2.1 Study area

Nayarit (23º 05’-20º 36’ N; 103º 43’-105º 46’ W) is a state in central Mexico with an extension of 27 621 km², limiting to the north with Sinaloa and Durango, to the east with Durango, Zacatecas and Jalisco, to the south with Jalisco and the Pacific Ocean, and to the west with the Pacific Ocean and Sinaloa. Twenty-five weather stations (WS) distributed across Nayarit and bordering states (whose geographical locations are listed in Table I) were selected for this study.

Source: ^{SMN, 2019}.

The types of climate in Nayarit and the percentages of the state area they encompass are: hot humid with heavy rainfall in summer (Am), 0.56%; hot sub-humid with summer rainfall (A[w]), 60.63%; warm sub-humid with summer rainfall (ACw), 30.96%; temperate sub-humid with summer rainfall (C[w]), 6.15%, and semi-arid very hot and hot (BS1[h’]), 1.7% (^{INEGI, 2018}) (Fig. 1). The average annual temperature is 25 ºC, the minimum average temperature is 10 ºC in January, whereas the maximum average temperature is about 35 ºC in May and June. The raining season is in summer (May through September), averaging an annual precipitation of 1100 mm (^{García, 2004}) (see Table I for greater detail).

Fig. 1 Climates of the state of Nayarit, Mexico. 

Nayarit has a wide range of soil types (Table II): Acrisol, Andosol, Arenosol, Cambisol, Fluvisol, Gleysol, Leptosol, Luvisol, Nitisol, Phaeozem, Regosol, Solonchak, Umbrisol and Vertisol (^{INEGI, 2018}). The type of land is distributed within the state as follows: agriculture (15.9 %), grassland (14.6 %), livestock (61.2 %), woodland (4.9 %), and non-productive (3.4 %) (^{Bojórquez et al., 2006}). Nayarit is divided into four hydrological regions: Presidio San Pedro, Lerma-Santiago, Huicicila and Ameca. The main rivers across the state are Acaponeta, San Pedro, Santiago, Huicicila and Ameca. Freshwater bodies and salty ponds are the Aguamilpa, San Rafael and Amado Nervo dams, and the Agua Brava lake.

2.2 Methodology

The agroclimatic zoning was based on PCCS (^{Papadakis, 1970}), which relies on climate variables such as average, maximum and minimum temperatures, annual and monthly rainfall, and potential evapotranspiration. The information from the 25 WS was gathered from data records (1951 to 2010) provided by Mexico’s National Weather Service (^{SMN, 2019}).

PCCS classifies geographic regions into 10 climatic groups, according to three dimensions: winter and summer types, and humidity patterns. These groups are cold land, desert, Mediterranean, Pampean and tropical (nine subdivisions each); polar alpine and subtropical (five subdivisions each); marine and steppe (eight subdivisions each); and continental humid (three subdivisions). PCCS includes a list of 186 special diagnostics used to define each group’s subdivisions.

PCCS underlines the concept of monthly climate based on thermic and hydric characteristics, identifying 29 monthly thermic climates according to the average temperatures and extremes (maximum and minimum); and seven monthly hydric climates, composed by monthly precipitation plus the volume of water stored in soil due to previous rainfalls.

Arid climate (a) is defined when a set of climate variables entails less than 25% of the potential evapotranspiration; dry (s), between 25 to 50%; dry hydric intermediate (i), 50 to 75%; humid hydric intermediate (y), 75 to 100%; post-humid (p), more than 100%; humid (h), if rainfall denotes more than 100%; hyper-humid (w), when precipitation plus the soil moisture stored from previous rainfalls account for more than 200% of the potential evapotranspiration (^{Papadakis, 1970}). PCCS computes evapotranspiration based on mid-day saturation deficit. The three PCCS dimensions were calculated using the climatic information from each WS.

