Habitat Suitability of Fig (Ficus carica L.) in Mexico under Current and Future Climates

Abstract

Emerging or alternative crops are those that have been recently introduced in response to new demands and commercial opportunities. The fig has been introduced as an alternative fruit crop in Mexico due to its high nutritional, nutraceutical, and antioxidant capacity. A total of 644 points of presence were downloaded and filtered according to climatic ranges and agricultural areas, leaving 68 records. The MaxEnt algorithm was used to develop the habitat suitability models for current and future climate. The bioclimatic variables of the global circulation models, Hadley Centre Global Environment Model version 2-EarthSystem (HADGEM2-ES) and Max Planck Institute for Meteorology-Earth System Model Low Resolution (MPI-ESM-LR), were used under scenarios 2.6 and 8.5. The changes in the fig area for Mexico were analyzed based on the generated models. Under the current climate, 359,575.76 km² were estimated for 2050, and a loss of area for the excellent category was estimated for both models and scenarios; however, for the MPI-ESM-LR model projected to 2070, an increase of 5.51% and 0.39% was estimated for scenarios 2.6 and 8.5, respectively. The effect of climate change on agronomic species such as figs will be expressed in variations in climatic ranges and areas suitable for their development. The results of this study reveal the negative and positive effects of climate change on fig habitat suitability in Mexico. The dynamics of changes in surface area will be reflected mainly in northern and central Mexico.

1. Introduction

Soil-based agricultural production is the main source of food for all countries in the world [1]; therefore, it requires specific attention to maintain its productivity due to the remarkable diversity in size, complexity, economy, and type of crop in each country [2,3].

The increase in temperature reduces crop production, availability, and distribution and facilitates the proliferation of weeds and pests, affecting human well-being [4,5,6]. By 2100, the mean sea level is projected to rise 0.1 m, while the temperature will increase from 1.5 °C to 2 °C from intensified anthropogenic activities [5,6,7].

Environmental problems are complex and require a combination of new methods and tools, such as geographic information systems, which allow handling greater amounts of data to develop and generate knowledge through cartographic products [8]. Ecological Niche Modeling is a useful tool for risk assessment potential development of plant species as a response to environmental conditions [5]. This model is based on computer technology development, which allows a methodological approach for an extensive question solution in different fields, improving their interpretations [9].

Emerging or alternative crops have been recently introduced in response to new demands and commercial opportunities [10]. Alternatives crops have characteristics that allow them to replace or complement the crops traditionally produced and to improve profits from either higher commercial price, higher international demand, or both [11].

The fig, a fruit tree native to the Mediterranean basin, has expanded rapidly throughout the world due to its adaptability to different soils and climates, and the economic importance of its fruit, due to its excellent source of nutrients and antioxidants) [12,13], and its varied consumption option (fresh, dried, or processed) [14].

Bare-root fig plants can be transplanted during the dormant season, from December to the end of February [15]. However, asexual propagation, by rooting stem cuttings, is the main method of propagation of this crop because it is a simple method, which allows to multiply and obtain in a relatively short time, homogeneous plants of good commercial quality [16].

De Fina (1979) points it out as one of the fruit trees with greater plasticity to behave satisfactorily between 23°19′ and 40°50′ Latitude [17] and at elevations of 900–1800 m above sea level [18], being a typically Mediterranean species, which prefers subtropical climates with warm winters and dry and cool summers, but it is also able to withstand non-extreme cold [19].

Fig represents a potential alternative crop because of its tolerance to salt and its adaptability to climatic and edaphic conditions in semi-arid areas. This fruit tree is traditionally considered rainfed since, under these conditions, it produces sweeter fruit, but in case of extreme drought, its harvest will be null, and it will have a reduced number of small leaves [17,20]. Excess irrigation in this fruit tree is detrimental because it produces figs that are too thick, very watery, rot easily, and are difficult to dry. In addition, the tree is quite sensitive to root rot [20,21].

