National trend analysis using the Mann-Kendall test

The Mann-Kendall test was used to analyze the trend in the national data. It is a non-parametric statistical test used to detect trends in time series data. It tests the null hypothesis that there is no trend in the data against the alternative hypothesis that there is a monotonic trend (either increasing or decreasing). The test statistic is based on the differences between pairs of observations in the time series. If there is a trend, then the signs of the differences will tend to be either all positive or all negative. The Mann-Kendall test statistic is the sum of the signs of the differences, and its distribution under the null hypothesis is known. A large positive or negative value of the test statistic indicates a strong trend in the positive or negative direction, respectively. The p-value of the Mann-Kendall test indicates the strength of evidence against the null hypothesis. A small p-value (e.g., less than 0.05) suggests strong evidence against the null hypothesis and in favor of the alternative hypothesis of a monotonic trend.

The test statistic for the Mann-Kendall test can be calculated using the following formula: (1)

Where x_i is the i^th observation, x_j is the j^th observation, sgn() is the sign function, which returns -1 if (x_i-x_j) is negative, 0 if (x_i-x_j) is zero, and 1 if (x_i-x_j) is positive. The test statistic S follows a normal distribution with mean 0 and variance: (2)

Where n is the number of observations, t_j is the number of tied groups in the j^th group. The standardized test statistic Z can be calculated as: (3)

Where 1 is the expected value of S under the null hypothesis of no trend. The p-value can be calculated from the standard normal distribution, using the Z statistic and the appropriate tail(s) of the distribution.

Decadal analysis using CAGR and CV

For the national and state level trend analysis, we divided the data into 5 decades D1 (1970–1980), D2 (1980–1990), D3 (1990–2000), D4(2000–2010), D5(2010–2020) and calculated the Compound Annual Growth Rate (CAGR) and Coefficient of Variation (CV). CAGR is the annualized average rate of growth between two given years, assuming growth takes place at an exponentially compounded rate. The formula for CAGR is given below: (4) where v_final is the last year of the decade, v_initial is the beginning year of the decade, t is the number of years which is 10 for decadal analysis.

CV is the measure of the relative variability or dispersion of a dataset. It is calculated by dividing the standard deviation of the dataset by the mean. A low CV indicates that the data points are relatively close to the mean, while a high CV indicates that the data points are more spread out around the mean. The formula for CV is: (5)

Spatial autocorrelation using Global Moran’s I

Spatial autocorrelation has been widely used to explore the spatial pattern of variables for various applications. It measures how one object is similar to others surrounding it. There are various indices used to measure spatial autocorrelation at different levels, depending on the application. Spatial autocorrelation analysis can identify spatial clusters (positive autocorrelation) and spatial outliers (negative autocorrelation) of a regionalized variable [19]. Global Moran’s I is a measure of the overall clustering of the spatial data. Values of I usually range from −1 to +1 with values tending towards +1 indicating positive spatial auto-correlation and vice versa. The z-score measures the degree to which the observed Moran’s I coefficient differs from what would be expected under the null hypothesis of no spatial autocorrelation. A high absolute value of the z-score indicates that the observed Moran’s I coefficient is significantly different from the expected value, suggesting the presence of spatial autocorrelation. (6) where I is the Moran’s I statistic, x_i is the value of variable at location i, x_j is the value of a variable at location j, and w_ij is the weight that determines the relationship between i and j.

Hot spot analysis (Getis-Ord Gi*)

Hotspot analysis was used to identify hotspots of soybean production and how they have shifted over the study duration. The Getis-Ord Gi* statistic can be used to perform hotspot and cold-spot analysis to identify statistically significant clusters of high or low values in a dataset. It is based on the concept of spatial autocorrelation, which refers to the tendency of nearby values to be more similar than those that are further apart. The test measures the degree of spatial association between each feature in a dataset and its neighboring features, and then assigns a score to each feature based on its level of association. G* is high where the sum of values within a neighborhood of a given distance or arrangement is high relative to the global average and negative where the sum of values within a neighborhood are small relative to the global average and approaches 0 at intermediate values [20, 21]. The neighborhood function determines which features are considered neighbors of a specific feature based on their spatial proximity or other defined criteria. The function assigns weights to these neighboring features to measure the degree of association or interaction between them. There are different types of neighborhood functions, depending on the application and analytical requirements: distance-based neighborhood, contiguity-based neighborhood, and kernel-based neighborhood. The appropriate conceptualization of spatial relationships and the best distance band or threshold distance value depend on the specific context and the nature of the data being analyzed [22].