3.1 Determination of winter climate types

Temperature data from 1951 through 2010 of almost all CS records, identified January as the coldest month of the year, except in La Yesca (December), San Marcos and Zacualpan (both in February). The lowest T _pb , T _min and T _max on that month were measured in Las Bayas (-6.6, -1.7 and 17.7 ºC) and Tenzompa (-3.9, 0.8 and 22.1 ºC) (Fig. 2); the highest, in El Carrizal (14.5, 17.7 and 32.4 ºC).

Fig. 2 Temperature records for coldest months (December-February) in participating climate stations (1951 to 2010). 

CS were assigned to the following type of winter climates: 17 stations in tropical zone (Tp [hot], tP [warm]); five in the citric zone (Ct [tropical], Ci [typical]); and three in the oat zone (av) (Table V).

3.2 Determination of summer climate types

All of the CS recorded T _pb > 2.0 ºC all year round (i.e., frost free), except for Amatlán de Cañas (11 months of FFP _min ); Bolaños, Hostotipaquillo and San Juan Peyotán (nine months); Tepic (five months); San Gregorio and Tenzompa, (four months); and Las Bayas (none). Also, all WS recorded a 12-month FFP _avail except Las Bayas, San Gregorio and Tenzompa, which registered FFP _avail during one, six and seven months, respectively.

La Yesca (40.6 ºC) was the WS with the highest T _max value, whereas Las Bayas (23.4 ºC) had the lowest; T × 6 of La Yesca was 37.6 ºC and for Las Bayas 22 ºC; T × 4 for the same WS was 37.6 and 22.55 ºC, respectively (Fig. 2). According to the data obtained, the number of CS assigned to types of summer climates were as follows: 23 to the cotton zone (17 to the hottest [G]; five to the mild hot or warm [g]); one to an alpine zone (a), and two to the wheat zone (t) (Table V).

3.3 Humidity pattern determination

According to the humidity patterns, CS with extreme rainfall values were Paso de Arocha (1685.4 mm) and Bolaños (593.3 mm); extreme PET values were observed in San Juan Peyotán (2797.68 mm) and Las Bayas (1188.32 mm); in the rest of the CS, intermediate amounts were recorded (Fig. 3). However, all CS had higher PET values than P, which indicates an important water flow across the soil-plant-atmosphere system. As for the water balance, the highest PET value was recorded at La Yesca (-2028.16 mm). The WS with lowest P-PET values were San Gregorio (-89.2 mm) and Paso de Arocha (-106.59 mm).

Fig. 3 Annual hydric pattern of the climate stations in the study (1951 to 2010). 

The number of CS that exhibited a particular hydric pattern were: one humid (Hu) (HI _A > 1 and L _n > 20%) and 24 monsoon (14 dry monsoon [Mo; 0.44 < HI_A < 0.88 and L_n < 20%]; 9 semi-arid monsoon [mo; HI_A < 0.44]; and one rainy monsoon [MO; HI_A>0.88]) (Table IV). CS were associated with three climate groups, according to the winter and summer types, hydric patterns and special diagnostics (Table V). Nevertheless, subtropical and tropical climatic groups are found only in Nayarit. The cold land group was registered in WS located in surrounding states (Table V).

3.4 Agroclimatic zoning and recommended crops

Figure 4 shows the agroclimatic zoning of the state of Nayarit, followed by the analysis based on the three climatic groups and the respective subgroups identified.

Fig. 4 Agroclimatic zoning of the state of Nayarit, Mexico. 

3.4.1.1 Cool-winter tropical subgroup

Non-irrigated winter potato (Solanum tuberosum L.) and rice (Oryza sativa L.) render high yields. Olives and grapes also grow well; however, since the summer is long and humid, they require fungicides and growth regulators for pest control and the use of irrigation during the dry season as well. Marginal progress is observed on winter cereals such as oat (Avena sativa L.), barley (Hordeum vulgare L.), rye (Secale cereale L.) and wheat (Triticum aestivum L.); and summer cereals such as rice, corn (Zea mays L.), temperate millet (Panicum miliaceum L.), pearl millet (Pennisetum cinereum Stapf & C.E. Hubbard) and sorghum (Sorghum bicolor [L.] Moench). All of them need irrigation, fertilizers and growth regulators, and require close attention to sowing dates: grains are hard to store when harvested during the wet season, whereas in the dry season, they could be damaged.