Mexico, with favorable climatic conditions, where July is the warmest month, with an average of 26.7 °C, and December is the coldest, with an average of 16.3 °C [22], has the potential to cultivate fig trees with parthenocarpic fruits in the absence of their natural pollinator (Blastophaga psenes) [23]. The most common varieties are Mission Black, Black Turkey, Sierra, and White Kadota, which are characterized by being bifeda; that is, they produce fruit at two times of the year, from June to July, early figs and from August to September, figs, and the tiger, which only produces figs (unifeda), although Mexican figs are generally believed to be of the European Black Mission variety [23,24,25].

Research on fig dorp has been developed in different aspects; in India, fig crops are more common in arid areas [26], and its range in the northwest [27] in Africa Fig (Ficus carica L.) vulnerability to climate change: Combined effects of water stress and high temperature on ecophysiological behavior of different cultivars [28], in Palestine Impact of bioclimatic and climatic factors on Ficus carica L. yield: increasing the economy and maintaining the food security of Jerusalem [29], while in Mexico research has focused on the evaluation of crop establishment and fruit quality [30,31], on the analysis of biostimulants and nutrients [32] and on the seasonal variation of foliar nutrient concentration [33]; however, there is no research related to climate change and fig cultivation in this country.

According to data from FAOSTAT (2020), in 2018, the world production of figs exceeded one million tons (t), with Turkey being the main producer with almost 270 thousand t. Mexico has recently increased fig production to 7 thousand tons, obtaining half from Morelos state [34,35].

The MaxEnt program is a specialized algorithm that is frequently used to predict the current and future habitat suitability of various species. This is a program that employs absence-presence records and has been widely used to estimate the habitat suitability of species of various biological groups [36], useful for the development of programs focused on conservation and/or ecological restoration [37,38,39].

Analysis of potential areas for the establishment of crops such as figs is necessary, as it is important for the sustainable agricultural planning of the species, the conservation of natural ecosystems, and the achievement of food security [40].

The determination of climate change scenarios aims to determine the impacts on land properties [41,42]; therefore, the objective of this research was to determine areas of fig (Ficus carica L.) crop habitat suitability under current climate conditions and climate change scenarios for the 2050 and 2070 horizons with the HADGEM2-ES and MPI-ESM-LR models for the RCP 2.6 and 8.5 scenarios.

2. Materials and Methods

2.1. Study Area

The study area was delimited by the extreme coordinates 32°17′44.52″ latitude and 92°30′00″ longitude, with an estimated surface area of 721,990 km² and an elevation gradient from 0 to 5417 m above sea level. Precipitation ranges from 800 to 1600 mm per year, with an average of 1200 mm, while the mean annual temperature ranges from 16 °C to 25 °C [43].

2.2. Fig Geographic Records

A total of 644 fig crop occurrence records for Mexico were obtained from the Global Biodiversity Information Facility database [44]. A first filter was performed to eliminate data that were within 1 km of each other using CONABIO’s NicheToolBox platform [45]. Then, a second filter was conducted using ArcGIS ver. 10.8 software [46] by the vector layer of land use and vegetation series VII of INEGI from CONABIO [47], the points of crop presence within agricultural areas were selected, reducing the number of data points to 68. This process avoids spatial autocorrelation and overestimation in distribution models [48,49].

2.3. Current and Future Climate Variables

Climate data were obtained from the WorldClim v. 2.0 database [50], with a resolution of ~1 km² (30 arc seconds) per pixel, with global average information from 1970 to 2000. For future climate, information from CMIP5 (Coupled Model Intercomparison Project Phase 5 2013) was downloaded for the periods 2041–2060 (2050 horizon) and 2061–2080 (2070 horizon), with a spatial resolution of 30 arc seconds, using the Global Climate Models (GCMs) HADGEM2-ES and MPI-ESM-LR, models recommended by the National Institute of Ecology and Climate Change for the Mexican Republic [6], using the Representative Concentration Pathways (RCP) 2.6 y 8.5.

