Introduction

Although chickpea is known for its greater drought tolerance compared to other grain legumes, drought reduces its yield and can cause complete failure of the culture (^{Turner et al., 2001}). In Mediterranean and Sub-tropical climates the filling of the chickpea grain, when planted mainly under conditions of residual moisture, is subjected to terminal drought which limits its yield. Thus, the terminal drought is considered as one of the adverse factors of globally greatest impact for chickpea (^{Fang et al., 2009}).

One of the main attributes of the chickpea is the ability of its root system to explore the soil for moisture and achieve yield with less water than other crops. This feature is important because itrepresents an opportunity for saving irrigation water and to include chickpea on favorable agronomic rotations (^{Guriqbal et al., 2016}). Upadhyay et al. (2013) described selection methods for different characteristics related to drought tolerance such as early maturity (escape), large and deep roots, high water efficiency, small leaves, reduced canopy temperature, carbon isotope discrimination, high chlorophyll content in the leaf (drought avoidance) and selection strategies to improve drought resistance. The plants also control the water losses by the control of stomatal opening under conditions of high vapor pressure deficit (VPD), in a transient way, both processes (leaf development and stomatal opening) are mostly controlled by hydraulic processes (^{Vadez, 2014}).

In the Costa de Hermosillo, Sonora, chickpea sowings are carried out in two production systems; planted with drip irrigation strips separated at 1.6 m and planted in rows at 50 cm apart with only pre-sowing irrigation. In the latter production system at the end of the culture cycle terminal drought is shown coinciding the reproductive stages with high temperatures and lack of precipitation so it is common to obtain low grain yields, decreased size and export percentage (^{Durón et al., 2004}). In El Bajío with residual moisture plantings (from September to December), rainfall decreases and drought periods occur matching the reproductive stage, when the culture is more sensitive to the lack of moisture (^{Acosta et al., 1999}).

The effect of terminal drought depends on its duration, on the soil’s ability to store water for the roots, on the atmospheric conditions that influence the evapotranspiration rate, and on the genetic constitution of the plant that conditions the reaction to this abiotic factor (^{Nielsen and Nelson, 1998}). ^{Pushpavalli et al. (2014)} mention that there is a high genotype*environment interaction due to genetic variation for yield and its components in the treatment of controlled drought used where daily water loss becomes equal for all plants from the soil, revealed genotypic differences in the sensitivity of the reproduction process to drought.

Genetic improvement of chickpea for drought tolerance represents one of the best alternatives to increase chickpea yield under these production conditions and considering that one of the main attributes of the chickpea is the ability of its root system to explore the soil searching for moisture and achieve yield with less water than other crops. This feature is important because it represents an opportunity for saving irrigation water and for including chickpea on favorable agronomic rotations (^{Frahm et al., 2003}; ^{Guriqbal et al., 2016}).

However, the development of improved varieties with drought tolerance is difficult, slow and costly, since genotypes show inconsistency in its yield, due to differences in severity, time of occurrence and duration of drought across localities and years (^{Acosta et al., 1999}; ^{Rosales-Serna et al., 2000}), also there is a strong genetic environmental interaction that prevents the rapid advance of genetic improvement. Very encouraging results demonstrating the efficacy of molecular marker assisted selection for stress tolerance to terminal drought in chickpea have been obtained (^{Samineni et al., 2015}). ^{Sarmah et al. (2012)} mention that wild chickpeas have high levels of resistance to pod borer and to conditions of water deficit where the marker assisted selection and genetic engineering of chickpea are exploited to increase resistance-tolerance levels to these limitations in the future.

A way to achieve results in less time is to evaluate genotypes without moisture limitation (irrigation) and with irrigation suspension in the reproductive stage to identify outstanding genotypes by selection indices (^{Rosales-Serna et al., 2000}). Among the most used indices there is the one proposed by ^{Fisher and Maurer (1978)}, in which the yield average of all the genotypes in both humidity conditions is used to calculate the intensity index and drought susceptibility; there is also the geometric mean where the observed yield of each genotype under irrigation and drought conditions is used, and has proven to be an effective selection criteria (^{Abebe et al., 1998}).

Relative efficiency index of ^{Graham (1984)} allows the classification and selection of high yield genotypes under irrigation conditions and moisture deficiencies. Among various environmental constraints, high temperature is one of the most important for the growth and yield of chickpea in a range of environments (^{Summerfield et al., 1990}; ^{Singh et al., 1994}; ^{Basu et al., 2009}). The objective of this paper was to characterize the productive response of 12 ‘Kabuli’ chickpea genotypes with and without the application of supplementary irrigation in the reproductive stage.

Materials and methods

Two trials were established in the costa de Hermosillo, Sonora (28º 45’ 5.87” north latitude and 111º 27’ 37.69” west longitude), at 56 masl. The predominant climate according to the classification of Köppen modified by ^{García (1973)}, is BWh, very dry semiwarm; and BW (h’) very dry very warm and warm with average low temperatures of 14 to 16 °C in the months of January and February and extreme temperatures of 31 to 47 °C in the months of July and August. Rainfall in the coastal region occurs in June, July, August and September with an annual rainfall of 75 to 200 mL, the predominant soil type is Yermosol, mainly distributed in the central zone of the coast and the northeast.

