Scholarly article on topic 'SALTMED Model Performance on Simulation of Soil Moisture and Crop Yield for Quinoa Irrigated Using Different Irrigation Systems, Irrigation Strategies and Water Qualities in Turkey'

SALTMED Model Performance on Simulation of Soil Moisture and Crop Yield for Quinoa Irrigated Using Different Irrigation Systems, Irrigation Strategies and Water Qualities in Turkey Academic research paper on "Agriculture, forestry, and fisheries"

CC BY-NC-ND
0
0
Share paper
Keywords
{"SALTMED Model" / quinoa / "saline water" / "water quality"}

Abstract of research paper on Agriculture, forestry, and fisheries, author of scientific article — Çiğdem Ince Kaya, Attila Yazar, S. Metin Sezen

Abstract The effective usage of limited water resources and ensuring of agricultural sustainability especially in case of using poor quality water for irrigation in arid and semiarid regions require a good irrigation management and management tools. The SALTMED model has been developed as an integrated model that can simulate soil moisture, salinity distribution, leaching requirements and crop yield by considering irrigation systems, soil types, crops, water application strategies and different water qualities. The objective of this research was to evaluate the SALTMED model performance on simulation of soil moisture, total dry matter and yield of Chenopodium quinoa Willd. var. Titicaca grown in the Mediterranean region of Turkey under different irrigation methods, irrigation strategies and water qualities (fresh, saline and drainage water). For this purpose, field data from two different locations, Tarsus and Adana, were used in SALTMED Model. Different irrigation treatments with drip irrigation system consist of full and deficit irrigation using fresh water, full and deficit irrigation using saline water at ECw 30, 20 and 10 dS m-1 salinity levels, and partial root drying (PRD) with fresh water along with rainfed (DRY) treatment were considered in Adana study. In Tarsus, line-source sprinkler system was used to create four irrigation levels (full and three deficit irrigation treatments) using drainage canal water. The field experiments results were compared with simulation results on soil moisture, grain and dry matter yield. The model was able to simulate with good precision, soil moisture values, total dry matter and grain yield for quinoa under various irrigation strategies, irrigation methods, and different water quality conditions for quinoa grown in the Mediterranean environment.

Academic research paper on topic "SALTMED Model Performance on Simulation of Soil Moisture and Crop Yield for Quinoa Irrigated Using Different Irrigation Systems, Irrigation Strategies and Water Qualities in Turkey"

Available online at www.sciencedirect.com

ScienceDirect

Agriculture and Agricultural Science Procedia 4 (2015) 108 - 118

IRLA2014. The Effects of Irrigation and Drainage on Rural and Urban Landscapes, Patras, Greece

SALTMED Model Performance on Simulation of Soil Moisture and Crop Yield for Quinoa Irrigated Using Different Irrigation Systems, Irrigation Strategies and Water Qualities in Turkey

Cigdem Ince Kayaa*, Attila Yazara, S. Metin Sezenb

aIrrigation and Agricultural Structures Department, Qukurova University, Adana, 01330, Turkey bAlata Horticultural Research Institute, Soil and Water Resources Research Station, Tarsus, 33400, Turkey

Abstract

The effective usage of limited water resources and ensuring of agricultural sustainability especially in case of using poor quality water for irrigation in arid and semiarid regions require a good irrigation management and management tools. The SALTMED model has been developed as an integrated model that can simulate soil moisture, salinity distribution, leaching requirements and crop yield by considering irrigation systems, soil types, crops, water application strategies and different water qualities. The objective of this research was to evaluate the SALTMED model performance on simulation of soil moisture, total dry matter and yield of Chenopodium quinoa Willd. var. Titicaca grown in the Mediterranean region of Turkey under different irrigation methods, irrigation strategies and water qualities (fresh, saline and drainage water). For this purpose, field data from two different locations, Tarsus and Adana, were used in SALTMED Model. Different irrigation treatments with drip irrigation system consist of full and deficit irrigation using fresh water, full and deficit irrigation using saline water at ECw 30, 20 and 10 dS m-1 salinity levels, and partial root drying (PRD) with fresh water along with rainfed (DRY) treatment were considered in Adana study. In Tarsus, line-source sprinkler system was used to create four irrigation levels (full and three deficit irrigation treatments) using drainage canal water. The field experiments results were compared with simulation results on soil moisture, grain and dry matter yield. The model was able to simulate with good precision, soil moisture values, total dry matter and grain yield for quinoa under various irrigation strategies, irrigation methods, and different water quality conditions for quinoa grown in the Mediterranean environment.

© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Peer-review under responsibilityofDataResearchandConsulting

Keywords: SALTMED Model; quinoa; saline water; water quality

1. Introduction

The demand for food is rising by rapid world population growth and food security is one of the main concerns of this century. On the other hand, agricultural production is increasingly faced with environmental constraints such as

* Corresponding author. Tel.: +90-322-344-1717; fax: +90-322-344-1515. E-mail address: cigdemincekaya@gmail.com.

2210-7843 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of Data Research and Consulting

doi:10.1016/j.aaspro.2015.03.013

drought, salinity and negative impacts of climate change. Therefore, the effective usage of limited water and soil resources and the inclusion of new stress tolerant crops such as quinoa in cropping systems have become important.

Quinoa (Chenopodium quinoa Willd.) is an annual grain plant native to the Andean region of South America. This plant is known by resistance to various abiotic stresses such as salinity, drought and frost (Jensen et al., 2000; Geerts et al., 2008). Quinoa has been cultivated in diverse environmental conditions include low precipitation, high evapotranspiration rate, frost and soil salinity for thousands years (Jacobsen, 2003). It has adapted to different agro-ecological zones range from sea level up to 4000 m altitude with its rich genetic diversity (Garcia et al., 2003; Jacobsen, 2003). Its high gluten free nutritional component is another characteristic of quinoa. It is the unique food crop that contains all the essential amino acids, trace elements and vitamins, and is also gluten-free (FAO, 2013). Quinoa has been indicated as a good candidate to offer food security, especially in the face of the predicted future world scenario of increasing salinization and aridity (Ruiz et al., 2014). In recent years, it has attracted interest all over the world and quinoa cultivation has begun expand in many countries in Europe, Africa and Asia (FAO, 2013).

It is well known that irrigation plays a key role in increasing agricultural productivity. On the other hand the irrigated agriculture is already the largest water user with an about 70% of global fresh water consumption. The climate change effects are further threat for limited fresh water resources especially in arid and semi arid region. It is possible to obtain high yields using less water with appropriate deficit irrigation strategies. Irrigation with poorer quality water such as saline or drainage water instead of fresh water may be another alternative solution especially in arid and semi-arid zones. In these circumstances, more crop production for growing population in a sustainable and environmental friendly way requires good management tools for limited soil and water resources.

SALTMED model is one of the integrated models that can simulate soil moisture profiles, salinity distribution and nitrogen dynamics in the soil, leaching requirements, crop growth and yield by considering irrigation systems, soil types, crops, water application strategies and different water qualities (Ragab, 2002). The well-known physically based equations are utilized by SALTMED model to calculate time-varying water and solute flow in the soil, water uptake by the roots and crop water use (Ragab, 2002; Ragab et al., 2005a).

The SALTMED model has been successfully tested for tomato irrigated with saline water using furrow and drip irrigation systems in Egypt and Syria (Ragab et al., 2005b). The impact of climate change on the irrigation requirements of three main crops, cabbage, carrot and castor bean of Brazil were investigated using SALTMED model after a successful calibration (Montenegro et al., 2010). The predicting performance of the model on soil moisture distribution and crop yield has been evaluated for quinoa in Italy (Pulvento et al., 2013). In Morocco SALTMED model has been calibrated and validated for quinoa, sweet corn and chickpea (Hirich et al., 2012, 2014). The estimated yield and soil salt profile by SALTMED model was compared with observed data from field experiments of three years for sugar cane in Iran (Golabi et al., 2009). In Portugal, plant growth under different irrigations strategies for five different chickpea varieties simulated with SALTMED model (Silva et al., 2012). According to results of these investigations, SALTMED model can predict accurately water and salt distribution in the soil profile, yield and total biomass of several crops grown under several irrigation applications and environmental conditions.

In this study the SALTMED model ability on simulation of soil moisture, total dry matter and grain yield of quinoa var. Titicaca was investigated by comparing predictions of model and observed data from two field experiments conducted in two locations in south part of Turkey.

2. Materials and methods

2.1. Field experiments

The field experiments were carried out in two locations, Tarsus (37°0r N, 35°0r E; 10 m a.s.l) and Adana (36o59'N, 35o18' E; 50 m a.s.l), during the growing season of 2012. Climatic conditions in both of the experimental site show typical Mediterranean climate characteristics which have 656 and 603 mm long term annual average precipitations respectively in Adana and in Tarsus.

In Adana, the experimental site has clay textured soils which have 198 mm available water holding capacity throughout the 120 cm soil profile. Soil pH values in this soil depth vary from 7.61 up to 7.87 and the initial soil salinity in terms of the electrical conductivity of the saturation extract (ECe) range between 0.12 and 0.19 dS m-1.

Soil moisture contents at field capacity and at permanent wilting point are 37-41% and 24-26% by volume respectively. The soil bulk density ranges 1.14 to 1.30 g cm-3.