Trend analysis at national level

Soybean area, production and yield are plotted annually at the national scale for India from the 1970 to 2022 (Fig 1) which depicts an overall upward trend in area, production and yield, however, the yield and production can be seen to plateau in recent years. To further analyze the trends, we performed the Mann-Kendall test and the results suggest that there is a significant upward trend in soybean area, production, and yield over time. For Area, the Mann-Kendall test yielded a z-score of 10.32 and a very low p-value of 2.2e-16, indicating a strong and statistically significant upward trend in the soybean area over time. The large value of S (1.38e+03) and tau (9.67e-01) further supports the presence of a significant positive trend. Similarly, for production, the Mann-Kendall test yielded a z-score of 9.53 and a very low p-value of 2.2e-16, indicating a strong and statistically significant upward trend in soybean production over time. The large value of S (1.27e+03) and tau (8.93e-01) supports the presence of a significant positive trend. For Yield, the Mann-Kendall test yielded a z-score of 4.81 and a p-value of 1.494e-06, indicating a strong and statistically significant upward trend in soybean yield over time, although it appears to be stagnant over the last decade. The sample estimates show a value of S (6.46e+02) and tau (4.51e-01), indicating a less pronounced but still positive trend.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Growth trajectory of soybean at the national level.

Area (Hectares), Production (Tonnes) and Yield (Tonnes/Hectare) at the national level for India from 1970 to 2022.

https://doi.org/10.1371/journal.pone.0292005.g001

CAGR and CV in Table 1 and Fig 1 show that the soy harvested area has consistently increased over time, while production has had some significant dips with the highest production achieved in 2012. Average yield increased steadily but flattened out starting in 2012. In the first decade, 1970–80 the total national soy area increased at a CAGR of 34% (from 0.029 MHa to 0.54MHa), while production increased from a CAGR of 38.6% (0.016 MT to 0.42 MT). This was the period of greatest growth in soybean. For each of the subsequent decades the rate of growth in area and production approximately halved. In the last decade (D5), production went into decline (negative growth). The rate of growth in yield decreased steadily with each decade (D1 3.4% to D4 1.7%). During 2010 to 2020, the national growth rate in yield went negative (D5-2.4%) bringing down overall production despite of increase in area under the crop.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Comparison of soybean area, production and yield using the CAGR (annualized average rate of growth rate between two given years) and CV (standard deviation over mean during each decade) for 5 decades 1970 to 2020.

The table has been color-coded with dark green being the highest CAGR of 34% and yellow as lower with a CAGR towards zero or negative CAGR.

https://doi.org/10.1371/journal.pone.0292005.t001

Overall, the area, production, and yield have improved in the past five decades, however, the rate of improvement has slowed in each of the five decades consecutively. For the last decade, the yield and production have shown negative CAGR, while the area increased only by 2% CAGR because of the poor yield during last few years at the national level. There is considerable variation in each decade especially for the first decade as shown by the Coefficient of Variance (CV) statistics. In the first decade (1970–80), total of 1786% of increase in area (CAGR 34%) with CV of 95% and total 2469% of increase in production (CAGR 39%) with CV of 100% was reported (S1–S3 Tables in S2 File). For subsequent decades, the CV for the national area was 9%, and 16% for production, which means the area was increasing slowly but production was not stable in the last decade. For the same period, there was negative growth rate in yield (-2.45%) with a CV of 20%, showing the turbulence in yield in recent decades.

Interstate comparison of soybean

Data were plotted data for the five primary soy-producing states (Madhya Pradesh, Maharashtra, Rajasthan, Karnataka, and Gujarat), which together currently account for more than 98% of national soybean production (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Comparison of soybean area, production and yield at state level for major producing states.

(Madhya Pradesh, Maharashtra, Rajasthan, Karnataka, and Gujarat.) In addition, the yield graph shows the average national yield compared to the state yield.

https://doi.org/10.1371/journal.pone.0292005.g002

Though soybean was grown traditionally in India as black soybean [12], the yellow soybean was introduced in Madhya Pradesh in the 1970s. The cultivation of soy was subsequently adopted by farmers of Gujarat and Rajasthan around 1980, and Maharashtra around 1990. Although being adopted in Rajasthan and Gujarat earlier than in Maharashtra, the area under soy increased more rapidly in Maharashtra. The rate of growth in Maharashtra began to accelerate around 2000. Madhya Pradesh contributed around 97% of India’s soybean production around 1980 but in 2020, its contribution declined to 52% in terms of sown area and 44% in production. Over the same period, Maharashtra’s contribution increased from 1% to 45% in 2020. Currently Rajasthan contributes 8% and Karnataka 3% of production respectively. Clearly, the crop has drastically expanded in other parts of the country in past two decades and Madhya Pradesh can no longer be called the soybean bowl of India.