Twelve genotypes of white chickpea (Kabuli) were evaluated, eight varieties originated in the chickpea breeding program of INIFAP; Blanoro, Blanco Magdalena 95, Tequi Blanco 98, Costa 2004, Blanco Sinaloa 92 and Desierto 98, in addition to the introduced varieties Troy and Sierra and 4 elite lines Hoga 067, Hoga 2001-2-2, Hoga 021 and Cuga 08-743 originated in the same Genetic Improvement program.

The experimental design used was randomized complete blocks with three replications, in plots of 1 groove 5 m in length and 1.6 m in width with double row separated to 50 cm, where the useful plot corresponded to the same experimental plot. Planting took place on December 27, 15 seeds per linear meter were sown for a population of 175 000 plants ha^-1. In one trial the crop had no water restriction during its development and in another trial the application of water at the beginning of the reproductive stage (beginning of flowering) was restricted. The response variables were: 1) grain yield, which was calculated in kg ha^-1 from the weight of the grain harvested from each plot; 2) weight of 100 seeds, measured in 100 seeds of each plot in grams; 3) number of grains per plant, measured in 5 plants per plot; and 4) plant height measured in 10 plants per plot. For soil moisture monitoring, moisture sensors were placed in the soil profiles in both tests at 0-30 and 30-60 cm of deep.

Variance analysis of the variables in both moisture conditions was performed, and in the case of grain yield, the data were analyzed in random block design in split plots arrangement, considering each moisture condition as a large plot and the genotypes as small plots. Data were analyzed using the SAS statistical package, version 7.2 (^{SAS Institute, 1999}) and when significance was detected between treatments, the least significant difference test (DMS, 0.05) was applied.

The effect of drought on the average grain yield of each genotype was estimated using the drought susceptibility index (ISS), using the geometric mean (MG) and the relative efficiency index (IER).

The ISS of each genotype was calculated using the equation proposed by Fischer and Maurer (1978): ISSi= 1-(Yii/Yci)/IIS, where: Yii= average yield of each genotype without irrigation from flowering beginning; Yci = average yield of each genotype with irrigation application during the cycle. The drought intensity index (IIS) was obtained using the formula: IIS= 1-(Yi/Yc), where Yi= average yield without irrigation from the flowering beginning and Yc= average yield with irrigation application during the cycle.

The MG proposed by ^{Samper and Adams (1985)}, was calculated with the equation: MG= (Yii*Yci)1/2, where MGi= geometric mean of each genotype; and Yii and Yci= yield of each genotype under additional irrigation conditions and without irrigation from the flowering beginning, respectively.

The IER described by ^{Graham (1984)}, was calculated by the equation: IER= (Yii/Yi) (Yci/Yc), where: IERi= relative efficiency index of each genotype; Yii= yield of i genotype without irrigation application from flowering onset; Yi= average yield with irrigation suspension from flowering onset; Yci= yield of i genotype with irrigation during the crop cycle; and Yc= average yield with irrigation during the crop cycle.

Results and discussion

Soil moisture conditions and climatic variables

The moisture distribution of water in the two tests was monitored with humidity sensors, in Figures 1 and 2 its behavior is shown in the profiles 0-30 and 30-60 cm in depth and in the two humidity conditions, with (CRS) and without supplemental irrigation (SRS), the latter implemented at the beginning of flowering, and from this stage the readings increase from 67 to reach in only 12 days 200 centibars indicating total absence of moisture in both profiles on the contrary, in the test with supplementary irrigation in the 0-30 profile it showed greater variation in the moisture distribution with readings from 15 to 88 millibars and in 30-60 profile this was more stable with readings ranging from 27 to 67 centibars. The environmental conditions during the development of the study are shown in Table 1 taken from the weather station located in the Costa de Hermosillo Experimental Field, which can be found on http://agroson.org.mx

Figure 1 Distribution of soil moisture in the 0-30 cm profile in the two moisture conditions. 

Figure 2 Distribution of soil moisture in the 30-60 cm profile in the two moisture conditions. 

Table 1 Climate information during the chickpea cycle in the 2014 and 2015 autumn-winter period, Costa de Hermosillo, Sonora. 

Effect of moisture regime with yield

Under irrigation conditions, grain yield differences between genotypes were detected, humidity conditions and in the interaction of genotypes with humidity conditions, (p≤ 0.01), weight of 100 seeds (p≤ 0.01), number of grains per plant (p≤ 0.05), and plant height (p≤ 0.01). In the treatment with drought differences between genotypes were detected in all evaluated variables except in number of grains per plant.