In Tarsus, the soil of the experimental site is silty clay loam with 158 mm available water holding capacity in 90 cm profile. The volumetric soil water content at field capacity varies from 34 to 39% and wilting point varies from 17 to 23%. Soil pH values range between 8.3 and 8.35. The initial soil ECe ranges between 0.15 and 0.34 dS m-1. Mean bulk density varies from 1.17 to 1.31 g/cm3.

The experiment was conducted using a randomized block design with three replications in Adana. The experimental plot size was 6 m long and 2.5 m wide and each plot included 5 crop rows in 0.5 m apart. In this trail, fourteen irrigation treatments were considered: full irrigation using fresh water (FIF), full irrigation using saline water at ECw 10, 20 and 30 dS m-1 salt concentrations (SI10-100, SI20-100 and SI30-100), two deficit irrigation levels 67% (DIF-67, SI10-67, SI20-67 and SI30-67) and 33% (DIF-33, SI10-33, SI20-33 and SI30-33) of full irrigation both with fresh and saline water, partial root drying (PRD) using fresh water and a non-irrigated (DRY) treatment. Fresh water was provided from the irrigation canal running next to the experimental site. For saline irrigation treatments, saline water supplied from Mediterranean Sea was diluted with fresh water to ECw 10, 20 and 30 dS m-1 and storage in tanks each with 5 m3 capacity. Irrigation water was applied by the surface drip irrigation system at fixed irrigation interval of seven days. Soil water in the 60 cm soil depth in full irrigation treatments (FIF, SI10-100, SI20-100 and SI30-100) was restocked to field capacity. In deficit irrigation treatments, water was applied at 67 and 33% of those in full irrigation treatments while PRD treatment received 50% of water applied to FIF plots.

In Tarsus, the experiment was carried out using a line-source sprinkler irrigation system which allows a gradual variation of irrigation, in direction at right angle to the source (Hanks et. al., 1976). Drainage water obtained from the drainage canal running through the experimental area was used for irrigations. Drainage water is classified as C2S1 by USSL (1954). During the experiment, the electrical conductivity of drainage water varied between 0.57 and 1.69 dS m-1 and average water pH value was 7.1. Four irrigation levels, namely a full (I1) and three deficit (I2, I3 and I4) irrigations along with a rain-fed treatment were envisaged with four replications. I2, I3 and I4 treatments represent deficit irrigation of approximately 80, 50 and 20%, respectively. Double-nozzle sprinkler heads (4.5 mm x 4.8 mm) placed at 6 m intervals on the laterals provide linearly decreasing wetting pattern under the operating pressure of 300 kPA.

In both field trails, quinoa seedlings with 4-6 leaves produced in a greenhouse were transplanted into the field at 20 cm apart in the row and 50 cm row spacing on 11 April, 2012. Quinoa was harvested on July 3, 2012 in Adana and July 10, 2012 in Tarsus.

In both experiments the soil moisture content changes in each treatment were monitored using both gravimetric sampling and neutron scattering methods during the growth season. A series of plant measurements included plant height, biomass, leaf area index (LAI) were made once a week throughout the growing season at each trial plots. The grain yields and total biological yields of experimental treatments were determined at harvest. Harvest index (HI) was computed as the ratio of the grain yield to the biological yield. Water productivity (WP), the ratio of the seed yields to seasonal evapotranspiration, was calculated. The irrigation water productivity (IWP) which is defined as the grain yield per unit applied irrigation water for a particular treatment was determined.

2.2. Modelling

The data required to run SALTMED model were obtained from two field trails and literature. Crop-specific input data such as plant height, LAI, maximum and minimum root depth, length of the growth stage were obtained from field measurements. The crop coefficients (Kc), basal crop coefficients (Kcb) and fraction cover (Fc) for each growth stage were literature based crop characteristics. Soil characteristics included depth of each soil horizon, saturated hydraulic conductivity, saturated soil water content, initial soil moisture and salinity profiles were obtained from field and laboratory measurements. Irrigation data for different treatments and daily climate data of experimental sites also were used as input data.

The model was calibrated separately for each field trial. The calibration has been done by taking as reference full irrigation treatment with fresh water (FIF) for drip irrigation experiment carried out in Adana and full irrigation treatment with drainage water (I1) for line-source sprinkler irrigation experiment carried out in Tarsus. By fine-tuning on literature-based data, Kc, Kcb, Fc and photosynthesis efficiency, a good agreement was obtained between simulated and observed values of grain yield, total dry matter and soil moisture.

