Optimization of deficit irrigation and nitrogen fertilizer management for peanut production in an arid region

Deficit irrigation (DI) has been emerging as an important technique for enhancing crop water productivity (WP). However, advantage of DI under varying nitrogen (N) application rates remains unclear. Field experiments were conducted during 2012–2014 to investigate the impacts of six irrigation levels[FI (full irrigation), DI10, DI20, DI30, DI40 and DI50, with irrigation amount of 100, 90, 80, 70, 60 and 50% of ETc, respectively) and four N application rates (N0, N10, N20 and N30, with 0, 10, 20 and 30 kg N ha−1, respectively) on WP, yield, quality, and net economic return of peanut in hot arid region of India. We used Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) method to obtain the optimal combination of irrigation and N rates. Both irrigation level and nitrogen dose had significant effects on yield and quality parameters examined in the study. Relative to FI, DI40 and DI50 significantly reduced yield (40.2–62.1%), economic benefit (70.8–118.5%), WP (8.2–33.0%), and kernel oil content (7.5–11.9%), but DI20 increased WP by 17.1% with only marginal reduction in economic benefit (2.6%), and yield (3.2%). Compared to N0, the N30 had 1.7, 1.1, and 1.6-folds increased yield, oil content in the kernel, and WP, respectively. Among all treatments, DI0N30 had the greatest yield and net return; DI20N30 had greatest WP and oil content in the kernel. TOPSIS analysis showed that DI20N30 was optimal in balancing of WP, yield, net return, and quality of peanut in northwestern arid India. The results have direct implications for improving irrigation water and N management for irrigated crops in arid regions.

Demand for agricultural products will further increase 1 to satisfy the needs of an increasing population. However, the availability of water for agriculture has been declining due to an increasing demand of water for nonagricultural sectors 1,2 . Shortage of irrigation water is seriously affecting agricultural production particularly in arid and semi-arid regions because irrigated agriculture is required for agricultural production in these regions. In the light of diminishing water resources for agriculture and increasing demand for crop production, irrigation strategies need to be devised to maximize crop water productivity (WP) 3 . Deficit irrigation (DI) which involves an application of the amount of irrigation water lesser than the full crop evapotranspiration (ET) is emerging as an important technique to enhance WP 4,5 . It had been reported that DI increased WP with considerable saving of irrigation water in many crops particularly in arid and semi-arid regions [6][7][8][9] . Adu et al. 10 reported huge variations in yield response of different crops while analysing relative yield performances of 43 crops grown in 14 countries under DI and full irrigation (FI) suggesting that DI require crop and region specific information on suitable magnitude of irrigation deficit.Thus, implementation of DI requires crop specific information related to identification of suitable magnitude of irrigation deficit, and agronomic management for its effective use 11 .
The crop yield and WP are affected by climate, crop species, soil, crop management practices, and choice of cultivar [12][13][14] . Li et al. 14 reported that agronomic practices influenced WP more than climatic factors; and from among various agronomic management, fertilizer rate and irrigation contributed 42.3% and 32.8%, respectively to the increase of WP. Soil nutrients directly influence photosynthesis and improve utilization of water by crops 13,15 .

Results
Amount of irrigation, AET and yield-water relationship. Amount of irrigation application varied from 330.6 mm to 591.2 mm during three seasons ( Table 1). The DI 10 , DI 20 , DI 30 , DI 40 and DI 50 had 8.8, 17.6, 26.5, 35.3 and 44.1% reduction in amount of irrigation application compared to FI. The irrigation levels (I), N application rates (N) and their interaction (I × N) had significant (P < 0.05) effects on crop water use measured in terms of actual evapotranspiration (AET) ( Table 1). The AET varied from 515.6 mm to 763.2 mm. The FIN 30 had significantly higher AET than all other treatments. The AET declined with a decrease in irrigation; and averaged across years and N rates, the AET under DI 10 , DI 20 , DI 30 , DI 40 and DI 50 was decreased by 6.1, 15.9, 20.6, 24.1 and 31.0%, respectively than under FI. The N application enhanced AET; and effects of N application on AET varied with irrigation levels. The N 30 under FI and DI 10 had significantly (P < 0.05) greater AET than in other N rates. Under DI 40 and DI 50 , the difference in AET among N 10 , N 20 and N 30 was not significant (P > 0.05), but these treatments had significantly greater AET than under N 0 .