2.4. Topographic and Edaphological Variables

Topographic elevation information (ELE) was extracted from WorldClim 2.0 with a spatial resolution of 30 arc seconds (~1 km²), while edaphology data were obtained from the CONABIO information portal [4]. Values for sand (ARE), silt (LIM), clay (ARC), rock fragments (FR), pH (PH), and organic matter (MO) were downloaded at 90 cm depth, with a resolution of 90 m², that were re-scaled to 30 arc seconds (~1 km²). All variables were then transformed into ASCII format.

2.5. Variable Selection

To select the variables, a minimum convex polygon (MCP) was generated with the points of presence of the crop in agricultural areas. In total, 10,000 background points were positioned based on a non-discriminatory test (without duplication) of information relevant to the environmental heterogeneity of fig tree climatic information, and information was extracted for each variable according to Becerra-López [51].

To avoid autocorrelation between variables, those with a correlation higher than 0.70 (p < 0.05) were excluded, thus avoiding the effect of multilinearity between variables [52]. The selected variables were determined at the same spatial resolution of 30 arc seconds (~1 km²), taking into account the minimum convex polygon (Figure 1), which covers the geographic space where the species is recorded and delimits its biology and distribution [53].

2.6. Model Calibration

The model was calibrated using the standardized Akaike information criterion coefficient (AICc), with the lowest value fit selection [54]. This calibration was performed with the ENMeval library [55] in R statistical environment v. 3.5.3 [56].

2.7. Modeling of the Current Climate

The MaxEnt ver. 3.4.4 software algorithm was used [36]. The variables selected after cleaning were BIO5, BIO9, BIO12, BIO17, MO, ELE, FR, and ARC (

Table 1. Variables used in ecological niche modeling in MaxEnt.

Variable	Description	Range (Unit)
BIO12	Annual precipitation	54–4816 (mm)
MO	Organic matter	1–3 (%)
BIO9	Mean temperature of driest quarter	−2.4–28.5 (°C)
ELE	Elevation	0–5417 (msnm)
BIO17	Precipitation of driest quarter	0–335 (mm)
FR	Rock fragments	1–3 (%)
BIO5	Max temperature of warmest month	5.4–42.4 (°C)
ARC	Clay	1–3 (%)

). The MaxEnt setup criteria consisted of internal replication by cross-validation (1000 iterations), logistic type output (100 repetitions), and a convergence threshold of 0.0001, where 75% of the records were selected for training the model and 25% for validation [57].

2.8. Modeling under Climate Change Scenarios

To obtain the habitat suitability models under climate change conditions, the calibration parameters of the model with the best fit from the current climate to future climate were transferred to MaxEnt v. 3.4.4 [58].

To estimate the area of current and future fig tree habitat suitability zones, a reclassification was performed with ArcMap software ver. 10.8 through the reclass tool [46] of the continuous values in each of the projections, using the four-level Likert scale for suitability (poor, fair, good, and excellent) [59].

2.9. Validation of the Models

The distribution models were evaluated using the statistical test of the area under the standard ROC curve (AUC), which ranges from 0 to 1, where values of 0.7–0.9 indicate a good fit and values above 0.9 classify it as excellent. Additionally, the models were evaluated with the partial ROC, with values ranging from 1 to 2, where a mean value of radius 1 represents a random model, and a value of 2 represents the best model performance [60] and with the TSS statistic, which considers the specificity and sensitivity of the model [45].

2.10. Percentage Change in Surface Area due to the Effects of Climate Change

The trend of surface area change due to the climate change effects was quantified using the equation proposed by Gutiérrez and Trejo [61], both RCP 2.6 and 8.5 scenarios, considering the current climate as a reference, as well as the change between scenarios.

$$\% Change=\left[\frac{S_1-S_0}{S_0}\right]\times 100$$

where S₀ is the total area, according to the baseline scenario, and S₁ is the area under climate change conditions.

2.11. Relevant Variables in the Modeling of Habitat Suitability

The contribution of environmental variables to the present and future distribution of the species was estimated by the Jackknife test, identifying the percentage contribution of the variable to the development of the crop in Mexico [36].

