Species studied

Salvia hispanica L. (chia), a species that formed an essential part of the Mesoamerican culture, with a wide geographic distribution and a high value as input for human consumption.

Databases and GIS. We used the monthly and annual precipitation data, maximum, minimum and average temperature for the periods 1961-1990 (climatology of reference) and from 2040 to 2069, to determine potential areas of S. hispanica. These climate data were obtained from the data portal "Earth System Grid" (ESG) WorldClim and working through raster images with a resolution of 2.5 min of arc in the Idrisi Selva system (^{Eastman, 2012}). For the period 2040-2069 were considered the GCM: MPIM-ECHAM5, MIROC (medres) and UKMO^_HADCM3 under the emission scenario greenhouse gas A2 (^IPCC,2007).

These three models are the most used for Mexico and have shown good fit to the climatic conditions (^{Conde et al., 2006}). For determining the potential areas other diagnostic variables were also included, such as the use of agricultural land, land-slope and soil texture; which were obtained from the Environmental Information System (EIS) ofthe National Research Institute of Livestock, Agriculture and Forestry (INIFAP) (^{Díaz et al, 2012}); except for the agricultural land use and texture, land-slope was extracted from the image using soil series III of the National Institute of Statistics, Geography and Informatics (^{INEGI, 2009}).

Statistical analysis

The Kolmogorov-Smirnov test was used to test normality in the data series ofprecipitation and temperature (climatology 1961-1990) in the three altitudinal zones. The test was run through the SPSS Statistics 19 software (^{IBM Corp, 2010}). Since the test of normality reported for all the cases that the data of temperature and precipitation were not normally distributed, we proceeded to analyse the variance with a nonparametric statistical method, proposed by Kruskal and Wallis, which is also known as test H and uses ranges sample data from three or more independent populations (^{Kruskal and Wallis, 1952}). In the present study, the test H was used to identify significant differences between temperature and precipitation data from three altitudinal strata. Statistical described by the expression:

Where: j= number of samples; ni= number of observations in the sample í; N= E_m, number of observations in all the combined samples; R_i= sum of the ranks in the í sample.

In order to make the statistical analysis we used data of temperature and precipitation derived from each cell in the raster images of these variables for the three altitudinal strata studied. This information was obtained by transforming raster images to vector of points, which were exported in Idrisi Selva to ascii type file system and these were opened and manipulated in Microsoft Excel system.

Diagnostic potential areas

Potential areas were determined using a multi-criteria analysis in the Idrisi-Selva system and considering a qualitative scheme of three categories: areas with optimal ecological conditions (Op), areas with suboptimal ecological conditions (Sp) and areas with marginal agro-ecological conditions (Mg). Established with optimal ecological conditions, the areas in which all the environmental factors analysis were at an optimum level for forage species; as suboptimal areas in which at least one variable of diagnosis was found in non-optimal conditions (sub-optimal or supra-optimal) for growing; finally taken as marginal areas where at least one of the diagnostic variables held values that restricted the development of forage species under study.

The information was obtained to establish categories of literature review reported by ^{Hernández and Miranda (2008)} ; ^{Jamboonsri et al. (2012)} ; ^{Capitani (2013}). The Table 1 shows the intervals of three variables used to diagnose potential areas. The slope of the ground for the species was assessed by assigning optimal condition for slopes 0-8%, suboptimal condition for slopes 8-20% and marginal condition for slopes larger than 20%. The analysis was performed considering potential areas as diagnostic only surface areas for agricultural use.

Table 1 Agro-ecological intervals to diagnose potential areas of S. hispanica.  

Op= óptima; Sp= subóptima y supraóptima; Mg= marginal; Gr= gruesa; Md= media; Fn= fina.

Statistical analysis

As shown in Table 2, and according to the results of the Kolmogorov-Smirnov test (p< 0.0001), the temperature and precipitation data have a normal distribution in any of the studied altitudinal strata.

Table 2 Results of normality test Kolmogorov-Smirnov for temperature and precipitation data from three altitudinal strata. 

The Table 3 shows the basic statistics of temperature and precipitation by altitudinal strata. According to the results of the Kruskal and Wallis method, both the temperature and precipitation vary significantly (p< 0.001) between the three altitude layers, so that all three can be regarded topographical regions climatically different.

Table 3 Basic statistics of temperature and precipitation in three altitudinal strata. 

Climate changes in the agricultural areas of Mexico

The temperature variation ranging from 14.3 °C in the highlands to 22.8 °C in the lowlands, in combination with the variation of precipitation, produces great environmental diversity in agricultural areas of Mexico. Temperature projections of the three GCM show a heat increase in a range from 2.6 to 2.9, 2.7 to 3.3 and 2.4 to 3.1 °C at low, intermediate and high-lands, respectively (Table 4) passing the 1961-1990 period from 2040 to 2069 areas, which translates into a warming per decade of0.32 to 0.37, 0.34 to 0.42 and 0.30 to 0.39 °C. This resembles the increase recorded by ^{Brohan et al. (2006)} and ^{Trenberth et al. (2007)} , who reported that, the temperature of the Earth's atmosphere per decade between 1979 and 2005, increased about 0.268 ± 0.069 °C.

Table 4 Average annual values of mean temperature and precipitation accumulated in two climate scenarios for three altitudinal strata of the agricultural area. 