Relationship between AET and irrigation amount was linear and had a high (0.98) coefficient of determination (R 2 ) (Fig. 1a). Slopes of the linear relationship of AET with irrigation water applied were found 0.90, suggesting that about 90% of applied water is converted to AET. Pod yield exhibited a quadratic relationship with amount of irrigation applied (Fig. 1b), implying that after a certain level of irrigation amount (~ 500 mm), a further increase in irrigation amount will not result into the same amount of increase in pod yield rather it will be less. Growth, yield and quality. The growth (leaf area index, dry matter production) (Fig. 2), yield components (pod number, kernel number, 100-kernel weight), biomass partitioning (pod harvest index, shelling fraction), yield (biomass yield, pod yield, kernel yield) (Table 1), and quality (Table 2) of peanut were significantly (P < 0.05) affected by irrigation levels, N application rates and their interaction.
The growth, yield components and yield decreased with a reduction in irrigation and N application rates. Relative to FI, the DI 50 had significant reduction in growth (24.2 to 30.1%) (Fig. 2), yield components (15.2 to 60.8%) and yield (55.9 to 67.6%) ( Table 1). Averaged across years and irrigation levels, N 30 enhanced yield (1.6 to 2.0-folds), yield components (1.3 to 2.0-folds), and growth (1.4 to 1.8-folds) compared to that of without N application (N 0 ). The effects of N application rates on growth, yield components and yield were modified with irrigation levels. The N 30 under FI and DI 10 had greater (P < 0.05) growth and yield components than other treatments. Under DI 40 and DI 50 , there was no significant (P > 0.05) difference in pod number and kernel number among N 30 , N 20 and N 10 , but the values were greater (P < 0.05) than those of N 0 . The N 30 under FI had greater (P < 0.05) biomass yield (11.0 to 237.9%) and pod yield (10.9 to 346.3%) than other treatments. The FIN 30 and DI 10 N 30 had greater (P < 0.05) kernel yield (7.4 to 481.1%) than other treatments. The N 30 under FI, DI 10 and DI 20 had significantly greater pod and kernel yields than other N rates. The difference in yields among N 10 , N 20 and N 30 under DI 50 were not significant, but yields were greater (P < 0.05) than those of N 0 .
The biomass partitioning measured in terms of pod harvest index and shelling fraction enhanced with a reduction in irrigation from FI to DI 20 ; and further reduction in irrigation declined biomass partitioning (  20 had significantly greater pod harvest index and shelling fraction than other treatments. The quality attributes of kernel measured in terms of oil and protein contents significantly affected by irrigation levels, N application rates and their interactions ( Table 2). Oil and protein content of kernels varied from 4.29 to 5.26 g kg −1 and 2.23 to 2.74 g kg −1 , respectively. The oil contents in kernel declined with a reduction in irrigation and N application rates. The N 30 under DI 20 irrigation level had highest oil content followed by N 30 under DI 10 and FI irrigation levels. A significant increase in oil contents were recorded up to application of N 20 under all irrigation levels. The protein content of kernel enhanced with a reduction in irrigation. Averaged across years and N rates, DI 30 had highest protein contents in kernel followed by DI 20 , DI 40 , DI 50 , DI 10 , and FI. The DI 20 , DI 30 and DI 40 had greater (P < 0.05) protein content than those under other irrigation levels. The N application enhanced protein content in the kernel. The DI 30 N 30 had greater (P < 0.05) protein content in kernel than other treatments.
Water productivity and nutrient-use efficiency. Irrigation levels, N rate and their interaction had significant (P < 0.05) effect on WP (WP AET : water productivity in terms of AET, and WP I : water productivity in terms of irrigation water applied) (Fig. 3), and NUE (Fig. 4). The WP AET and WP I varied from 0.176 to 0.597 kg m −3 and 0.275 to 0.791 kg m −3 , respectively. The DI 20 N 30 had highest WP which was significantly greater and had advantage to the tune up to 239.6% for WP AET and up to 193.0% for WP I ) over other treatments. Averaged across years and N rates, the WPs increased with a reduction in irrigation up to DI 20 . However, with further reduction in irrigation (DI 40 and DI 50 ) declined WPs. The DI 20 had highest WP (WP AET : 0.481 kg m −3 , WP I : 0.633 kg m −3 ) which was significantly greater (from 9.5 to 110.2% for WP AET and from 8.3 to 74.8% for WP I ) than other irrigation levels. The N application augmented WP, and N 30 had 64.4 and 66.6% greater WP AET and WP I , respectively than with N 0 .