Fig 3 shows the decadal analysis of area, production, and yield for 5 major producing states of India. Madhya Pradesh alone was the major producer of soybean contributing more than 95% to the national production but recently Maharashtra is contributing more, despite Madhya Pradesh having a larger area under the crop. This is because of the poor yield in Madhya Pradesh in recent years due to various factors including extreme weather conditions. The overall negative growth of production for the last decade could be attributed to the poor yield due to major flooding and waterlogging during the maturity stage of soybean in 2019 and extensive drought in 2020 in Madhya Pradesh. As a result, we see that although yield declined in Madhya Pradesh, in Maharashtra the productivity continued at 1.3 T/Ha in c.2020. Madhya Pradesh and Rajasthan have had consistent yields over the years averaging around 0.9 T/Ha and 0.7 T/Ha respectively, with some fluctuations but no major increase or decrease. Gujarat has shown a significant increase in yield from c.1980 (0.4T/ha) to c.2020 (1.3 T/Ha).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Soybean area, production and yield decadal comparison at the state level.

(a) Contribution to national soybean acreage by major states in c1980 inner circle, c2000 middle circle, c2020 outer circle. (b) National soybean production contribution by each state; c1980 inner circle, c2000 middle circle, c2020 outer circle. (c) soybean yield for years c.1980, c2000, c.2020.

https://doi.org/10.1371/journal.pone.0292005.g003

Spatial autocorrelation and trend analysis at the district level

Fig 4 shows the spatial distribution of area, production, and yield of soybean for three decades along with the Global Moran’s I values and z scores significance level (*** indicates p-value < 0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Soybean area, production and yield distribution in India for c. 2000, c. 2010 and c. 2020 along with the Moran’s I value and corresponding z scores.

*** represents p value < 0.001.

https://doi.org/10.1371/journal.pone.0292005.g004

All three variables (Area, Yield, and Production) show positive spatial autocorrelation in all three years, as indicated by the positive values of Moran’s I coefficient. The z-scores are also very high for all variables and years, with p value less than 0.001 indicating that the observed Moran’s I values are significantly different from what would be expected under the null hypothesis of no spatial autocorrelation. Overall, this suggests that there is a spatial pattern in the distribution of the variables, and nearby observations tend to have similar values.

Yield distribution for 20 major producing districts

To further analyze yield distribution, we compared the area and yield using a scatter plot for all districts and further analyzed the major 20 soybean producing districts in detail. Fig 5 shows a scatter plot between yield and area for all districts cultivating soybean for c.2020. We collected data from all the districts in India that cultivated soybean in the recent years i.e. circa 2020. Using these data, we created scatter plots to visualize the relationship between yield and acreage for each district. Furthermore, from this dataset, we identified the top 20 soybean producing districts for c. 2020 and separately plotted their data in a zoomed version within the same Fig 5. This allowed us to gain more detailed insights into the behavior of the major producers concerning their yield and overall acreage and production. Those 20 districts contribute around 57% of total production in year c 2020. The scatter plot of yield vs. acreage for c.2020 shows that the districts with higher area have lower average yields and the districts reporting highest yield mostly in the northeastern hill region have the lowest area under the soybean crop. Lower yield averages for districts with large areas under soybean crop may be due to the various biotic (pests and diseases) and abiotic factors causing yield losses resulting in a low average. Also, it is noteworthy that there were unusually heavy rainfall in 2019, while in 2000 there was drought in some part of Madhya Pradesh resulting in average lower yield and production for c. 2020 [23]. The impact of weather variability on yield is beyond the scope of this paper and is a topic for future research.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Scatter plot of area vs yield for all districts for c2020 and the zoomed-in window shows the scatter plot for the 20 major producing districts.

