Historical climate information

It worked with daily and monthly precipitation and temperature data for the period 1961-2010, from 133 weather stations in the state of Jalisco (Figure 1). In addition solar radiation data for the period 1984-2014, which were obtained from the website of climatological resources from NASA (^{NASA, 2015}) were used.

Figure 1 Spatial distribution of 134 weather stations in the state of Jalisco, Mexico. 

Climate change information

Climate change information used in this study consisted of monthly average values of maximum temperature, minimum temperature, average temperature and precipitation for the periods 2041-2060 (represented by year 2050) and 2061-2080 (represented by year 2070) under representative concentration paths of greenhouse gases (Rcp) 4.5. This information derives from the application of the assembly model reported by Ruiz et al. (2016) for Mexico and were used in this study through raster images with a resolution of 30"arc. The assembly model considers the consensus of 11 general circulation models (GCM) reduced in scale and calibrated (^{Walton et al., 2013}.): and belonging to CMIP5 (Coupled Model Intercomparison Project Phase 5) BCC-CSM1-1, CCSM4, GISS-E2-R, HadGEM2-AO, HadGEM2-ES, IPSL-CM5A-LR, MIROC-ESM-CHEM, MIROC-ESM, MIROC5, MRI-CGCM3, NorESM1-M.

Determination of the growing season with historical climate information

With climate data for the period 1961-2010 the growing season of the 134 localities studied were determined. Such determination was made only considering the criterion of moisture availability in the soil and it was assumed that the occurrence of frost does not interfere with the growth period under rainfed conditions. The calculation of the EC was made with the method from FAO, which determines the beginning of EC when the rainy season starts, p≥ 0.5 ETP, where P is rainfall and ETP is the potential evapotranspiration or of reference (^{FAO, 1978}). In this study the precipitation was considered in terms of the expected monthly rainfall at 70% probability, for which the probability of rain was calculated with the theoretical distribution that best fits to each series of monthly precipitation data; for this procedure SICA 2.5 (^{Medina et al., 2003}) program was used, which considers normal distributions, log-normal and incomplete range. Monthly ETP values were estimated using the Penman-Monteith method using the ETo Calculator program Ver. 3.2 (^{FAO, 2012}). The end date of EC was estimated when p< 0.37 ETP (^{Ruiz, 1994}). Once all the above calculations are made, the values of probabilistic rain and ETP obtained were emptied into an Excel spreadsheet through formulation and plotting to obtain the beginning date of the growing season (FIEC), the end date of the growing season (FFEC) and duration of the growing season (DEC) for each of the weather stations under study.

Modeling parameters of the growing season

In order to determine the parameters of the growing station, scenarios for climate change and for climatology 1961-2010, proceeded to generate models to estimate FIEC, FFEC and DEC. For this were analyzed as potential explanatory variables (variables associated with the variation of FIEC, FFEC, DEC) monthly, seasonal and annual normal values of precipitation, mean temperature and precipitation ratio/average temperature. Explanatory variables with a higher correlation coefficient regarding FIEC, FFEC and DEC to form models were selected.

Evaluating impact of climate change on the growing season

With the predictive models obtained for FIEC, FFEC and DEC generated state maps of these variables for the years 2050, 2070 and 1961-2010 (reference climatology).

Determining the growing season

The results from the calculation of probability of rain and estimation of potential evapotranspiration allowed to estimate the basic parameters from the growing season in the 134 localities studied (Figure 2).

Figure 2 Variation of the growing season in 134 localities from the state of Jalisco, Mexico. 

All localities analyzed showed growing season, except for the town Ojuelos, which showed probabilistic precipitation values that were not enough to overcome the ETP value required to begin EC. In the graphs from Figure 2 in can be observed this fact since the bar from this site remains at axis level "X". This locality and region from the state had been previously reported with zero days of growing season by other authors (^{Ruiz et al., 2000}). As can be seen in Figure 2, the parameters of the growing season show variation among sites, which is the result of varying weather conditions both in space (^{Medina et al., 1998}) and time (^{SMN, 2015}) through their different agro-ecological regions.

Modeling growing season

Figure 3 graphically depicts the relationship of the variables that led to models to estimate the parameters of the growing season. By analyzing the correlation coefficients indices from the independent variables considered, with the parameters of the growing season, it was found that the beginning date (Julian day) of the growing season was the parameter that best correlated with the independent variables. Within these correlations, the variable precipitation ratio / average temperature from June (CPT6) correlated best with beginning date of the growing season (DjIEC) (r= -0.87; significant p< 0.05). In the upper graph from Figure 3 it can be observed that the relationship between these two variables is logarithmic, which gave rise to the following estimation model:

1)

Where: DjIEC= julian day of the beginning date of the growing season; Ln= natural logarithm, CPT6= precipitation ratio/average temperature for June.

Figure 3 Relationship between CPT in June and beginning date of the growing season (above), and the relationship between the beginning date and duration of the growing season (below). 