SALMED model validation was done for each experiment by comparing simulated and observed total dry matter, grain yield and soil moisture data of treatments other than FIF and I1. The agreement between simulated and observed data was evaluated by statistical and graphical procedures. In the graphical approach, the measured and simulated data were plotted against time in order to visually assessment of model performance. For statistical comparison the relative error (RE), root mean square error (RMSE), normalized root mean square error (NRMSE) and coefficient of determination (R2) were used. Beside these statistical indices, the index of agreement (d) (Willmott 1981) was calculated as an efficiency criterion. The mathematical expressions of these statistics are:

where O is observed data obtained from field trails, S is predicted data by using model and n is the number of samples. A value of 1 for R2 and d indicates a perfect match between observed and simulated data, while the ideal values of RE, RMSE and NRMSE are zero.

3. Results and discussions

3.1. Field experiments

In the experiment conducted in Adana, a total of 140 mm irrigation water was applied to all full irrigation treatments (FIF, SI10-100, SI20-100 and SI30-100). In the deficit irrigation treatments of DIF-67, SI10-67, SI20-67 and SI30-67 a total of 100 mm water applied while DIF-33, SI10-33, SI20-33 and SI30-33 treatments received 60 mm of water. Seasonal evapotranspiration (ET) varied between 253 and 360 mm. In each salinity level, full irrigation treatments resulted with higher ET values than deficit irrigation treatments. The highest ET value was observed in FIF treatment. Crop water use was significantly affected by irrigation water salinity and soil water deficit (Table 1).

The grain yield of quinoa ranged from 1.28 to 1.96 t ha-1 under different irrigation strategies. Irrigation water salinity and lack of the water in the root zone caused yield reductions at different rates in comparison FIF treatment that represents non-stressed conditions. Under the full irrigation conditions, irrigation water salinity had mild effects on grain yield of quinoa. Moderate deficit irrigations (67%) and PRD resulted in greater yields than severe deficit irrigations (33%). The highest yield reduction occurred in the SI30-33 treatments with 34.7% while the yield reduction was 23% in the non-irrigated DRY treatment. This results show that the combination of drought and salinity stress created more effects on quinoa yields than only drought or salinity stress. The highest water productivity (WP) was calculated in SI30-67 treatment. The irrigation water productivities (IWP) in the treatments were pretty low. Because quinoa has produced grain yield even dry conditions with an amount of yield reduction (Table 1).

Table 1. Seasonal irrigation water applied (I), evapotranspiration (ET), grain yield (GY), water productivity (WP), irrigation water productivity (IWP) and relative yield reduction (YR) of quinoa for different irrigation strategies in Adana.

Treatments I (mm) ET (mm) GY (t ha-') WP (kg m-3) IWP (kg m-3) YR (%)

FIF 140 360a 1.96a 0.54 0.32 0.0

DIF-67 100 313bcde 1.68abc 0.54 0.17 14.3

DIF-33 60 284efg 1.64abcd 0.58 0.28 16.3

SI10-100 140 343ab 1.92a 0.56 0.29 2.0

SI10-67 100 294def 1.69abc 0.57 0.18 13.8

SI10-33 60 269fg 1.46cd 0.54 0.00 25.5

SI20-100 140 337abc 1.87ab 0.56 0.26 4.6

SI20-67 100 309bcde 1.64abcd 0.53 0.15 16.3

SI20-33 60 287efg 1.43cd 0.50 0.16 27.0

SI30-100 140 330abcd 1.91a 0.58 0.29 2.5

SI30-67 100 286efg 1.74abc 0.61 0.24 11.2

SI30-33 60 281efg 1.28d 0.46 0.00 34.7

PRD 80 304cdef 1.67abc 0.55 0.20 14.8

DRY 0 253g 1.51bcd 0.60 - 23.0

LSD (0.01) 36.2 0.36

In the experiment carried out in Tarsus, the amount of irrigation water applied by line-source sprinkler system was changed between 74 and 310 mm. Applied water decreased with distance from sprinkler line in a fairly linear manner. Seasonal crop water use varied from 222 mm in I5 and 456 mm in I1 treatment plots. ET was significantly influenced by irrigation levels and decreased with increasing distance from line-source.

The highest grain yield of 6.38 t ha-1 was obtained from I1 treatment plots which adjacent to the line-source. Grain yields significantly decreased with decreasing the amount of irrigation water and the lowest yields were attained from non irrigated treatment, I5. Relative yield reduction was calculated as 65.36% for this treatment. These results show that grain yield of quinoa varies considerable depending on differences in soil water contents and severe water deficit markedly decreased grain yield of quinoa. The WP ranged from 1.0 to 1.57 kg m-3 and the maximum WP recorded in I4 (Table 2).

Table 2. Seasonal irrigation water applied (I), evapotranspiration (ET), grain yield (GY), water productivity (WP), irrigation water productivity (IWP) and relative yield reduction (YR) of quinoa irrigated with drainage water in Tarsus.

