for the development of this crop would be positively influenced by the effect of climate change and would be maintained at least until 2070. Considering the RCP 8.5, the surface with a percentage of 40 to 50 % of hours with sub-optimal temperature in January for cactus pear would have the same behavior, but a greater decrease in this surface in the 2070 scenario. In other words, the lower sub-optimal temperature in January would decrease, which could promote the development of cactus pear by decreasing the risk of frost (^{Nobel, 1996}). However, the moderate lack of winter cold could reduce the number of reproductive buds, which would result in loss of productivity (^{Nerd & Mizrahi, 1995}). Temperature increases in producing regions of the world are a constant concern (^{Inglese et al., 2017}).

Average temperature of January sub-optimal higher (> 14 °C)

Table 9 shows five ranges of percentages of the number of hours with an average temperature of January sub-optimal superior (> 14 °C) for the development of this plant in the north-central region. In the current scenario, 84.3 % of the surface of the producing municipalities are located in the ranges 30-40 and 40-50 % with the number of hours with an average temperature of the month of January sub-optimal superior (> 14 °C). By 2030, with an RCP 4.5, the surface area will decrease from 30 to 40 %, from 25.4 to 7.7 %. This trend would continue steadily in the 2050 and 2070 climate scenarios and would be dramatic for all three RCP 8.5 scenarios (Table 9). In contrast, the surface of the 40 to 50% for hour range above 14°C would increase from 58.9 to 61.7 % in 2030, and then decrease in the 2050 and 2070 scenarios with RCP 4.5. The range 50-60 % of hours, whose current scenario had 12.3 % of the surface of the cactus pear fruit-producing municipalities, would increase to 60.9 % in 2070 with an RCP 4.5 and this behavior would be similar with an RCP 8.5. This indicates that the sub-optimal upper temperature of January for the development of this crop would be influenced by the effect of climate change and would increase in the future. In other words, an increase in the January upper sub-optimal temperature could negatively affect crop productivity due to the absence of low temperatures during the winter (^{Nerd & Mizrahi, 1995}).