Therefore, it is recommended to sow at the beginning of the wet period, even though the harvest will take place under rainy weather. For instance, corn is sowed at the start the wet season, whereas sorghum and millet are sowed afterwards in order to be harvested and stored in the dry period. Non-irrigated rice is planted if the dry season is longer than five months, but the storage may turn difficult when it is harvested during rainfall. In all cases, there is a need of fertilizers, growth regulators and grain driers. Table VI shows recommended crops for Nayarit according to PCCS groups and subgroups identified.

Table VI Crops with improved yields, according to climate groups in ^{Papadakis’ (1970)} system. 

Climate group

Subgroup

Winter cereal

Summer cereal

Other crops

Oats

Rye

Barley

Wheat

Rice

Corn

Millet

Sorghum

Olives

Cotton

Banana

Sugarcane

Citric

Forage

Potato

Beetroot

Tea

Grapes

Tropical

Cool-winter tropical

-

-

-

-

-

-

-

-

-

Continental savanna tropical

-

-

-

-

-

-

-

-

-

-

Marine savanna tropical

-

-

-

-

-

-

-

-

-

-

Semiarid tropical

-

-

-

-

-

-

-

-

-

Subtropical

Marine semitropical

-

-

-

-

-

-

-

-

-

Continental semitropical

-

-

Subtropical continental

-

Cold land

High Andine

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

High cold land

-

-

-

-

-

-

-

-

3.5 Agroclimatic zoning of PCCS and its relation to the climatic conditions of Nayarit

Based on ^{García’s classification system (García, 2004}), the main climate types distributed across the state’s surface are hot subhumid with summer rainfall (60.63%), and warm subhumid with summer rainfall (30.96%) (^{INEGI, 2018}). Comparing this information with our study results, we found similar characteristics matching the tropical and subtropical climatic groups of the PCCS. respectively, where most of the WS are located (Fig. 1).

The largest surface area when zoning with the PCCS corresponded to semiarid tropical climate (1 010 722.42 ha), whereas the marine tropical savanna subgroup had the least (72 603.56 ha) (Table V). Irrigated farmland area (219 817.95 ha) was smaller than dryland agriculture area (299 361.56 ha) (Fig. 5).

Fig. 5 Irrigated and non-irrigated agricultural zones in the state of Nayarit, Mexico. 

^{Hernández et al. (2018)} compared the Köppen Climate Classification (modified by ^{García, 1964}) with the Worldwide Bioclimatic Classification System (^{Rivas-Martínez et al., 2011}) in Mexico, identifying coincidences regarding humid conditions. For Nayarit, ^{Hernández et al. (2018)} determined tropical macrobioclimate, tropical-season rain bioclimate, and tropical xeric, as the three highest ranking units according to summer climates and intermediate humidity patterns (in both systems). These results have also similar features than the tropical and subtropical climate groups of the present study.

3.6 PCCS applications versus other agroclimatic zoning methods

PCCS is a useful tool for adequate crop planning, based on the main climatic factors of a particular area. ^{Halabian et al. (2014)} used this methodology in Iran (East-Azerbaijan province), where they identified the climate types across the province and determined the best crops for each. The study also recommended sowing months for sugar beet, which is a very important crop for the country’s economy. On the other hand, ^{Alonso et al. (2012)} analyzed the impact of climate change on the agriculture potential of 17 municipalities in Cantabria (Spain), using agroclimatic zoning by means of the PCCS. Their study disclosed the ecological requirements for 14 crops, as well as the agricultural boost of cranberries (Vaccinium uliginosum L.) and grapevines (Vitis vinifera L.) in the territory.