The NUE measured in terms of partial factor productivity of N (PFP n ) varied significantly among different treatments (Fig. 4). The PFPn declined with a reduction in irrigation and an increase in N application rates. The N 10 under FI, DI 10 and DI 20 irrigation levels had similar PFPn, which was significantly greater than other Table 1. Amount of irrigation applied, actual evapotranspiration (AET), yield components and yields of peanut under different irrigation and nitrogen application rate treatments [Values are mean ± standard error, n = 9 (3 × 3; 3 growing seasons and 3 replications per treatment)]. Values within a column for each variable followed by different letter/s are significantly different at P < 0.05 level. FI: full irrigation means irrigation equaled to 100% ETc; DI 10 :10% deficit irrigation means irrigation equaled to 90% ETc; DI 20 : 20% deficit irrigation means irrigation equaled to 80% ETc; DI 30 : 30% deficit irrigation means irrigation equaled to 70% ETc; DI 40 : 40% deficit irrigation means irrigation equaled to 60% ETc; DI 50 : 50% deficit irrigation means irrigation equaled to 50% ETc. N 0 : 0 kg N ha −1 ; N 10 : 10 kg N ha −1 ; N 20 : 20 kg N ha −1 ; N 30 : 30 kg N ha −1 . Economic benefits. The cost of cultivation varied considerably (from US $ 979.7 to US$ 1299.0) among different treatments ( Table 3). The irrigation, N rates and their interaction significantly affected economic returns. The returns increased with an increase in irrigation and N application rates. The DI 40 along with N 0 and DI 50 with all N rates had economic loss (negative net return).The FI, DI 10 and DI 20 had similar net return (from US $ 804 to 826 ha −1 ) which was significantly greater (from US $ 349.5 to 776.7 ha −1 ) than those in other irrigation levels. The N application enhanced returns. The FIN 30 had greatest return, and the DI 50 N 0 had lowest return.
Comprehensive evaluation based on TOPSIS. The evaluation of peanut production under 27 treatments by using TOPSIS (  50 irrigation levels and N 0 and N 10 nitrogen application rates were not conducive for acquiring comprehensive benefits. The comprehensive benefit was positively related to pod yield, net return, oil content in kernel, and WP I ; and negatively related to amount of irrigation water applied.
Discussion. Shortage of water is the major constraints for limiting crop yield in arid and semi-arid areas 34,35 ; and improving effective utilization of water is an urgent need for sustainable crop production in these areas 6,7,36 . Deficit irrigation (DI) has been emerging as an effective practice to improve WP, and saving of water [4][5][6][7] . Prior to this study, little information exist on productivity, resource use efficiency (WP and NUE), and quality for peanut under varying irrigation and N application rates. Our results showed that the studied parameters were In the present study, the DI saved irrigation water (from 8.8 to 44.1%) and decreased crop water consumption (AET, from 6.1 to 31.0%) compared to FI (Table 1). This reduction in water saving and AET compromised yield; and magnitude of yield decline depended on reduction in irrigation. DI with > 20% reduction in irrigation (DI 30 , DI 40 and DI 50 ) caused significant reduction in yield compared to FI. An adequate production of dry matter and its translocation to sinks are major determinants of yield; and water-deficit strongly influences them [37][38][39][40] . The yield decline with a reduction in irrigation observed in this study could be attributed to decrease in dry matter production (Fig. 2d-f) and biomass partitioning (Table 1), which resulted in decrease in yield components (Table 1). Another reason for a reduction in yield with reduced irrigation application might be an adverse effect on pegging 39,40 . The pod yield and kernel yield exhibited less reduction than biomass yield under moderate DI (DI 20 ) level, suggesting that moderate DI stimulated biomass partitioning (higher pod harvest index and shelling fraction) resulting in marginal reduction in pod and kernel yield under DI 20 observed in the present study.