Treatments I (mm) ET (mm) GY (t ha-') WP (kg m-3) IWP (kg m-3) YR (%)

I1 310 456a 6.38a 1.40 2.06 0

I2 235 397b 5.77ab 1.45 2.45 9.6

I3 164 348c 4.55b 1.31 2.77 28.7

I4 74 262d 4.11b 1.57 5.55 35.6

I5 0 222e 2.21c 1.0 - 65.36

LSD (0.01) 39.7 0.51

Considering the both of two field experiments, although high salinity in the root zone caused yield reductions at different rates depending on salinity level, quinoa survived in all cases and produced grain yield. At the same time, quinoa has maintained its productivity even under rainfed conditions but moderate deficit irrigation show better results in terms of grain yield. Similar results have been obtained in Denmark (Razzaghi et al., 2011), Italy (Cocozza et al., 2012) and in Morocco (Lavini et al., 2014). Koyro and Eisa (2008) mentioned that quinoa survives even when irrigated with 100% seawater. Razzaghi et al. (2012) referred that soil-drying during the grain filling stage significantly decreased the seed yields of quinoa Titicaca. Cocozza et al. (2012) suggested that a certain amount of

water supplied during flowering and grain filling is enough to stabilize quinoa yield even for severe deficit irrigation. Deficit irrigation can stabilize quinoa yields at a level that is significantly higher than under rainfed cultivation (Geerts et al., 2009).

3.2. SALTMED model

After successful calibration processes, SALTMED model predictions on soil moisture, grain yield and dry matter yield were compared separately with two field experiment results. The simulated and observed total dry matter and grain yield for drip irrigation treatments in Adana are presented in Table 3. The relative error ranged between -13.47% and 9.16% for simulation of dry matter. Although there were slightly overestimated and underestimated values, the RE values for fourteen treatments were lower than 10% except two treatments and mean RE was 0.57%. The RMSE and NRMSE were calculated as 0.39 and 0.07, respectively. These indices that close to zero point to an acceptable agreement between simulated and observed dry matter yields. The calculated index of agreement (d) strengthens this result with the value of 0.86. Figure 1a also shows a good correlation between observed and simulated dry matters with R2 equal to 0.78.

Table 3. Comparison between simulated and observed total dry matter and grain yield of quinoa irrigated with different irrigation strategies in Adana.

Dry matter yield (t ha-1) Grain yield (t ha-1)

Treatments Observed Simulated RE (%) Observed Simulated RE (%)

FIF 7.04 6.45 8.38 1.96 1.81 7.65

DIF-67 6.04 6.21 -2.81 1.68 1.74 -3.57

DIF-33 5.27 5.81 -10.25 1.64 1.8 -9.76

SI10-100 6.52 5.99 8.13 1.92 1.74 9.38

SI10-67 5.38 5.4 -0.37 1.69 1.67 1.18

SI10-33 5.11 5.13 -0.39 1.46 1.5 -2.74

SI20-100 6.64 6.12 7.83 1.87 1.71 8.56

SI20-67 5.21 5.4 -3.65 1.64 1.68 -2.44

SI20-33 4.9 5.56 -13.47 1.43 1.61 -12.59

SI30-100 6.99 6.35 9.16 1.91 1.71 10.47

SI30-67 6.02 5.56 7.64 1.74 1.61 7.47

SI30-33 5.12 5.19 -1.37 1.28 1.30 -1.56

PRD 6.09 5.99 1.64 1.67 1.61 3.59

DRY 5.11 5.24 -2.54 1.51 1.57 -3.97

Mean RE (%) 0.57 0.83

RMSE 0.39 0.12

NRMSE 0.07 0.07

d 0.86 0.84

SALTMED Model predicted grain yields for different treatments with a maximum of -12.59% differences among the observed grain yield values. The mean RE was 0.83%. The RMSE was equal to 0.12 and NRMSE to 0.07. The index of agreement for grain yield was determined as 0.84 (Table 3). The relationship between observed and predicted grain yield obtained with 0.64 R2 value (Figure 1b). The efficiency criterion, d, confirms the closeness of the simulated grain yield to observed ones with the calculated value of 0.84. All these statistical parameters revealed that SALTMED model predicts dry matter and grain yield in a reasonable level of accuracy although slightly over or under-predictions.

Fig. 1. Correlations between observed and simulated dry matter (a); and between observed and simulated grain yield (b) for quinoa grown in Adana.