To estimate the duration of the growing season it was observed that the explanatory variable that best correlated with this parameter was precisely the beginning date of the growing season (Figure 3 lower graph; r= -0.85, significant p< 0.05), being this relationship of exponential type, so the estimation model obtained was:

2)

Where: DEC= duration of the growing season (days)

As for the end date of the growing season, there were no high and significant correlations with any independent variable, so it was decided to estimate this variable as follows:

3)

Impact of climate change on the growing season

With the application of expressions (1), (2) and (3) on normal climatological images for the periods 1961- 2010, 2050 and 2070, the state images of DjIEC, DjFEC and DEC were obtained. Figure 4 shows the expected changes for DEC by 2050 and 2070 regarding reference climatology 1961-2010. As can be seen, the spatial distribution of DEC shows a gradual decrease of it at the expense of more favorable intervals. That is, the gradual reduction of the interval 150-180 days to almost nothing in 2070. At the other extreme, the condition of zero days of DEC currently prevailing in the municipality of Ojuelos, won’t improve with the advance of climate change, this municipality will continue without DEC days. According to the corresponding climatology 1961-2010 from Figure 4, currently the predominant DEC interval is 120 to 150 days however by 2050 this predominance will fall to an interval of 90-120 days; this is that climate change will cause a reduction of this parameter.

Figure 4 Expected changes for the duration of the growing season in 2050 and 2070 regarding reference climatology (1961-2010). 

This reduction will accentuate by 2070, as the territorial coverage of the interval 120-150 days will decrease even more, while the interval 90-120 days will gain more ground, causing even the appearance by 2050 of a very visibly area with DEC <90 days in the central part of the coastal region from the state, which will advance territorially by 2070 (Figure 4). This could be explained by a possible regional evolution towards drier weather with climate change, this is the prevalence of atmospheric conditions that do not promote precipitation. The region is currently classified between semi-dry climate BS₁ and warm humid Aw0 (^{Garcia, 2004}), so that the weather would tend more to BS₁ with climate change from the twenty first century.

This panorama of reduction of DEC in the XXI century, points out to an adverse scenario that climate change will bring to agricultural productivity in the state, situation that has been previously pointed out for low-latitude regions (^{Parry et al., 1999}). The reduction in DEC results in a reduction of crop yields (^{Kang et al., 2009}); however, the agricultural areas of the state with high storage capacity of soil moisture could cushion this yield loss to some extent, so that the impact of climate change would not feel so significantly.

A strategy to cope with this reduction in DEC could be the development of new varieties, earlier but with yields equal to or higher than current, which is a challenge, as it usually shorter maturity cycles produce lower grain yields and dry matter. Hence other options to adapt to these new DEC conditions could be changing crop pattern, sowing species with lower water requirements and growing season, and promoting practices that improve storage capacity of moisture.

The reduction of DEC days in the State is due to changes occurring at beginning and end date of the growing station. In Figures 5 and 6 can be seen the changes that climate change will bring by 2050 and 2070 for FIEC and FFEC respectively. Figure 5 shows that the earliest dates of IEC will be replaced territorially by latest dates. So areas with IEC before June 15 will virtually disappear by 2070. The current dominant FIEC is from June 15 to 30, which barely will maintain for 2050 but for 2070 will the predominant FIEC will be from June 30 to July 15. In Figure 6 is looks like the most favorable interval of FFEC, that is, November 15 to 30, by 2050 will be reduced and by 2070 will disappear. In addition, the most representative interval of FFEC in Jalisco is October 31 to November 15, which by 2050 will decrease in surface 2050 and by 2070 it will be displaced as the most representative by a FFEC prior to October 31.

Figure 5 Expected changes for beginning date of the growing season in the years 2050 and 2070, regarding reference climatology (1961-2010) 

Figure 6 Expected changes for end date of the growing season in the years 2050 and 2070, regarding reference climatology (1961-2010). 

In Figure 7 it can be shown more clearly the negative effects from climate change by 2050 and 2070 on EC parameters. In this figure appear the maps that show regions with different levels of absence of DEC days. By 2050 this loss is predominantly from 0 to -14 days, a situation that allows visualizing the type of agro-climatic conditions which will face the state agricultural production systems. By 2070, the loss in predominant DEC ranges between -7 and -14 days, depending on the region from the state, but with considerable territorial presence of other more drastic losses such as those ranging from -14 to -28 days in the southwestern portion of the state, where the shortening of EC will be alarming.

Figure 7 Loss in days for: DEC in 2050 (a) and 2070 (b); for: IEC (delay) in 2050 (c) and 2070 (d); as well as for: FEC (advance) in 2050; and (e) and 2070 (f). 

The second pair of maps in Figure 7, describes the loss in days for growing season, due to a delay in the beginning of EC. So by 2050 the delay in IEC will be from 0 to 21 days, whereas in 2070 it will be from 0 to 28 days.

The third pair of maps in Figure 7 refers to the loss in days of the growing season, but now caused by an advance on the end date of EC. As can be seen in these maps, the impact of climate change will be lower on FEC, this is that their impact on moving the end date of EC will be less significant than the effect that this phenomenon will have on the delay of the beginning date of EC. Hence the effect of climate change in the reduction of DEC both in 2050 and 2070, will focus mainly on a delay in the beginning of EC.

With the above, it is necessary to start designing strategies to reduce such effects of climate change. These strategies should address to use and management of water and soil, including the reduction of intensive tillage; dry plantings (where feasible); breeding crop varieties aimed at developing shorter cycle and tolerant climatic stress; crop rotation, strip cropping with different moisture requirements and others.