In this study, N application increased yield (Table 1), and magnitude of which depended on irrigation levels (Fig. 5). Adequate N availability is essential for boosting peanut productivity 29,41 . The significantly higher yield with N-fertilization in our study was directly attributed to greater expression of yield contributing characters ( Table 1) that were a result of the enhanced DM ( Fig. 2d-f) and biomass partitioning (Table 1) due to N application. Our results showed that yield responses to N rates decreased with a reduction in irrigation. An increase in soil water availability positively affects mineralization and physical transport of N to roots, all of which increase plant-available N for uptake 42 . Limitation to mineralization and/or transport of N to roots surface in reduced    Table 1. www.nature.com/scientificreports/ water-supply conditions might be responsible for lack of yield response to higher N rates with a reduction in irrigation observed in the present study. The findings of our study provide direct evidence that application of N-fertilizer is effective to increase yield of peanut in the areas having soil with low content of N, and proper combination of irrigation and fertilizer can achieve the optimal coupling effect and optimize yield. There are reported contrasting effects of DI for WP of peanut, ranging from an increase 43,44 to a decrease 20,24,45 of WPs. In the present study, WPs initially increased up to 20% reduction in irrigation (DI 20 ); and further decreased with greater reduction in (DI 30 , DI 40 and DI 50 ) (Fig. 3). Relatively greater reduction in AET (15.9% reduction compared to FI) and amount of irrigation water application (17.6% reduction compared to FI) than reduction in pod yield (3.3% reduction compared to FI; Table 1) is responsible for higher WPs under moderate DI (DI 20 ) as compared to FI observed in this study. Conversely, relatively greater reduction in yield (from 40.1 to 62.2% reduction in pod yield than FI) than reduction in AET (from 25 to 30% reduction than FI) and amount  Table 1. Table 3. Cost and economic benefits of peanut under different irrigation and N application rates [Values are mean ± standard error, n = 9 (3 × 3; 3 growing seasons and 3 replications per treatment)]. Values within a column for each variable followed by different letter/s are significantly different at P < 0.05 level. The details of treatments are given in Table 1.  (Fig. 3), because of improved yield without much affecting crop water consumption (AET) ( Table 1). Enhanced leaf area index (Fig. 2a-c) which lead to reduction in evaporation component of ET 13 along with increase in yield due to enhanced dry matter production ( Fig. 2d-f) and partitioning of dry matter to sink (evident by higher pod harvest index and shelling fraction) might be responsible for enhancement of WPs due to N application observed in this study. It has been reported that addition of N in N-deficient soil increased WPs, provided irrigation water is not limited 6,7,44 . Based on the above-mentioned benefits of both moderate DI and N application, the highest WPs were observed with DI 20 N 30 . The NUE declined with a reduction in irrigation and an increase in N rates (Fig. 4).The N10 under FI, DI 10 and DI 20 had significantly greater NUE than other treatments. Adequate soil water positively affects availability (via mineralization and physical transport of ions in soil), uptake, and utilization of nutrients by plants and resulting in greater yield and NUE. Restricted soil water availability reduces the access of root to the mass flow of N leading to low uptake and utilization of N and a subsequent decrease in NUE 42,46 . The lower increase in yield than corresponding increase in N application rates is the major reason for declining in NUE with an increase in N rates observed in the present study. Our results showed that increasing N application rates, the WP increased (Fig. 3) and that of NUE declined (Fig. 4); suggesting the trade-off relationship between WP and NUE. But maintaining a higher WP with better yield at cost of NUE is more suitable for peanut production in water scarce areas like northwestern hot arid region of India.

Treatments Cost of cultivation (US $ ha −1 ) Gross return (US $ ha −1 ) Net return (US $ ha −1 )
Previous researchers reported contrary results regarding effect of water deficit on oil and protein content of kernel. There are reports that water-deficit had no effects 47 , and decline 27 in oil content in kernel; no-consistent effect 47 , and increase 48 in protein content in kernel of peanut. In this study, oil content of kernel declined, and protein content increased with a reduction in irrigation. Both oil and protein contents enhanced with an increase in N rates ( Table 2). The enhanced oil and protein content of kernel due to N application is associated with the Table 4. TOPSIS analysis of irrigation water applied, yield, water productivity, oil content in kernel, and economic benefit for different irrigation and N application rate treatments for peanut. X 1 , X 2 , X 3 , X 4 and X 5 represent normalized value of pod yield, economic benefit, kernel oil content, irrigation water productivity (WP I ), and amount of irrigation water applied, respectively. Z + and Z − are the positive and negative ideal solutions, respectively. Di + and Di − are the distance between each alternative and positive and negative ideal solution, respectively. C i is the comprehensive benefit evaluation index for different treatments. R is Spearman correlation coefficient between comprehensive benefit index rank and single attribute index rank and the asterisk (*) means that the R was significant at α = 0.05. www.nature.com/scientificreports/ facts that adequate N supply is required for both protein and oil synthesis 28 . Inadequacy of photo-synthates (due to water and N-deficit induced reduction in leaf area and photosynthesis), and increased oxidation of fatty acids 49 might be responsible for the reduction in oil content of kernel with a decrease in the irrigation and N application rates observed in the present study. Besides, water-deficit hinders the fat synthesis more than that of protein formation 50 is responsible for greater reduction in oil content than that of protein content with a reduction in irrigation observed in this study. Net return is an important criterion for evaluating irrigation and fertilizer management strategies for crop production. In this study, the net return increased with an increase in N rates under FI, DI 10 and DI 20 irrigation levels implying that increased cost with an increase in N rates in these irrigation levels balanced by value of improved yield due to the N application (Table 3). Cost incurred in all N rates under DI 50 was not compensated by value of yield under these treatments and thus had economic loss.