For the experiment carried out in Tarsus, the comparisons between simulated and observed total dry matter and grain yield are presented in Table 4. SALTMED model show better performance on estimation of dry matter and grain yield of quinoa with lower differences for Tarsus than those in Adana. The RE ranged between -3.85 and 0.98% for dry matter and -3.74 and 0.49% for grain yield. The RMSE and NRMSE were obtained as 0.19 and 0.02 for dry matter and 0.10 and 0.02 for grain yield, respectively. The index of agreement, d, point to a nearly perfect estimation of both dry matter and grain yield by SALTMED model. Figure 2 shows very good correlations between observed and predicted values of dry matter and grain yield with R2 value of 0.99. The model has correctly estimated the effects of water deficits on dry matter and crop yield of quinoa irrigated with drainage water in Tarsus.

Table 4. Comparison between simulated and observed total dry matter and grain yield of quinoa irrigated with drainage water in Tarsus.

Dry matter yield (t ha'1) Grain yield (t ha'1)

Treatments Observed Simulated RE (%) Observed Simulated RE (%)

I1 14.24 14.23 0.07 6.38 6.38 0

I2 10.94 10.97 -0.27 5.77 5.81 -0.69

I3 8.31 8.63 -3.85 4.55 4.72 -3.74

I4 7.13 7.06 0.98 4.11 4.09 0.49

Mean RE (%) -0.77 -0.99

RMSE 0.19 0.10

NRMSE 0.02 0.02

d 0.99 0.99

Fig. 2. Correlation between observed and simulated dry matter (a); and between observed and simulated grain yield (b) for quinoa grown in Tarsus.

The results obtained for total dry matter and grain yield are in agreement with Hirich et al. (2012) who predicted dry matter of quinoa grown in south Morocco with SALTMED model obtaining a R2 of 0.98. In the mentioned research, authors referred that the SALTMED model has been successful to predict also grain yield with an average

RE values of 2.57 and with a good correlation (R2 = 0.96) between simulated and observed data. Pulvento et al. (2013) found a relationship between measured and predicted final yield with R2 of 0.95 and RMSE of 0.19 for quinoa under different irrigation strategies with saline and fresh water. Silva et al. (2013) reported R2 of 0.99 simulating grain and biomass yields of chickpea in Portugal under wet and dry year conditions.

The correlations between the observed and estimated soil moisture in three soil layers in Adana conditions are shown in Figure 3 a-c. The linear relationships obtained between measured and predicted soil water content for 0-30 cm, 30-60 cm and 60-90 cm soil layers with R2 values of 0.86, 0.93 and 0.80. The calculated mean RE, RMSE and NRMSE values in estimation of soil moisture content for 0-30 cm soil layer were 0.02, 0.03 and 0.09 respectively. These statistics and the obtained index of agreement, d, value of 0.95 indicated that the SALTMED model predicted reasonable well the soil moisture in the first 30 cm soil layer. For second soil layer (30-60 cm), the model estimated soil moisture with -3.82% RE. The RMSE was computed as 0.02 and the NRMSE as 0.05. According to these statistical indices, the estimation of soil water content for second 30 cm soil depth was fit to observed data. The d value of 0.95 for this layer clearly confirms the accuracy of predictions by the SALTMED model. For 60-90 cm soil layer, although the high determination coefficient R2 value (R2 = 0.80), -8.60% mean RE indicated that the model-predicted soil water content values were slightly differ from observed values. In addition, the computed index of agreement value (d = 0.58) for this soil depth was lower than d values calculated for first two soil layers. These differences between the observed and estimated soil water content in this layer might be due to running the model by neglecting the soil drainage properties because of the missing data. Nevertheless, the RMSE and NRMSE point out to a good agreement between the model-simulated and observed soil moisture with values of 0.04 and 0.11, respectively. The SALTMED model estimated the soil water content within an acceptable accuracy level for last 30 cm soil layer.

Fig. 3. Correlation between observed and model-simulated soil water content (SWC) for 0-30 cm (a); 30-60 cm (b) and 60- 90 cm (c) soil layer in Adana.

For the experiment carried out in Tarsus, SALTMED model was run for simulation of soil water content in a total 90 cm soil depth. The observed and predicted soil moistures in 0-90 cm soil layer for four treatments are shown in Figure 4 a-d in order to visually comparison of the data. The soil moisture data of I1 treatment was used for calibration. Therefore almost all points that represent to observed and simulated data overlap with each other. There was a good overlap in I2 treatment but in I3 and I4 treatments the model show a tendency to a slight overestimation of soil moisture beginning from the mid-season to the end of the growth season. Even so, soil moisture estimations of the model seem matched with observed data. There was a good correlation between observed and model-simulated soil water content in 90 cm depth and the coefficient of determination R2 was over 0.75 (Fig. 5). The RMSE and NRMSE were calculated as 0.03 and 0.09 respectively. The Willmott's index of agreement, d, was equal to 0.81. Although mean RE value of -6.54% call attentions to above mentioned mild overestimations of the model, the RMSE, NRMSE and d clearly confirmed that the SALTMED model predicted the soil water content with high accuracy.