To statistically validate the partial ROC ratios, z-tests of the models were performed, and the best model with the highest partial ROC test value, lowest standard error, and reliable z-value was selected. The results of the selected models for each period were produced as distribution maps in ArcMap 10.8 software [46].

3. Results

Habitat suitability modeling for the current climate. For the 100 replicates performed of the current habitat suitability models, AUC values were obtained between 0.918 and 0.924 for training and from 0.831 to 0.981 for validation, indicating that the models performed adequately. The partial ROC value for the best model was 1.85, with an AUC of 0.924 for training and 0.846 for validation (Standard deviation = 0.0001).

The estimated area of fig cultivation for the current climate was 719,097.84 km² and is located mainly in the north, through the Sierra Madre Occidental, to the Neo volcanic Transversal Axis (Figure 2). The states with the largest surface area were Chihuahua (86,978.20 km², 12.10%), Jalisco (35,346.02 km², 4.95%), Michoacán and Coahuila with 33,302.92 km² (4.62%) each. The values of surface area by the state for current climate can be found in the Appendix A.

3.1. Habitat Suitability Modeling under Climate Change Effects (RCP 2.6 and 8.5)

The values obtained from the standard ROC AUC test for the HADGEM2-ES and MPI-ESM-LR global circulation models for the periods 2050 and 2070 were 0.918 to 0.924 for model training and 0.831 to 0.981 for validation (Figure 3a–d and Figure 4a–d). The partial ROC values for both general circulation models and the two RCP 4.5 and 8.5 scenarios were between 1.81 and 1.83, in all cases significant (p < 0.01).

3.2. Relevant Variables in Current and Future Distribution

The most relevant variables for fig crop distribution in Mexico for the current climate were BIO12, MO, BIO9, ELE, BIO17, and FR, which explain a total variability of 92.1% of the model. The most important variables for the HADGEM2-ES and MPI-ESM-LR model for RCPs 2.6 and 8.5 projected to the 2050 and 2070 horizons were BIO12, MO, BIO9, ELE, BIO17, and FR, which explain a total variability of 95.4%.

3.3. Percentage Change of Fig Area in Mexico

The areas classified as optimum for the establishment of fig cultivation in Mexico for the current and future climate are shown in

Table 2. Optimum suitable area for fig cultivation under current and climate change scenarios.

Scenario	Area km2
Current	359,575.76
HADGEM2-ES 2050 (RCP 2.6)	344,784.20
HADGEM2-ES 2070 (RCP 2.6)	330,195.82
HADGEM2-ES 2050 (RCP 8.5)	307,785.98
HADGEM2-ES 2070 (RCP 8.5)	338,323.05
MPI-ESM-LR 2050 (RCP 2.6)	312,171.10
MPI-ESM-LR 2070 (RCP 2.6)	305,834.01
MPI-ESM-LR 2050 (RCP 8.5)	313,462.54
MPI-ESM-LR 2070 (RCP 8.5)	353,942.12

. The analysis of the percentage change in the surface area predicted a decrement for all possible scenarios, finding a percentage loss considering the current climate as the base period of 4.11% when considering the HADGEM2-ES model (2050; RCP 2.6), 8.17% for HADGEM2-ES (2070; RCP 2.6), 14.40% for HADGEM2-ES (2050; RCP 8.5), 5.91% for HADGEM2-ES (2070; RCP 8.5), 13.18% for MPI-ESM-LR (2050; RCP 2.6), 14.94% for MPI-ESM-LR (2070; RCP 2.6), 12.82% for MPI-ESM-LR (2050; RCP 8.5) and 1.56% for MPI-ESM-LR (2070; RCP 8.5). Habitat suitability maps for Mexico considering future climate are shown in Figure 3a–d and Figure 4a–d.

According to the future projections of the two models and the scenarios analyzed, the current fig distribution areas prevail in all scenarios (Appendix A), increasing in some regions and decreasing in others (

Table 3. Percentage change in fig crop area under different climatic scenarios.