The warming projected for the region of study is quite important, as some areas will vary its temperature regime; such as the areas of intermediate elevation which will change from a temperate condition (12-18 °C mean annual temperature, ^{García, 1988}) to a semi-warm conditions (18-22 °C, ^{García, 1988}; ^{Medina et al, 1998} ), which will have positive effects on the optimal conditions surface for the cultivation of tropical and subtropical species (^{Ruíz et al, 2011}). However, the temperature increase is considered harmful to current cropping patterns, since an increase in mean seasonal temperature can advance the time of harvest of the current varieties, and thus reducing the final yield; even if no adaptation measures are implemented (^{Gornall et al, 2010}). In the lowlands, which already maintain a warm temperature, near the maximum physiological thresholds of the crops, the temperature increase projected in the present study might be harmful, due to the increasing of heat stress and water loss by evaporation (^{Gornall et al., 2010}).

In the lowlands, which already maintain a warm temperature, near the maximum physiological thresholds of the crops, the temperature increase projected in the present study might be harmful, due to the increasing of heat stress and water loss by evaporation (^{Gornall et al., 2010}).

Regarding precipitation, the projections of the three GCM do not maintained a coincidence as high as in the case of temperature, because while the MIROC (medres) and "UKMO^_HADCM3" models reflect a decrease in the precipitation of 13 to 6% in low areas, the model "MPIM-ECHAM5" projects a slight increase of 0.7% in annual rainfall (Table 4). Meanwhile, for the intermediate zones an increased is expected for the annual rainfall from 2 to 4%, with the models "MPIM-ECHAM5 and UKMO_ HADCM3" respectively, and a decrease of 9.5% with the model "MIROC" (medres) projects. In the highlands a decrease in precipitation is also expected using the models "MPIM-ECHAM5 and MIROC" (medres) of2 and 12% and, a 5% increase with the model UKMO_HADCM3. This lack of consistency in the modelling offuture precipitation by the different GCM has been previously reported (^{IPCC, 2007}) and is accentuated in desert and semi-desert areas (^{Johnson and Sharma, 2009}), a condition that prevails in the Mexican territory. This is an important aspect because precipitation is a relevant variable for hydrometeorological assessments and crop productivity (^{Kumar et al., 2004}; ^{Sivakumar et al., 2005}). Even small changes in precipitation can affect productivity (^{Lobell and Burke, 2008}).

Potential areas for S. hispanica in agricultural areas of Mexico

The optimum conditions for growing chia are located primarily in low and intermediate elevation lands during the reference period (Figure 1), demonstrating that, the climatic conditions of the areas between 0 and 2 200 m have better advantages for agro-climatic requirements of the species (^{Jamboonsri et al., 2012}; ^{Capitani, 2013}). These areas are located primarily in the States of Jalisco, Sinaloa, Nuevo León, Tamaulipas, Guerrero, Puebla, Oaxaca, Nayarit, Morelos, Veracruz, Michoacán, Yucatán and Chiapas.

Figure 1 Potential areas for Salvia hispanica under climatologies: a) 1961-1990; b) 2040-2069 UKMO_HadCM3; c) MPIM-ECHAM5; d ) MIROC (medres ) with scenario A2 greenhouse gases. 

For this reason, these regions offer an important opportunity for improving human nutrition since chia provides a natural source of omega-3 fatty acids, antioxidants and dietary fiber; and the ability of the seed to enrich several products being added directly to food or as part of the diets of animals, providing an opportunity to develop products with the seed and venture into novel functional product markets or "nutraceuticals" (^{Hernández and Miranda, 2008}).

In the maps of Figure 1, the effect of climate change on potential areas of chia can also be observed. In this sense, the predictions of the potential areas of GCM show that, the optimal surface increase in highlands and intermediate elevation with respect to the climatology of reference, in the States of Jalisco, Michoacán, Guerrero, Puebla, Morelos and Oaxaca; decreasing in lowlands, mainly in the States of Sinaloa, Tamaulipas, Chiapas and disappearing in Yucatán.

The climate predictions of the three GCM defined potential areas very similar to the optimum surface in the three altitudinal strata (Figure 1). However, the potential areas established with the GCM: MIROC (medres) was significantly different for lowlands compared to the other stages. These results are attributable mainly to the differences ofthe simulations of precipitation ofthe GCM (Table 4). This fact evidence that, the potential growing are indeed sensitive to the climatic variations ofthe simulation of the GCM, even among the models considered similar on the prediction of climate change for Mexico (^{Conde et al., 2006}).

A similar trend of the GCM was also observed in determining suboptimal surfaces. The Table 5 shows that there is a tendency to decrease in the lands of intermediate elevation, mainly in Jalisco; and increase in highland and lowland areas, located in the State of Mexico, Tlaxcala, and Yucatán.

Table 5 Potential areas for S. hispanica in three altitudinal strata and two climate scenarios in Mexico 

Sp=subóptimas

Regarding the marginal areas, there is a decrease between 5 and 10% with MIROC (medres) and UKMO_HadCM3 models, respectively, with respect to the climatology of reference therefore climate change during this century will have a positive impact on chia production in Mexico.

The dynamics ofidentified potential areas to climate change expects the emergence of optimal land areas ofintermediate elevation and also in the highlands of Central and South West regions of the country and disappear in the lowlands in the northeast and in the Yucatan Peninsula (Figure 1). These regions have already been reported with changes in cropping patterns due to climate change (^{Ramírez et al., 2011}; ^{Santillan et al., 2011}; ^{Mardero et al., 2012}).

These results confirm that, the presence of climate change on the optimal potential growing areas ofchia move towards the high and intermediate lands in the future. Therefore, a redistribution of farmland will likely take place in these areas ranging in the same elevation, in the present century. Considering that today, the lands of intermediate elevation include 45% of the agricultural area, this redistribution would not be a high impact; however, when climate change impulse crops to migrate to higher lands, a difficult situation will be presented, since currently only 18% of the agricultural areas are located within the Highlands.