https://doi.org/10.1371/journal.pone.0292005.g005

For c. 2020, the highest yield was observed in the northeastern districts of India, while these are not major producers as the area under soybean is considerably lower as compared to central India. The Southwest Garo Hills and West Garo Hills Districts of Meghalaya show the highest yields with 5.9 T/ha and 5.8 T/ha respectively, although the area contribution from these districts is very low. This comparatively high yield with small area could be in part because of the high variability in yield in major producing districts including farms with poor yields in many fields, which would average out to a lower yield for the entire district. Another reason for these yield gaps could be the varieties used or the climatic or soil conditions. The most likely explanation could be that the districts having comparatively less area started cultivation very recently on virgin soils. Further, less area facilitates the farmers to adopt the recommended agronomic practices at the right time and the crop can be managed for insect-pests and diseases. Similarly, the soybean areas of northeastern states have more soil organic carbon content and are thus able to give higher yields, as compared to the high intensity, large area soybean growing states/districts which are comparatively more prone to biotic and abiotic factors. In either case, such yield gaps are an important topic for further research.

Some exceptions can be found, where districts have both good yield and high cultivation area, for example the Satara District in Maharashtra has an area of 0.07 MHa with a yield of 1.7 T/ha, with a production of 0.11 MT, making it one of the major producer districts, despite having less area than some other districts. It is interesting to note that there is no single district from Madhya Pradesh in the top 70 high yield soybean districts. The maximum yield in Madhya Pradesh is observed in Indore with an area of 0.2 MHa and a yield of 1.3 T/Ha. The Barwani District (MP) had the poorest yield of 0.3 T/ha for c. 2000 and 2010, but in 2020 improved to 1.2 T/Ha.

A zoomed-in plot in Fig 5 shows the scatter plot for 20 major producing districts for c.2020. These 20 districts include 9 districts from Maharashtra state and 11 districts for Madhya Pradesh, contributing to 57% of national soybean production in c. 2020. The blue represents the districts in Madhya Pradesh state and red represent the districts in Maharashtra state. The lowest soybean yield is observed in Ujjain (Madhya Pradesh) 0.9 T/Ha while it has the highest area among all districts of India. The highest yield was observed for Hingoli District (Maharashtra) 1.5 T/ ha with area 0.2 MHa. For most of the top producing districts the yield is around 1.1 T/ha. For Madhya Pradesh State, Indore and Dhar districts reported the highest yield.

Hotspot analysis with respect to agroclimatic zones and spatial mean

The hotspot maps generated using production statistics (Fig 6) shows that in c. 2000 the soybean was concentrated in the Malwa region of Madhya Pradesh and it was major contributor to national production but in c. 2020 the soybean hotspot had spread to the western plateau hill region and covered some of the districts of Maharashtra and Rajasthan. The Malwa region of central India is no longer the major hotspot of soybean production as it was in the 1970’s, as the entire central belt of Madhya Pradesh and Maharashtra combined can now be called ‘the soybean bowl’. There has also been a trend in soy expansion towards western India, which includes the eastern districts of Gujarat and Rajasthan. The cold spots where low values cluster (99% confidence) are observed in north-eastern districts which means this region has the lowest producing districts as compared to the major producers. As of 2020, 42 districts are hotspots of soybean production, up from 29 in 2000.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Hotspots and coldspots of soybean production in c. 2010, c. 2000 and c. 2020.

https://doi.org/10.1371/journal.pone.0292005.g006

Fig 7 shows that the hotspot districts are mainly concentrated in two agroclimatic zones on India viz. Central Plateau and Hills Region and Western Plateau and Hills as shown in the Fig 7(A). Agroclimatic zones can help identify areas that may be susceptible to specific weather-related hazards such as droughts, floods or frost, allowing farmers to take steps to mitigate those risks. India is divided broadly into fifteen agroclimatic zones based on soil, water, temperature conditions and agricultural practices. Soybean is mainly grown in three agroclimatic zones of India viz.

1. Central Plateau and Hills Region (Aravali Malwa upland) (50 cm-100 cm)
2. Western Plateau and hills (Maharashtra plateau region) (Rainfall 25-75cm)
3. Eastern plateau and Hills Region (Southeastern Plateau region) (Rainfall 80-150cm)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Hotspot expansion in three decades and distribution of soybean in agro-climatic zones.

