All the agricultural methodologies that can be managed but the unpredictable nature of weather remains beyond human control. Weather is a key reason to decide the fate of agricultural production. It shows magnifying impact on operations and farm management practices. To provide the reliable medium range forecast of particular district, the District Agro-Meteorology Units (DAMUs) project was initiated by the India Meteorological Department (IMD) in 2018, in collaboration with the Indian Council of Agricultural Research (ICAR). This weather based agromet advisory service useful to increase the awareness, helpful for perception and control of economic loss and obtaining better yields (Ramachandrappa et al., 2018). Agromet Advisory Services (AAS) beneficious to the farming community to take advantage of favorable weather and mitigate the impacts of extreme weather conditions through the inputs in advisory. The risks associated with agricultural activities can be reduced by convert weather information into needful advisory for agricultural implications. DAMUs contribute to collect and organize crop, soil and weather information and to process them with forecast information to help the farming community in decision making (Standard Operating Procedure for Agromet Advisory Services, 2020).
       
Cotton grown in the monsoon season of Telangana, which is supporting the seasonal agricultural cycles. After the paddy production (14,485.0 thousand tons in Kharif and 11,182.0 thousand tons in Rabi) cotton occupies good area with large production in Telangana. Telangana had cotton area 17.82 lakh ha and production of 48.95 lakh bales of 170 kg with 467 kg ha^-1 productivity (Anonymous, 2025).
       
Inter cropping is well known practice in crop production to utilize the resources efficiently, produce the additional income, enhance soil health, pest-disease prevention and weed suppression (Tandale et al., 2018). This is a traditional practice to create interaction between two or more crops at a time, these crops coexist for a significant part of their life and they interact among themselves and with ecosystem (Maitra and Gitari, 2020). In other hand without extra allocation of land to extra crop farmers can harvest a larger yield proportion from both the crops (Pandagale et al., 2019).
       
Red gram occupies 176.4 thousand hectares with 145.0 thousand tons production and green gram area 69,000 acres and production 20,000 tons (Telangana state at a glance, 2024). The large range of cotton area can provide the opportunity of intercropping with pulses in Telangana. Cotton and Red Gram intercropping benefit the soil health and resource management through organic matter accumulation in soil and increasing of organic carbon in soil through root residues and shedding of leaves (Ammaiyappan et al., 2023 and Adepu and Latha (2025).

Experimental site
 
Present study conducted by DAMU, Krishi Vigyan Kendra, Mancherial during three consecutive kharif seasons during the year of 2019-2020 to 2021-2022 at 5 locations (3 locations in Mancherial and 2 locations in K.B Asifabad district) of Telangana state.
       
The DAMU, Mancherial provided the information to the farmers of both the districts about Bi-weekly medium range weather forecast and AAS. In present investigation to understand the superiority and scope of cotton-red gram inter cropping system in Telangana provided AAS (Bi-weekly agrometeorological advisory bulletins) to cotton red gram inter cropping farmers based on forecast and real time rainfall data and suggested the management and cultural practices from land preparation to till the harvesting to obtain maximum yields.
 
Rainfall information
 
The Mancherial district received rainfall 1314.9, 1199.6 and 1145.0 mm in 2019-20, 2020-21 and 2021-22, respectively. K.B. Asifabad district received rainfall 1452.8, 1275.2 and 1662.7 mm in 2019-20, 2020-21 and 2021-22, respectively (Table 1).


       
In three consecutive kharif seasons (2019-20 to 2021-22) yield and economic results from the all 5 locations were observed in both AAS followed cotton- redgram and non-followed sole cotton fields. 
 
Yield analysis
 
Yield of AAS followed intercropping as well as traditional sole cotton was observed and converted into hectares for convenance.
 
Land equivalent ratio (LER)
 
Land equivalent ratio is one of the important calculations to evaluate the performance of an intercropping system given by Beets (1982). Later number of researchers improved the concept of LER (Fukai, 1993). LER is described as the proportionate land area required under pure stand of crop species to yield the same produce as obtained under an intercropping at the same management level.

Crop equivalent yield (CEY)
 
Crop equivalent yield (CEY) is the expression of both crops (Main and inter crop) yield into as main crop yield. Crop equivalent yield, which is useful in mixed cropping or intercropping or sequential cropping (De Wit, 1960).

 
Economic analysis
 
Gross return (Rs. ha^-1), Net returns (Rs. ha^-1) and Benefit cost ratio. 
 