Our results showed that WP, yield, quality, NUE and economic benefit were determined not only by application rates of irrigation and N, but also by their interaction. Besides, our results also demonstrated that there was trade-off relationship among these parameters; consequently it was difficult to obtain the maxima of WP, yield, quality, and saving of irrigation water simultaneously. In this study, maximum yield and net return were achieved when FI with N 30 were applied, whilst WP and KOC were maximized with application of N 30 under DI 20 irrigation level. The TOPSIS was used to identify best alternative for optimizing WP, yield, quality, economic benefit, and saving of irrigation amount simultaneously. Since, the value of NUE is not possible to calculate in treatments involving N 0 nitrogen application rate, the NUE was not considered for TOPSIS analysis in this study. Results of TOPSIS showed that moderate DI (DI 20 ) in combination with N 30 (DI 20 N 30 ) could simultaneously achieve better yield, quality, return, irrigation water saving along with higher WP (Table 4).

Conclusions
Three-year field experimentation on investigation of effects of irrigation and N regimes on WP, yield, quality, and economic return from peanut cultivation in northwestern arid India demonstrated that all the parameters were determined by both irrigation level and application rate of N, and there is an optimal level of N rate for each level of irrigation. Therefore, N application rate should be adjusted according to level of irrigation available. Results also clearly showed that it was not possible to maximize yield, quality, return and WP simultaneously. Rather, TOPSIS analysis showed that moderate DI (~ 20% deficit) along with application of N @30 kg ha −1 could balance among productivity, economic return, quality, and resource utilization for peanut cultivation in northwestern arid India.   Experimental design. There were 24 treatments having six irrigation (I) levels and four nitrogen (N) fertilization rates. The experiment was conducted using a split-plot design with three replications for each treatment. Irrigation levels were laid out in main plots (9 m × 37 m) and N application rates were laid out in subplot (9 m × 7 m size) ( Supplementary Fig, SI 2). Irrigation amount equaled to 100% (representing no deficit and denoted by FI), 90% (denoted by DI 10  where E pan is pan evaporation (mm); K p is pan coefficient (0.75); and K c is crop coefficient, which varies for different growth stages of the crop 51 . Irrigation water was applied through sprinklers (double nozzle of 2.5 mm × 1.8 mm size with a discharge of 7.5 lpm at 2.5 kg cm −2 pressure, spaced at 10 m). A 70 mm pre-sowing irrigation was applied to each experimental plot. Subsequent irrigation water requirement (ET c ) was computed as ET c × irrigation efficiency (80%) as per irrigation levels. Irrigation was applied whenever cumulative ET c reached approximately 45 mm. The evaporation was measured daily using USWB Class A pan evaporimeter located at 500 m away from the experimental site. Nitrogen was applied through urea fertilizer (46%N) as per N levels during land preparation prior to planting. Measurements and analysis. Ten plants were randomly taken from each plot to measure the leaf area index (LAI) and dry matter (DM) at 30, 60 and 90 days after sowing (DAS). The leaf area was measured by a leaf area meter (Systronic India Ltd, Model 211), and LAI was calculated as the ratio of leaf area to harvested ground area 39 . The whole plants (leaf, stem, peg, and pod) were dried at 70 ºC to constant mass to determine DM. At maturity, total biomass yield and pod yield were recorded on central 4 m 2 from each subplot after manual harvesting. The yield components, i.e., number of pod m −2 , number of kernel m −2 , and 100-kernel weight were recorded at maturity from each plot. Harvested material was sun dried for 7 days and threshed separately for each subplot to record pod yield. The pods were shelled manually to determine kernel yield. The pod harvest index was calculated by dividing pod yield with biomass yield. The shelling fraction was calculated as the ratio of kernel yield to pod yield. For computing water consumed by the crop (AET), the soil profile water content was measured in 0-150 cm (with 20 cm increments for 0 to 100 cm, and 25 cm increments for 101 to 150 cm) soil by the thermo-gravimetric method. Volumetric soil water content was obtained by multiplying gravimetric water content by the corresponding bulk density and thickness of respective soil layers. The AET was calculated using the water balance method by monitoring the change in the soil water content prior to sowing and after harvest of crop:

Crop husbandry details.