Under irrigation conditions, Blanco Magdalena 95 and Hoga 067 showed superior yield and significantly different from the rest with 1.6 and 1.59 t ha^-1. Under drought conditions, Hoga 067 and Tequi Blanco 98 genotypes showed statistically higher yields compared to the rest with 0.377 and 0.324 t ha^-1 (p≤ 0.01). In addition to the aforementioned genotypes under irrigation conditions, the Sierra, Troy and Desierto varieties showed smaller reductions than the general average, the genotypes less affected by the drought condition were Sierra (72.1%), Troy (76.1%) and Hoga 067 76.3%), not necessarily showing the greater yield. The yield ranged between humidity and genotypes (p≤ 0.01), detecting statistically significant differences between factors (p≤ 0.01). With irrigation during the development cycle of the crop a superior average yield to that obtained with the moisture stress condition (p≤ 0.05) was recorded. A group of 6 genotypes was outstanding, the highest yields were shown by the lines Hoga 067 and Hoga 2001-2-2 in addition to the Blanco Magdalena 95, Blanoro, Tequi Blanco 98 y Costa 2004 varieties (Table 2).

Table 2 Effect of two moisture conditions on grain yield and weight of 100 seeds (g) of white chickpea (Kabuli) genotypes. Costa de Hermosillo, Sonora 2014. 

^* = genotipos estadísticamente superiores, según la diferencia mínima significativa (0.05)

Under irrigation conditions, the Hoga 067, Blanco Sinaloa 92, Cuga 08-743, Blanoro, Hoga 2001-2-2 and Costa 2004 genotypes showed the highest weight of 100 seeds, which is closely related to the caliber of these genotypes, with In the present study, showing weights of 76.3, 75.7, 73.3, 70.9, 69.8 and 59.8 g by100 seeds, under drought conditions Hoga 067, and Cuga 08-743, showed the highest weights with 66 and 64.3 g by 100 seeds, the less affected genotypes were Troy, Desierto 98, Hoga 021 and Cuga 08-743 with reductions in weight of 100 seeds below the general average with 7.4, 7.9, 11.7 and 12.3%, respectively (Table 2).

Regarding the number of seeds per plant under irrigation, the Hoga 2001-2-2, Blanco Sinaloa 92, Hoga 021 and Blanco Magadalena 95 genotypes showed more grains per plant and were statistically different to the rest (p≤ 0.05), with 33, 32, 28 and 25, under conditions of moisture stress at the beginning of the reproductive stage, the quantity of grains per plant was strongly affected, the least affected were the genotypes Troy, Costa 2004, Hoga 067 and Blanoro showing reductions of 50, 56.6, 60.3 and 63.4% in that same order being below the general average for this variable (Table 3).

Table 3 Effect of two moisture conditions on number of grains per plant and plant height (cm) of white chickpea (Kabuli) genotypes. Costa de Hermosillo, Sonora 2014. 

^* = genotipos estadísticamente superiores, según la diferencia mínima significativa (0.05); ns= no significativo.

Drought susceptibility indices

Reduced yield due to lack of moisture was evident in all genotypes, but it was more severe in the Cuga 08-743, Blanco Sinaloa 92 and Blanco Magdalena 95 genotypes, which in turn showed the highest ISS values. The Sierra, Hoga 067, Troy and Desierto genotypes showed greater tolerance to drought stress, since they showed the susceptibility indexes closer to zero (^{Fisher and Maurer, 1978}). This index is an acceptable criterion to select genotypes that less reduce its yield under moisture stress, although they did not necessarily show the greatest yield (^{Rosales-Serna et al., 2000}).

In the absence of rain interference during the reproductive stage of the crop, it was possible to quantify the response of the genotypes to this quantitative trait. The lines Hoga 067 and Hoga 2001-2-2 and the varieties Blanoro and Tequi Blanco 98 showed the highest values of the MG and IER indices (Table 4), which indicate high yield in the two studied moisture conditions. Both indices are highly correlated and are based on the yield under the two humidity conditions (^{Mayek et al., 2003}; ^{López et al., 2006}). The results observed with the indices used are similar to those obtained by ^{Rosales-Serna et al. (2000)}, who suggested the combined use of a rate related to yield reduction (ISS) and another with yield between moisture conditions (MG or IER).

Table 4 Grain yield of chickpea genotypes in two moisture conditions, and indicators of susceptibility to drought and yield efficiency. Costa de Hermosillo, Sonora 2014. 

ISS= índice de susceptibilidad a sequía; MG= Media geométrica; IER= Índice de eficiencia relativa; * = genotipos estadísticamente superiores, según la diferencia mínima significativa (0.05).

Conclusions

The Sierra, Troy and Desierto varieties and the Hoga 067 line were the most tolerant to severe terminal drought, while the Cuga 08-743 line and the Blanco Magdalena 95 and Blanco Sinaloa 92 varieties were the most susceptible.

The Hoga 067 and Hoga 2001-2-2 lines and the varieties Blanoro and Tequi Blanco 98 showed the highest values of MG and IER indices, which indicate high average yield of the two edaphic moisture regimes.