Descriptive statistics and normality diagnostics
 
Data of above parameters categorized by district and village were gathered from two cropping systems over a three-year period (2019-20 to 2021-22).
       
The assumption of multivariate normality was confirmed prior to conducting any multivariate analysis. For tests like Wilks’ Λ, Pillai’s Trace, Hotelling-Lawley Trace and Roy’s Largest Root in MANOVA, as well as the Discriminant Function in Linear Discriminant Analysis, to be valid, the joint distribution of dependent variables must approximate a multivariate normal distribution (Royston, 1982).
       
Multivariate normality was assessed using Royston’s multivariate normality test, an extension of the Shapiro-Wilk W test to multiple variables. The procedure applies the Shapiro-Wilk test to each variable X_j (j = 1,2,....,p), obtaining statistics W_j, which are transformed into approximate standard normal scores: Z_j = - In (1 - W_j) These
 
are then combined into an overall test statistic: ,                           
where  and Z_j ~ N (0, 1) under the null hypothesis (H₀) of MVN.
       
Royston’s approach is better than alternatives like Mardia’s skewness-kurtosis tests and works well for small to moderate samples. A significant result (p<0.05) implies a departure from multivariate normality, whereas a non-significant p-value (p>0.05) shows no evidence against it.
 
Multivariate Analysis of Variance (MANOVA)
 
Once normality was established, a MANOVA was conducted to investigate how cropping systems affected several dependent variables: yield, cost of cultivation (CC), benefit-cost ratio (BCR), net returns (NR) and gross returns (GR). MANOVA controls Type I error inflation while enabling the testing of correlated dependent variables simultaneously (Wilks, 1932).
The model was specified as:

Y = XB + E
 

Where,
Y_nxp= The matrix of dependent variables (n observations in p dependent variables).
X_nxk= Design matrix of predictors (k groups or predictors).
B_kxp= Matrix of regression coefficients and E_nxp is error matrix.                 
The group effect was tested under H₀: B ≠ 0 (No group effect) and H₁: B = 0 (presence of group effect).
       
The hypothesis matrix (H) and pooled within-group error matrix (E) were computed as:

 
Where, 
_i i^th= Group mean vector.
= Overall mean vector.
n_i= Sample size in ith group.
 
MANOVA test statistics
 
Each test statistic is based on the eigenvalues.
i. Wilks’ Lambda (Λ): Tests the proportion of variance in the dependent variables not explained by the factor.
Where,
s= The number of non-zero eigen values of E^-1 H or min (p,dfg).

    
ii. Pillai’s trace (V): A robust statistic, particularly effective when assumptions such as homogeneity of covariance are not fully met (Pillai, 1955).

 

iii. Hotelling-lawley trace (T²): Measures the sum of the explained variances across canonical variates (Hotelling, 1931).

 
iv. Roy’s largest root (Θ): Tests the maximum separation between groups based on a single canonical variate (Roy, 1953).
 

Θ = max (λ_i)

 
Univariate analysis of variance (ANOVA)
 
Univariate ANOVAs tested mean differences of individual dependent variables (NR, CC, GR, BCR and Yield) across the two cropping systems. The model is:
 

Y_ij = μ + τ_i + ∈_ij 

 
Where,
Y_ij= Observation of the dependent variable in the i^th group and j^th replication.
μ= Overall mean of the dependent variable.
τ_i= Effect of the ith group.
∈_ij= Random error term.
∈_ij ~ N(0,σ²) (Montgomery, 2019).
       
Significance was determined using p-values (p<0.05) and the percentage contribution of each variable’s sum of squares (SS) to the total SS, indicating which traits most strongly explained group differences.

Linear discriminant analysis
 
Using the following explanatory variables, Linear Discriminant Analysis (LDA) was used to categorise and differentiate between two cropping systems: intercropping and sole cropping: Net returns (NR), benefit–cost ratio (BCR), cost of cultivation (CC), gross returns (GR) and yield. In order to minimise within-group variance and maximise between-group variance, LDA builds a linear combination of predictors (Fisher, 1936).
       
For each group g∈ {1,2,…,G}, the discriminant function is expressed as:

 

Where,
X= (x₁, x₂,…,x_p)′= Vector of predictor variables.
μ_g= The mean vector of group k.
Σ= The pooled within-group covariance matrix.
π_g= The prior probability of group g.
       