where AET is the actual crop evapotranspiration (mm), P e is effective rainfall (mm), I is the irrigation (mm), U is upward capillary flow from ground water (mm), R is surface runoff (mm), D w is drainage out of the root zone (mm), and S is the change in soil water storage in the 0-150 cm layer (mm), from planting to harvest of crops. As the groundwater level at experimental site is > 75 m deep, the U was ignored. The D w was also considered zero as supported by the observation on negligible change in soil moisture content at soil depth > 120 cm. The R was also assumed negligible because the soil has good infiltration rate and each subplot was surrounded by 35 cm high bund. Therefore, the U, R and D w were considered zero for calculating AET 7 .
The water productivity (WP) was calculated using the equations: where WP AET and WP I , is water productivity (kg m −3 ) in terms of AET and volume of irrigation water applied, respectively; PY is pod yield (kg ha −1 ), and AET is actual evapotranspiration (m 3 ha −1 ) and W I is volume of irrigation water applied (m 3 ha −1 ). The partial factor productivity of N (PFP n ; kg kg −1 ) is an important measure for determining NUE. The PFP n was calculated by using the following equation: where PY is pod yield (kg ha −1 ); N fi is applied N as fertilizer in different treatments (kg ha −1 ).
The kernel oil content (KOC) was determined by the Soxhlet method 52 . Kernels were ground into a fine meal, and weighed quantity (2.5 g) of the meal was transferred to the thimble and extracted in 150 ml hexane in Soxhlet extraction assembly for 6 h. The solvent was then evaporated in an oven at 60ºC to a constant weight. The values of oil were calculated and expressed as g kg −1 (W W −1 ). To estimate kernel protein content, the N concentration was determined using the Kjeldahl method. A factor of 5.46 was used to convert N concentration into protein concentration 48 .
Economic benefit in terms of net return (NR) was calculated for each treatment. Components of cost of cultivation (CC) included investments of inputs (seed, fertilizer, pesticides), irrigation, labour and machinery cost for field operations (tillage, sowing, weeding, harvesting, threshing) and rental value of land. The cost of land was estimated using seasonal rental value of land prevailing in the region. Income from the sale of pod and straw yield was added to calculate gross return (GR). The income was calculated from current price prevailing in the study region. The NR was calculated using following equation: where NR is net return ($ ha −1 ), GR is gross return ($ ha −1 ), and CC is cost of cultivation ($ ha −1 ).

Multi-objective decision making and evaluation using TOPSIS.
The technique for order preference by similarity to ideal solutions (TOPSIS) was used to identify best alternative for achieving better yield, quality, WP, economic benefit simultaneously for peanut following Deng et al. 32 . Calculation involves five sequential steps: 1. Constructing the original evaluation parameter matrix by using 3-year averaged values of parameters: where x ij (i = 1,2,……..,n and j = 1,2,……,m) represents the j th measurement (pod yield, kernel oil content, irrigation water productivity, economic benefit and irrigation amount used) from i th treatment. In this study n = 24 and m = 5. 2. Construction of normalized decision matrix: For comparisons of attributes there is need of comparable scales which obtained by normalization. The vector normalization approach divides the rating of each attributes by its sum to calculate normalized value of Z ij as defined in Eq. (8).
The Z ij is the normalized x ij and W j is the weight of the jth evaluated index. In this study, all evaluated parameter treated equally, therefore, W j is 1. where 0 ≤ C i ≤ 1 When C i is closer to 1, the peanut had better comprehensive benefit in terms of yield, quality, net return and water productivity.
Statistical analysis. Data were statistically analyzed by ANOVA according to a split-plot design using Statistical Analysis System software. Data were subjected to Bartlett test for testing homogeneity of variance. The uniformity in error variance was not significant. The year × irrigation levels or year × N rate treatment effects on most of the parameters were not significant. Hence, the data were pooled and presented across the years. The means were compared using the least significant difference (LSD) test at 0.05 probability level.

Data availability
Data generated or analyzed during this study are included in this article (both as main file and supplementary information) and are available from corresponding author on request.