Current		HADGEM2-ES 2050		HADGEM2-ES 2070		MPI-ESM-LR 2050		MPI-ESM-LR 2070
		2.6	8.5	2.6	8.5	2.6	8.5	2.6	8.5
km2			%
Unsuitable	1,237,798.81	1.10	2.94	1.37	10.10	5.70	3.64	−3.62	0.22
Mod. Suit.	367,165.52	0.32	4.18	−5.18	−3.74	−6.30	0.27	−12.62	0.16
Suitable	230,353.83	−2.51	−12.35	−13.51	−9.77	−13.12	−13.02	−19.10	−4.64
Optimum	129,221.93	−6.97	−18.06	1.96	−1.29	−13.29	−12.48	5.41	0.39

).

4. Discussion

4.1. Spatial Modeling

Sixty-eight points of fig occurrence were used in the present study subsequent to debugging. According to Stockwell and Townsend [62], accuracy for habitat suitability modeling increases when the number of records is greater than fifty. The results found in this study allowed the identification of areas with environmental characteristics suitable for the establishment of fig crops, which agreed with the geographical areas of fig production in Mexico [54]. Methods have been developed to model habitat suitability for the current climate [63,64,65] and for climate change scenarios [66,67,68,69,70]. However, these studies have focused mainly on timber and non-timber forest species, while there is scarce or no information on agricultural crops.

4.2. Relevant Environmental Variables

BIO12 variable (annual rainfall) was the most relevant variable for the fig habitat suitability model in Mexico, with a contribution of (23.2%). The variable presented ranges from 36–1073 mm for Mexico, which according to the crop water requirement, can allow unrestricted development.

According to Interián-Ku et al. [71] and Rocha-Loredo [72], annual accumulated precipitation has been demonstrated to be important for deciduous trees because high precipitation leads to the vigorous growth of the tree, reduced flowering, and thick-watery figs hard to dry. The deficit of water in fig crops results in poor tree growth and a reduction of fruit size and crop yield from the reduction of cell expansion and limited stomatal opening and CO₂ uptake [20,73,74,75].

However, excessive rainfall can cause fruit cracking and the proliferation of pests and diseases as a result of high humidity [15]. This agrees with Mergalejo [20], who mentions that rainy climates or prolonged and intense periods of rain are detrimental to the quality of sycones. Annual rainfall of 800 mm spread throughout the year allows good fig production [17].

Organic matter represented 17.8% of the variability of the model; this edaphological variable is directly related to precipitation because low humidity reduces plant growth, which limits the production of food for the fauna that inhabits the deciduous forests, as well as the slow decomposition of organic matter, causing low nutrient productivity [76].

Organic matter stabilizes soil structure and favors erosion resistance, air, water, and nutrient flow through pore spaces; as organic matter increases, pore space increases, so it is considered an indicator of the volume occupied by porosity [77,78,79,80,81].

The fig tree is a crop considered a great generator of organic matter due to the drooping of leaves and ripe fruits to the soil surface; however, it is advisable to increase this contribution through fertilizers to improve the availability of nutrients, improve infiltration, aeration, and favor root development [19,20,21].

The variable BIO9 (average temperature of the driest quarter), with 17.5% of participation in the model, presented a temperature range from −2.4 to 22.6 °C, where temperatures between 32 and 37 °C aid fruit ripening and quality, conditions that occur mainly in January and the first half of February [21]. Temperatures above 38 °C can cause fruit drooping and faster ripening, while temperatures below −7 °C cause fruit loss and death of trees [20,21,82].

The edaphological variables were insignificant considering that the fig tree is an undemanding species, being able to adapt to infertile soils such as saline, semi-desert, limestone, poor and rocky soils, as well as resistant to iron chlorosis, and active limestone, tolerating levels up to 4.20 dS/m of electrical conductivity and pH of 8.5, the optimum being from 6 to 8; however, it does not tolerate drainage problems because it is sensitive to root rot [20,21,81].

4.3. Climate Change Scenarios

The models used for the scenarios under climate change have been used for the prediction of habitat suitability of vegetative species in Mexico because they show representative climatic variability for the country [6].