The pooled within-group covariance matrix is given by:

 
Here,
n_g= The sample size of group g.
S_g= The covariance matrix of group g.
       
For two groups, only a single discriminant function is extracted:
 

D(X) = W′X + c

 
Here,

W = Σ^-1 (μ₁ -μ₂)

 
The classification rule is:

 
The discriminant score for observation i is: z_i = w′x_i and the posterior probability of belonging to group k is computed as:

 
f_g (X)= The multivariate normal density for group g.
 
Implementation in R
 
R version 4.3.1′s tidyverse, psych and ggplot2 packages were used to perform descriptive statistics and graphical analyses for rainfall and yield by year, location and crop (R Core Team, 2023). The MVN package's Royston's MVN test was used to evaluate multivariate normality. The car package was used to perform the MANOVA.
       
The MASS, ggplot2 and pROC packages were used to implement the discriminant analysis, which produced discriminant scores and posterior probabilities to assess cropping systems' classification accuracy.

Yield performance
 
Over three seasons 2019-20, 2020-21 and 2021-22 and five locations, the cotton-redgram intercropping system in a 6:1 row ratio consistently outperformed sole cotton. The system maintained a uniform and robust plant stand, with about 15,357 cotton plants and 2,527 redgram plants per hectare. This configuration delivered a mean system yield of 2031 kg ha^-1, comprising 1537 kg ha^-1 from cotton and 494 kg ha^-1 from redgram, compared with 1690 kg ha^-1 under sole cotton (Table 2). The yield advantage reflects the complementary roles of the two crops: redgram fixes atmospheric nitrogen, improves soil physical conditions and helps buffer the microclimate within the canopy, which together support better boll retention and higher biomass in cotton.


       
These yield gains were also closely linked to the systematic use of Agro meteorological Advisory Services (AAS) throughout the study period. Location-specific weather-based advisories on sowing dates, irrigation scheduling, nutrient application and plant protection enabled farmers to time operations more precisely with short-and medium-range forecasts, lowering exposure to weather-related risks. Similar positive impacts of AAS on yields, income stability and risk reduction have been documented in other agro-climatic regions (Rakesh et al., 2025) and are consistent with national assessments of the strengthened GKMS/AAS framework (Rathore et al., 2025) and evidence that village-level agromet display boards and advisories promote adoption of weather-informed decisions (Roy et al., 2025).
 
Resource-use efficiency
 
Intercropping also clearly improved how land and inputs were used. The cotton–redgram system recorded a mean Land Equivalent Ratio (LER) of 1.30 and a mean Cotton Equivalent Yield (CEY) of 1993 kg ha^-1, representing roughly 15% higher productivity compared with sole cotton. An LER above 1.0 indicates that the intercropping system uses light, water and nutrients more efficiently per unit area than when crops are grown alone. This pattern agrees with findings from other cotton-based intercropping systems, including cotton-pigeonpea, where LER values typically exceed 1.0 and CEY is substantially higher under intercropping (Table 2). Latheef et al. (2018) gains better yield and economic returns through complementary canopy structures and more efficient resource use.
       
Our results are further supported by broader multicropping literature showing that intercropping tends to enhance both resource-use efficiency and overall system productivity. Gao et al. (2023), for instance, used MANOVA in daylily-based intercropping trials to demonstrate multivariate gains in yield and associated traits. While Huang et al. (2022) show that tea-legume intercropping improves nitrogen availability, soil organic matter and microbial activity, thereby enhancing overall resource-use efficiency. By biologically fixing nitrogen and improving nutrient cycling, legumes reduce fertilizer dependence and allow tea plants to use soil resources more effectively.
 
Economics
 
Economically, cotton-redgram intercropping proved clearly superior to sole cotton. The mean cost of cultivation under intercropping was lower (₹ 50,284 ha^-1) than under sole cropping (₹ 52,262 ha^-1), yet the system generated higher gross returns (₹ 117,069 ha^-1) and net returns (₹ 66,434 ha^-1) (Table 3 and 4). As a result, the Benefit-cost ratio (BCR) under intercropping was 2.30, compared with 1.71 for sole cotton, indicating that each rupee invested in the intercropped system yielded substantially higher returns. The additional grain yield from redgram, more efficient use of shared inputs and operations and the spreading of risk across two crops all contributed to this economic advantage.