In Venezuela, ^{Colotti (1997)} evaluated different classifications of common use in agroclimatology, among others the ^{Thornthwaite (1948)} method and PCCS, in order to identify convenient applications related to PET. ^{Velasco and Pimentel (2010)} used the same methodology for the zoning of the state of Sinaloa (northwest coast of Mexico), highlighting that PET estimation varies according to the method being used: underestimation by the Makkink method (53% of the registered evaporation) or very high values with the PCCS (96%). These authors recommend adapting the evaporimeter tank method and the use of monthly coefficients of evaporation (higher for dry months).

^{González et al. (2009)} found significant differences between measured PET and calculated PET when tracing the bioclimatic map of the Mendoza grasslands (Argentina): -27.6% using the Thornthwaite method, 7.2% with PCCS, 6.7% using the Penman-Monteith method, and -6.4% with the Penman standard method. The investigations of ^{Ortiz and Chile (2020)} in the Tumbaco Valley (Ecuador) also showed fluctuations in evapotranspiration values of the reference crop (ET ₀ ): based on FAO-56, the Makkink and evaporimeter tank methods generated lower values (11.59 and 7.29%, respectively), whilst the Hargreaves, modified Thornthwaite, FAO Radiation, Priestley and Taylor, Turcun, and Jensen and Haise methods produced higher values (41.59, 26, 11.87, 3.92. 1.46, and 0.18%, respectively).

Temporal and spatial variability of climate forced the use of different methodologies for PET estimation. Although based on meteorological data, some of these were empirically obtained through field experiments while others were theoretical approaches (^{Landon, 2004}), such as the evaporimeter tank and the Penman-Monteith model, respectively. Some others are based on temperatures, such as the Thornthwaite and Turc method, or in temperature and solar radiation, such as the Hargreaves, Jensen & Haise, Makkink, Priestley & Taylor, FAO Radiation models (^{Bhabagrahi et al., 2012}). Thus, model complexity, climatic variables diversity, and the varying conditions from which different models were designed, influence the fluctuating PET estimation.

It is important to bear in mind that PET implies a set of a simultaneous process of water loss: soil evaporation and plant transpiration. Apart from water availability in the surface horizon, soil evaporation depends on the solar energy it receives, which decreases along the crop cycle: at the beginning, water is lost through the former mechanism; later, as the crop progresses and the water layer covers up, the latter mechanism takes place. These are features taken into consideration by the FAO method for PET calculation, in addition to soil moisture, soil fertility and use of fertilizers, soil management practices, presence of hard breaking horizons and salinity, regarding soil-related variables; and plant cover and density, disease control and parasite presence regarding plant-related variables (^{Allen et al., 1998}, ²⁰⁰⁶).

For this reason, FAO (^{Allen et al., 2006}) suggests the FAO Penman-Monteith (FAO-56 PM) as the most precise method for PET calculation based on climatic parameters, suitable for arid and humid climate estimations, which incorporates physiological and aerodynamic parameters as well: (i) aerodynamic resistance (atmospheric evaporative demand according to climate variables such as temperature, relative humidity, sunlight hours, wind, altitude and latitude); (ii) crop surface resistance (water flood diffusion from roots to plant stomata and direct water evaporation from soil), and (iii) albedo (solar radiation reflected by the crop), using a reference crop (grass height of 0.12 m, irrigated with a total cover of soil surface) with surface resistance of 70 s m^-1 and 0.23 of albedo (^{Allen et al., 2006}).

From this standpoint, PCCS is an empirical method that uses equations with a minimum of meteorological variables (T _max and T _min ) for PET calculation, regardless of crop characteristics and soil factors; hence, calculated values may differ from those estimated by other methods. When computing water balance and analyzing historical monthly PET, this study found that all WS recorded the highest PET in May, except for La Yesca and San Gregorio (both in April), San Marcos (August) and Zacualpan (November).