It is important to note that future projections of suitability refer only to environmental conditions (climate change) and not to other conditions to which the species will be exposed, such as biological interactions with other organisms (pests, competition, etc.) and threats such as deforestation, overexploitation, and dispersal [83].

Several authors have reported results with the use of the MaxEnt program for different species under natural conditions of development and under agricultural practices, such as peanut, cotton, peanut, cocoa, and passion fruit [5,69,84], among others, obtaining good predictions with few bioclimatic variables but representing more than 90% of the variability.

The results obtained in this research work for the climate change scenarios are similar to those obtained by Durán-Puga et al. [85], who studied the distribution of Morus alba L. with the HadGEM2 model for the years 2050 and 2070, observing a reduction rate of the crop in the Mexican Republic, excluding Chihuahua, Baja California, and Sonora states, where it was found an increment or this crop. Altamirano and Leon [86] and Martinez-Sifuentes (2020); also reported a reduction in the distribution of tree species among pinus greggi in the mesophilic mountain forest of Oaxaca, Mexico.

The percentage change in the HADGEM2-ES model indicates an increase in the unsuitable and moderately suitable of fig cultivation in the country by 2050 and a decrease in the adequacy and the optimum of fig cultivation in the country by 2050 for both RCP 2.6 and 8.5, while by 2070 an increase in the optimum is observed for RCP 2.6, while for 8.5 an increase in the inadequacy and a decrease in the other ranges.

Martinez-Sifuentes (2020) Prasetyo et al. [87], and Priyanco et al. [88] suggest that an increase in precipitation and temperature (0.7–2.1 °C) in all RCP of this model allows monitoring and planning strategies to minimize the negative changes that could be caused by climate change in diverse species and different environmental areas.

The MPI-ESM-LR model obtained a higher aptitude in the suitability for the most catastrophic scenario (RCP 8.5) for the year 2070 than the observed for 2050 and 2070 by the HadGEM2-ES model with RCP 4.5; these results agreed with the reported by Martínez-Sifuentes [69] and Vargas-Piedra et al. [89].

In the percentage change for the MPI-ESM-LR model, there is an increase in the unsuitable area and a decrease in the moderately suitable, suitable, and optimum for RCP 2.6 in the year 2050, while in RCP 8.5 the increase is in the unsuitable and moderately suitable, while the suitable and optimum show a decrease.

There is greater variability with this model for the year 2070 since, for RCP 2.6, there is an increase of 5.41% in the optimal area and a decrease in the other ranges, and for 8.5, all ranges increase except the suitable area.

The predictions of both models and scenarios indicate that fig crop distribution will be present in northern Mexico, through the Sierra Madre Occidental and Eje Neovolcánico transversal, in a consistent way to the mean annual temperature range of 7–26 °C and precipitation of 580–2233 mm, considering warm temperatures and precipitation of up to 800 mm as ideal conditions for the crop [89,90,91]. Martínez-Meyer et al. [92] indicated that nutrient concentration, soil chemical and physical variables, species restriction, and some other type of disturbance that is not detected in the field can help to support the location of the suitability of the crop with respect to the analyzed variables.

5. Conclusions

The results of this study are of great importance because they represent an advance in the study of fig cultivation in our country and, at the same time, serve as a guide to take advantage of the areas where this crop has potential for its current and especially future production, since, due to the diversity of Mexico in flora, fauna, soils, and climate, it is susceptible to changes, which can harm or favor the production of this crop. The central region of Mexico, the Yucatan Peninsula, and Baja California North and South do not present positive changes in the future, being unsuitable for fig production, unlike the northern region, the Neo-volcanic axis, and the Sierra Madre Occidental, which present positive variations for the future of the crop in Mexico, which can be seen represented in the future projection maps and in the data contained in the tables of the Appendix A.

In addition, the results of the present work can be extended by taking into account other variables such as bulk density, cation exchange capacity, evapotranspiration, radiation, and nutrition, among others, which could positively or negatively modify the projections of habitat suitability for fig cultivation in Mexico and other regions of the world with similar or different climates in the future.