       
These outcomes also illustrate why intercropping should be evaluated using multi-trait criteria that combine yield and economic metrics rather than yield alone. Recent reviews on new statistical approaches for intercropping experiments have encouraged the use of multivariate tools such as MANOVA and discriminant analysis to capture integrated yield-income responses. In line with this perspective, our results corroborate earlier reports by Pandagale et al. (2019), which showed that cotton-legume intercropping not only increases profitability but also enhances economic resilience compared with mono-cropped cotton.
 
Multivariate statistical validation
 
Multivariate diagnostics and MANOVA
 
Before applying multivariate models, we assessed whether the data met key statistical assumptions. Royston’s multivariate normality test was non-significant (p = 0.473 > 0.05), indicating that the joint distribution of the response variables did not deviate significantly from multivariate normality and was suitable for both MANOVA and LDA. A similar diagnostic strategy was followed in the Brazilian study “Selection of Indicators to Discriminate Soil Tillage Systems” (Kazmierczak et al., 2020) where Shapiro–Wilk and Royston’s tests were applied prior to discriminant analysis, underscoring the importance of careful assumption checking in agronomic datasets.
       
In our study, MANOVA results (Table 5) a highly significant effect of cropping system (Wilks’ Λ = 0.2693), with the model explaining 73% of the total multivariate variance. Among the response variables, yield was the dominant contributor (65.9%, p<0.01), followed by cultivation cost, net returns and gross returns, while BCR made a comparatively minor and non-significant contribution. The Univariate ANOVA also results represented that the treatments clearly influenced NR, CC, GR and especially yield, as all of these traits show strong statistical significance. However, BCR did not respond much to the treatments, indicating that it remained relatively unaffected compared to the other traits.


 
Discriminant analysis and system separation
 
Linear Discriminant Analysis (LDA) provided additional insight into how well the two cropping systems could be separated using the measured variables. The first discriminant function was:
 

LD₁ = -0.00003499 × NR + 1.994 × BCR + 0.000392 × CC - 0.0000306 × GR - 0.00639 × Yield

 
The large negative coefficient for yield indicates that higher yields strongly pulled LD scores toward the intercropping system, highlighting yield as the main driver favouring that system. In contrast, positive coefficients for BCR and cultivation cost shifted scores toward sole cropping, indicating that these traits, when considered jointly with others, tended to characterize the mono-cropped fields. Mean LD scores clearly distinguished the two groups (Intercropping = -1.3; Sole = +0.64 showed Fig 1) and the model achieved 100% correct classification showed in Fig 2, demonstrating very strong multivariate separation between the systems in different representation.




       
Comparable discriminant or Linear Discriminant Analysis Effect Size (LEfSe) type analyses have been used in agronomic and ecological studies to differentiate treatments or management systems (Gao et al., 2023) and our findings are in line with these reports. In all cases, discriminant analysis helps reveal which traits or combinations of traits which most strongly contribute to group separation.
 
Integrated interpretation and implications
 
Overall, the combination of field performance and multivariate statistics makes a strong case for cotton-red gram intercropping as a superior alternative to sole cotton. The system delivers higher yields, better economic returns and more efficient use of land and inputs, while also satisfying key statistical model assumptions and showing clear multivariate separation from mono-cropping. These features echo earlier work on the multivariate stability and discriminant potential of diversified cropping systems (Gao et al., 2023).
       
The simultaneous improvements in yield, reduced cultivation costs and complementary use of light, water and nutrients position intercropping as both economically attractive and ecologically sound. This aligns with foundational intercropping concepts proposed by Raseduzzaman and Jensen (2017), who emphasized the potential of mixed cropping to increase overall system efficiency. In the present context, cotton-redgram intercropping, especially when supported by robust AAS-based management, emerges as a practical pathway towards sustainable intensification under variable and changing climatic conditions.

Indian meteorological department providing weather forecast and advisories through district agromet units. The following and practicing of agromet advisory service is beneficial for farming community to take necessary management practices in field according to weather.
       
According to present investigation, concluded that AAS followed farmers achieved maximum yields and productivity through practicing redgram as intercrop in cotton and weather-based need full actions in intercropping. Telangana offers significant potential for cotton-redgram intercropping, owing to the extensive acreage under cotton cultivation across the region. The synergetic impact of pulse crop increases the soil health by increasing organic matter and microbial activity.

On behalf of all the authors of the manuscript we declare that there is no conflict of interest.