There were five periods of six months each, where PET was highest: from January to June (Acaponeta, Amatlán de Cañas, Despeñadero, El Carrizal, Jumatán, Las Gaviotas, La Yesca, Ototitán, Pajaritos, Paso de Arocha, Rosamorada, San Gregorio and Tepic), February to July (Bolaños, Capomal, El Naranjo, Hostotipaquillo, Huaynamota, Las Bayas, San Juan Peyotán and Tecuala), March to August (San José Valle and Tenzompa), June to November (San Marcos), and July to December (Zacualpan). This allowed to identify that in the first two periods (January-June and February-July) PET records were the highest in 52 and 36% of WS, respectively.

Seven periods of different duration were also identified where PET was greater than P for all months: 56% of WS from October to June, and all-year round in Bolaños (Table VII). PET estimations with the PCCS method were conclusive to determine the humidity pattern of the WS in this study, and thus for the agroclimatic zoning of Nayarit. Results indicate that crop selection across the state must be done according to the water deficit of those months, in order to ensure optimal yields.

^{Jiménez et al. (2004)} state that, undoubtedly, climate is one of the main factors a farmer should always take into account, providing its influence over plants’ growth and maturation. Nevertheless, these authors outlined the importance of analyzing the soil and the crop as an integrated system, otherwise, there is a risk of an incomplete and unreliable zoning. For instance, in their study to determine potentially agricultural zones for sugar cane production in the south of Tamaulipas (northeast Mexico), they realized that if only climate was analyzed (agroclimate zoning), 86.9% of the territory was set as very suitable for that crop and 13.1% only suitable. However, when they included soil characteristics (such as units, phases, textures and slopes), only 40% of the territory was set as suitable for sugar cane (^{Jiménez et al., 2004}).

^{Díaz et al. (2000)} concluded that crop adaptability analysis is insufficient if no other information is taken into account, especially soil characteristics. In an agroecological zoning study of coffee (Coffea arabica L.) and camedor palm (Chamaedorea elegans Mart.) agroforestry systems conducted in Veracruz, Mexico, ^{Pérez-Portilla and Geissert-Kientz (2006)} found that temperature and water availability explain 70% of crop distribution within the study area, and soil characteristics only account for 45%. In other words, the distribution of agroforestry systems is defined in low temperatures by a less cold-weather tolerance of coffee plants, but not for fertility and stone content of the soil. Hence, the corresponding analysis between crop suitability and the presence of the agroforestry system only matched regarding climate. ^{Ohmann and Spies (1998)} support the findings when they assert the secondary role of soil in the distribution of species at low-level scale.

In similar studies, ^{Medina and Aldana (2019)} compared the Caldas-Lang & Holdridge climate classification method to the climate soil zoning in Valle del Cauca (Colombia). The outcomes of the study provided valuable knowledge about climate conditions associated with soil formation and enhanced the assessments over soil use and management in the basin. For this reason, ^{FAO (1997)} insists that by integrating soil components into zoning methodologies, the agroecological approach (as opposed to only climatic) has the advantage of evaluating suitability and land potential productivity as well. This underpins more advanced studies focusing on potential crop projection areas, harvest and production, which in turn allows for the evaluation of land degradation and sustainability capacity, as well as generating optimization models for land use in cattle farming.

All aforementioned highlights PCCS limitations: (i) PET estimations may differ from other methods, as PET is computed with a minimum of meteorological variables, however, this could become an advantage if data availability from WS is rather scarce (climatic, soil-related and of the crops themselves); (ii) overlook of crop characteristics and soil factors; (iii) few crops inclusion and, in the case of Mexico, very few fruit trees, and (iv) loose precision in crops of tea types.

In Mexico, agroclimatic zoning studies are scarce, and even more those comparing methodologies for the same zone aiming to suggest the most suitable crops to render optimal yields. Therefore, as a result of recurrent limited information available and its less complex construction, PCCS arise as a useful tool for crop planning according to the climatic variables of a region.