(a) Hotspot gradient diagram depicting the hotspot expansion over three decades; Yellow represents the soybean production hotspot areas for 2 consecutive decades (2000 to 2020); green and yellow together represents the soybean production hotspot districts for the last decade (2010 to 2020), and blue represents the districts that recently became hotspots i.e. c.2020. The red polygon represents the Indian state boundary while the polygons in black color represents the agroclimatic zones. The hotspots identified with 99% confidence level using Getis Ord G* statistics in ArcGIS. (b) The bar graph representing the area and production of soybean for various agroclimatic zones of India for c2000 and c.2020. The zones not producing soybean at all are not included in this graph.

https://doi.org/10.1371/journal.pone.0292005.g007

Overall average annual rainfall from 25 cm-150 cm and soil types in these three zones are mixed red, yellow and black. From Fig 7(B) it can be seen that in the two decades, the crop has expanded to the Gujarat Plains and Hills, which has rainfall ranging from 50-100cm. In the Western Plateau and Hills, soybean production has tripled from 1.6 MT to 5.1 MT over the past two decades, while it has increased only by 35% in Central Plateau and Hills region from 3.9 MT to 5.2 MT. We overlaid the hotpots for 3 circa years with the agroclimatic zones (Fig 7) and found that hotspots of soybean are concentrated in two agroclimatic zones only i.e. Central Plateau and Hills Region (Aravali Malwa upland) and Western Plateau and hills (Maharashtra plateau region). For c. 2020 and the average yield of Western Plateau and Hills is 1.1 T/Ha and for the Central Plateau and Hills region is 0.8 T/Ha for c.2020, indicating that the former provided more favorable conditions for soybean cultivation during recent years.

The spatial mean of the production was calculated using the mean distance and showed that the spatial mean of soybean production has moved towards south by approximately 91 km and west by 24 km. In 2000, the center of overall hotspot of soybean was the Dewas District, while at present in c. 2020, the mean has moved 93 km to the south to Khandwa (East Nimar) District of Madhya Pradesh, which started the cultivation of soybean two decades ago.

Drivers of trends and future projections

To better understand future prospects, in this section we address the drivers of soybean consumption in India and its future projections. We used long term population, price, import, export, domestic production and GDP per capita to examine the relation with the soybean consumption in India. Soybean is mainly consumed as edible oil and soy meal in India and hence we have performed regression and correlation analysis separately for soybean oil and meal. Domestic soybean oil and meal dynamics are influenced by multiple factors, including dietary habits and preferences, economic factors, and government policies like subsidies. Unfortunately, these data are not currently available at the national scale and were therefore not considered in this study. However, in future research, incorporating such ancillary variables could provide valuable insights into the drivers and future projections of oil and meal consumption. To achieve this, conducting a comprehensive national survey with robust sampling techniques will be essential for obtaining the necessary data. For this study we performed regression and correlation analysis with all the available relevant variables at national scale. Table 2(A) and 2(B) explains the correlation of soy oil and soy meal consumption with GDP per capita, import, export, production, population and commodity price while Table 2(C) and 2(D) shows the regression results of these five variables for soybean oil and meal consumption respectively. (The summary of regression results can be found in S5 to S7 Tables in S2 File).

Soy oil domestic consumption is highly positively correlated with population (0.904) and GDP per capita (0.962), meaning that as population and GDP per capita increase, the domestic consumption of soy oil also tends to increase. Soy oil price has a moderate positive correlation with soy oil Domestic consumption (0.600).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Potential drivers of soybean expansion in India.

(a) Pearson correlation coefficient matrix for soybean oil consumption, population, GDP per capita, import, export, and soybean oil price. (b) Results of a multiple linear regression analysis with soy oil as the dependent variable and population, GDP per capita, import, export, and soy oil price as independent variables. The table provides the coefficients, standard errors, t-statistics, and p-values for each independent variable, as well as the intercept. (c) Pearson correlation coefficient matrix for soybean meal consumption, population, GDP per capita, import, export, and soybean meal price. (d) Results of a multiple linear regression analysis with soybean meal as the dependent variable and population, GDP per capita, import, export and soy meal price as independent variables. The table provides the coefficients, standard errors, t-statistics, and p-values for each independent variable, as well as the intercept.

https://doi.org/10.1371/journal.pone.0292005.t002

The correlation analysis (Table 2(A)) reveals significant positive relationships between soybean oil consumption and domestic oil production, oil import, population, and GDP per capita. However, no significant correlation is found between soybean oil consumption and oil export or soy oil price, likely due to the influence of external factors such as market dynamics and international trade policies. Additionally, the dataset’s limited scope may not capture the full complexity of these interactions. Hence, to gain a deeper understanding, we conducted regression analysis (Table 2(B), and the results indicate that soy oil production, oil import, population, and GDP per capita are significant predictors of soybean oil consumption. A high F-statistic with a very low p-value indicates that the model is statistically significant and the independent variables together are strong predictors of soybean oil consumption. While it is intuitive that demand would lead to more import and domestic production of soybean oil, it’s interesting to notice that population and GDP per capita significantly influence oil consumption, indicating improved affordability for soybean oil among a larger population may lead to more consumption and hence demand.