These results are in agreement with the study was carried out in Italy by Pulvento et al. (2013) and the study by Hirich et al. (2012), which was carried out in Morocco. Hirich et al. (2012) obtained very good agreement between simulated and observed soil moisture in their study for quinoa under Moroccan climate conditions. Pulvento et al. (2013) found a good correlation with R2 of 0.84 between observed and simulated soil moisture content in 0-36 cm soil layer. Silva et al. (2013) obtained the linear relationships between observed and simulated soil water content data with R2 values over 0.78 for chickpea in Portugal and referred that there was a good agreement between mentioned data for all soil depths. Hirich et al. (2014) suggested that the SALTMED model was able to predict soil moisture in 0-30 cm and 30-60 cm soil layers for sweet corn with 0.06 and 0.05 RMSE values.

Fig. 4. Observed and simulated soil water content (SWC) in 90 cm soil depth for I1 (a); I2 (b); I3 (c); and I4 (d) treatments in Tarsus.

Fig. 5. Correlation between observed and model-simulated soil water content (SWC) data for 90 cm soil depth in Tarsus.

4. Conclusions

The SALTMED model performance was evaluated by comparing predicted and observed data from two field experiments conducted in two locations in south part of Turkey. According to experimental results, quinoa could cope with salinity in the root zone up to ECw 30 dS m-1, although high salinity caused an amount of yield reduction. In addition, quinoa has maintained its productivity even under rainfed conditions but moderate deficit irrigation show better results in terms of grain yield. Therefore, moderate deficit irrigation may be suggested as an irrigation strategy to obtain acceptable yields in water scarce arid regions. Saline water and drainage water may be used for quinoa irrigation but absolutely with an effective management strategy in order to conservation of soil and water resources. On the other hand, quinoa has the potential to be an alternative crop especially for salt affected areas of Turkey.

The SALTMED model simulated the effects of irrigation water salinity and water deficits on dry matter and grain yield of quinoa within the acceptable limits. The graphical and statistical comparisons confirm the ability of the SALTMED model to predict soil water content. Considering together the comparison results for two locations, it can be concluded that the SALTMED model was able to accurately simulate soil water content, dry matter and grain yield of quinoa under various soil water and salinity conditions. The SALTMED model proved to be an efficient

irrigation management tool for quinoa irrigated with different irrigation systems, irrigation strategies and water qualities in Mediterranean conditions of Turkey.

Acknowledgements

This research was financially supported by the EU 7th Framework Programme through the project "Sustainable water use securing food production in dry areas of the Mediterranean region (SWUP-MED)". The authors would like to thank Dr. Ragab Ragab for his supervision and his efforts during the SALTMED modeling workshop at Cukurova University, Adana in May 2013.

References

Cocozza, C., Pulvento, C., Lavini, A., Riccardi, M., d'Andria R., Tognetti, R., 2012. Effects of Increasing Salinity Stress and Decreasing Water Availability on Ecophysiological Traits of Quinoa (Chenopodium quinoa Willd.). Journal of Agronomy and Crop Science 199, 229-240. FAO, 2013. Food Outlook Biannual Report on Global Food Markets. ISSN: 0251-1959.

http://www.fao.org/docrep/018/al999e/al999e.pdf. Garcia, M., Raes, D., Jacobsen, S.-E., 2003. Evapotranspiration Analysis and Irrigation Requirements of Quinoa

(Chenopodium quinoa Willd.) in The Bolivivan Highlands. Agricultural Water Management 60, 119-134. Geerts, S., Raes, D., Garcia M., Condori, O., Mamani, J., Miranda, R., Cusicanqui, J., Taboada, C., Yucra, E., Vacher, J., 2008. Could Deficit Irrigation Be a Sustainable Practice for Quinoa (Chenopodium quinoa Willd.) in The Southern Bolivian Altiplano? Agricultural Water Management 95, 909- 917. Geerts, S., Raes, D., Garcia, M., Taboada, C., Miranda, R., Cusicanqui, J., Mhizha, T., Vacher, J., 2009. Modeling the Potential for Closing Quinoa Yield Gaps under Varying Water Availibility in The Bolivian Altiplano. Agricultural Water Management 96, 1652-1658. Golabi, M., Naseri, A.A., Kashkuli, H.A., 2009. Evaluation of SALTMED Model Performance in Irrigation and

Drainage of Sugarcane Farms in Khuzestan Province of Iran. Food, Agriculture & Environment 7, 874-880. Hanks, R.J., Keller, J., Rasmussen, V.P., Wilson, G.D., 1976. Line-Source Sprinkler for Continuous Variable