There is no projection available for GDP per capita, price, import and export of soybean oil, but according to the World Bank, the projected population of India in 2050 is approximately 1.67 billion people. Hence, we performed the regression for soybean oil domestic consumption with population only and results are shown in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Regression matrix showing relation between soy oil consumption and population.

https://doi.org/10.1371/journal.pone.0292005.t003

The regression statistics show that there is a strong positive correlation between domestic soybean oil consumption and population. The coefficient of determination (R-squared) is 0.8179, indicating that 81.8% of the variation in soybean oil consumption can be explained by the variation in population. The adjusted R-squared value of 0.8143 suggests that the model is a good fit for the data. The regression equation for this model is:

Based on the above regression equation and assuming a projected population of 1.67 billion, the estimated soybean oil consumption in India in 2050 would be around 5.72 MT compared to the current consumption which is around 5 MT. It should be noted that this is just an estimate and there are many other factors that could impact soybean oil consumption in the future. At present, India imports around 3MT of soybean oil worth approximately 5 billion U.S dollars.

Correlation and regression analysis for soy meal consumption in India (Table 2(C) and 2(D)), suggests strong positive relationships with meal production, population, and GDP per capita, indicating that these factors play significant roles in driving domestic consumption. Additionally, there is a moderate positive correlation with meal import and meal price, suggesting their potential influence. However, the correlation with meal export is relatively weak, implying limited impact on domestic consumption. Regression analysis for soy meal consumption shows that soy meal production and imports exhibit a strong association with soy meal consumption, which is intuitive as heightened demand corresponds to increased domestic consumption and imports, ensuring a sufficient supply to meet growing consumer needs. The regression coefficient for GDP Per capita is positive (0.658) and highly statistically significant (p-value: 6.71E-05), indicating that as GDP per capita increases, soybean meal consumption also increases, implying greater purchasing power and affordability, leading to increased consumption of soybean meal as a protein-rich food source. While it would be valuable to forecast soybean meal consumption in India, our limitations lie in the absence of future projections for variables other than population. Furthermore, the statistical significance of population as a predictor is not sufficiently strong to base estimates solely on population data. However, conducting future analyses with more comprehensive data, including subnational level long-term consumption data, dietary preferences, population trends, and subnational trade data, could lead to more reliable predictions of soybean meal as well as soybean oil consumption in India.

Indian soybean production in an international context

Placing Indian soybean production and demand in an international context is crucial. It can help identify global trends in soybean production and consumption, which can aid in predicting future demand and supply dynamics. Soybean contributes around 30% of the worlds current edible oil need. Major producers of soybean oil are China, USA, Brazil and Argentina, with India ranked 5th in production (Fig 8(A)). Major importers of soybean oil include China, India, Algeria, Bangladesh and Peru (Fig 8(B)). India is the world’s largest importer of soybean oil, surpassing China in 2013–14 (Fig 8(B)) despite being one of the major edible oil producers. This can mainly be attributed to the low production of soybean vis-à-vis the demand in India to feed its population (Fig 8(A)). The contribution of India in terms of the world’s soybean area is 10%, but the contribution to total world soybean grain is only 4%, indicating lower levels of productivity of the crop in India (<1 T/ha), as compared to other countries (world average 2 T/ha). Hypothetically, if the average soy yield for India was greater than 3 T/ha, and keeping other countries’ area and yield constant, India would become the 3rd largest producer, surpassing Argentina and China. Furthermore, the productivity of soybean crop in India is low, as compared to the other countries e.g., Brazil, USA, China and Argentina (Fig 8(A)) majorly because of its smaller duration to maturity (90–120 days) as compared to Brazil, Argentina and USA (140–160 days) and relatively unfavorable climatic and soil conditions. India needs to invest heavily in research and development of soybean varieties that are well-suited to its local growing conditions to be able to meet its domestic demands without increasingly depending on imports.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Comparison with international top producers and consumers of soybean.

(a) Area, production and yield for major 10 soybean producers of the world. (b) Major soybean oil importing countries of the world.

https://doi.org/10.1371/journal.pone.0292005.g008