Irrigation-Crop Production Studies. Soil Science Society of America Journal 40, 426-429. Hirich, A., Choukr-Allah, R., Ragab, R., Jacobsen, S-E., El Youssfi, L., El Omari, H., 2012. The SALTMED Model

Calibration and Validation Using Field Data from Morocco. J. Mater. Environ. Sci. 3, 342-359. Hirich, A., Ragab, R., Choukr-Allah, R., Rami, A., 2014. The Effect of Deficit Irrigation with Treated Wastewater

on Sweet Corn: Experimental and Modelling Study Using SALTMED Model. Irrigation Science 32, 205-219. Jensen, C.R., Jacobsen, S.E., Andersen, M.N., Núñez, N. S., Andersen, D., Rasmussen, L., Mogensen, V. O., 2000. Leaf Gas Exchange and Water Relation Characteristics of Field Quinoa (Chenopodium quinoa Willd.) During Soil Drying. European Journal of Agronomy 13, 11-25. Jacobsen, S.E., 2003. The Worldwide Potential for Quinoa (Chenopodium quinoa Willd.). Food Reviews International 19, 167-177.

Montenegro, SG., Montenegro, A., Ragab, R., 2010. Improving Agricultural Water Management in The Semi-Arid

Region of Brazil: Experimental and Modelling Study. Irrigation Science 28, 301-316. Koyro, H-W., Eisa, S.S., 2008. Effect of Salinity on Composition, Viability and Germination of Seeds of

Chenopodium quinoa Willd. Plant Soil 302, 79-90. Lavini, A., Pulvento, C., d'Andria, R., Riccardi, M., Choukr-Allah, R., Belhabib, O., Yazar, A., Ince Kaya, Q., Sezen, S. M., Qadir, M., Jacobsen, S.-E., 2014. Quinoa's Potential in The Mediterranean Region. Journal of Agronomy and Crop Science 200, 344-360. Pulvento, C., Riccardi, M., Lavini A., D'andria R., Ragab, R., 2013. SALTMED Model to Simulate Yield and Dry Matter for Quinoa Crop and Soil Moisture Content under Different Irrigation Strategies in South Italy. Irrigation and Drainage 62, 229-238.

Ragab, R., 2002. A Holistic Generic Integrated Approach for Irrigation, Crop and Field Management: The

SALTMED Model. Environmental Modelling & Software 17, 345-361. Ragab, R., Malash, N., Abdel Gawad, G., Arslan, A., Ghaibeh, A., 2005a. A Holistic Generic Integrated Approach for Irrigation, Crop and Field Management: 1. The SALTMED Model and Its Calibration Using Field Data from Egypt and Syria. Agric. Water Management 78, 67-88. Ragab, R., Malash, N., Abdel Gawad, G., Arslan, A., Ghaibeh, A., 2005b. A Holistic Generic Integrated Approach for Irrigation, Crop and Field Management: 2. The SALTMED Model Validation Using Field Data of Five Growing seasons from Egypt and Syria. Agric. Water Management 78, 89-107. Razzaghi, F., Ahmadi, S. H., Adolf, V. I., Jensen, C. R., Jacobsen, S.-E., Andersen, M. N., 2011. Water Relations and Transpiration of Quinoa (Chenopodium quinoa Willd.) under Salinity and Soil Drying. Journal of Agronomy and Crop Science 197, 348-360.

Razzaghi, F., Ahmadi, S. H., Jacobsen, S.-E., Jensen, C. R., Andersen, M. N., 2012. Effects of Salinity and Soil-Drying on Radiation Use Efficiency, Water Productivity and Yield of Quinoa (Chenopodium quinoa Willd.). Journal of Agronomy and Crop Science 198, 173-184. Ruiz, K.B., Biondi, S., Oses, R., Acuña-Rodríguez, I.S., Antognoni, F., Martinez-Mosqueira, E.A., Coulibaly, A., Canahua-Murillo, A., Pinto, M., Zurita-Silva, A., Bazile D., Jacobsen S.E., Molina- Montenegro, M. A., 2014. Quinoa Biodiversity and Sustainability for Food Security under Climate Change. A Review. Agronomy for Sustainable Development 34, 349-359. Silva, LL., Ragab, R., Durante, I., Lourenfo, E., Simoes, N., Chaves, MM., 2013. Calibration and Validation of SALTMED Model under Dry and Wet Conditions Using Chickpea Field Data From Southern Portugal. Irrigation Science 31, 651-659.

USSL, 1954. Diagnosis and Improvement of Saline and Alkali Soils. In: Richards L. A. (Ed.). USDA Handbook 60, Washington, pp.160.

Willmott, C. J., 1981. On The Validation of Models. Physical Geography 2, 184-194.