Adaptation to climate change and variability: a case of direct seeded rice in Andhra Pradesh, India

Food production in India is largely dependent on irrigated agriculture. Increase in population and urbanisation, in addition to climate change and variability, will have a significant impact on the availability of water resources. In arid and semi-arid regions of India, any shortfall in water supply due to climate change will enhance competition for water use to a wide range of economic, social and environmental applications. The population growth with the improved living conditions also increases the demand for food production and thereby increases the demand for water multiple-fold in the river basins (UNESCO-WWAP 2009). The major proportion of water available in Indian river basins is used for irrigation of rice and thus major challenges to water management are posed, especially during unfavourable weather or climatic conditions. Around 95% of the cultivated area under rice is irrigated, and requires water of about 12,000 m³ to 25,000 m³ per ha for one season depending on the soil texture, structure and profile conditions (Reddy & Reddi 1995). Of all the major cereal crops grown in India, rice constitutes 24% of the total food grains produced, contributing almost 96 million tonnes. The popular method of rice cultivation so far has been transplanting in ponded water that ensures steady yields (Chen et al. 2009). This may not be the case in the future with growing climate uncertainty, insufficient irrigation water in the canal commands, increased labour requirement and increased emissions. The temperature rise will also negatively impact rice yields as they are being grown to their threshold (Kelkar & Bhadwal 2007). The projected climate scenarios for mid (2021–2050) and end centuries (2071–2100) in the Krishna River Basin of India also show a change in the future water availability, which impacts rice production (Palanisami et al. 2011).

Labour requirement is also comparatively higher in transplanted rice cultivation. About 25 to 37 labourers (man-days) per hectare are required for rice transplanting, 50 for weeding/hectare and about 25 man-days/per hectare for harvesting (Prakash et al. 2013; Kakumanu et al. 2017). The availability of labour is becoming problematic in most Indian states, as rural people are migrating to nearby cities and towns for higher wages or being hired in the government social welfare programmes (e.g., the National Rural Employment Guarantee programme). Hence, the cost of hiring labour has significantly increased during recent years (Farooq et al. 2011). For example, in the lower Krishna River basin (Krishna, Guntur and Prakasam districts) of Andhra Pradesh state, costs for labour in rice cultivation accounted for 29% of the total cultivation costs during 2006–2007, and it increased to 49% in 2010–2011 (Kakumanu et al. 2017).

It is also a known fact that rice fields emit methane into the environment that contributes to global warming (Pathak et al. 2013). Emissions from flooded fields are higher than those from drained fields (Komiya et al. 2010). Methane is generated when organic matter decays in anaerobic conditions. Hou et al. (2000) reported that the factors affecting methane and nitrous oxide emissions are soil temperature and soil redox potential, net irradiance and organic matter content. The IPCC Fourth Assessment Report states that agriculture accounts for 50% of methane emissions, and 10–20% of these emissions come from rice production alone (Reiner & Aulakh 2000). Overall, South and East Asia are responsible for 82% of methane emissions from rice production. Annually, 4.5 million tonnes of methane are emitted from rice cultivation in India (Pepsico International 2009). Methane emissions from Indian rice fields are significant and therefore there is a need to shift to more climate resilient rice cropping systems that can also help in methane reduction (Lakshmanan et al. 2009).

Hence, to sustain rice production and ensure food security with these challenges, new adaptation strategies have to be developed and up-scaled. The delayed onset of monsoon, insufficient irrigation water at the tail ends of canal commands, increased labour requirement and costs are forcing farmers to adapt to different management practices like direct seeded rice (DSR) and alternate wetting and drying (AWD) for rice (Chapagain et al. 2011; Joshi et al. 2013; Mahajan et al. 2013; Li et al. 2014). However, the practices have not been adopted in large scale due to the lack of timely and sufficient scientific information and dissemination framework. Much of the scientific or technical information developed on DSR is at research station level (Yadav et al. 2007, 2011; Ali et al. 2014; Li et al. 2014) with modelling aspects (Liang et al. 2015; Palanisami et al. 2015). Hence, the present study analysed the performance of DSR in actual field conditions and documented the results along with the development of an upscaling strategy with policy prescriptions.

DSR can be practised through three methods, namely, (1) dry seeding, (2) wet seeding and (3) water seeding by avoiding the nursery and transplantation (Joshi et al. 2013). In the case of dry seeding, seeds are sown directly into dry soil at a depth of 2–3 cm, immediately after the pre-monsoon showers. The method is suitable for a rain-fed and irrigated environment with precise water control. In irrigated areas, the dry seeded fields can be converted into wet method based on the availability of water from canals. Irrigation can be provided 45–60 days after sowing and then managed as a wet system. In wet seeding, pre-germinated seeds are broadcast or sown in the puddled soils. It is most favourable to rain-fed lowlands and irrigated areas with good drainage facilities. The wet method is practised in Malaysia, Thailand, Vietnam, Philippines and Sri Lanka (Pandey & Velasco 2005). Drum seeded rice is recommended in India using the wet method, where seeds are sown in line on the puddled soil. In the water seeding method, seeds are broadcast in standing water (5–10 cm) in areas where red rice or weedy rice problem exists.

In the present study, the first dry seeded method was followed, which requires less water, labour and has lower cultivation costs with comparatively equal or slightly higher grain yields than traditional transplanting paddy rice method (Yadav et al. 2011; Li et al. 2014; Liang et al. 2015). The crop matures earlier than the transplanted paddy rice (Bhushan et al. 2007), but the weed growth is higher in DSR and requires the use of selective pre- and post-emergence herbicides to address the problem (Chauhan & Opeña 2012). Timely sowing of rice through the dry seeded method provides the possibility to take up a second crop (blackgram, greengram, etc.) on time during the year.

The study was conducted under Nagarjuna Sagar Project (NSP) area of the Krishna River basin falling in Guntur district of Andhra Pradesh (Figure 1). The NSP distributes water through right and left bank canals covering 0.47 and 0.42 million hectares, respectively. The right canal supplies water to Guntur and Prakasam districts, and the left canal to Nalgonda, Khamman and Krishna districts.

Guntur district (15°44′N, 80°55′E) was identified for the study covering the NSP right canal and Delta area in the district. The long-term mean annual rainfall in Guntur is 815 mm, where 70% falls between June and September from the southwest monsoon, and the remaining 30% from the northeast monsoon (October to December). Long-term mean annual air temperature is high in April and May (36 to 43 °C), and milder in August to October (17 to 35 °C). Weather data for the study areas were acquired from nearby meteorological stations located at the Regional Agricultural Research Station, Lam, in Guntur. The soils of the Guntur area are dominated by deep black cotton soils with a silty clay loam texture, and contain, on average, low organic carbon (0.55%), and pH range 8.1–9.3.

The DSR method was validated on farmers’ fields (n = 50) compared with the traditional transplantation method from Kharif 2010 to 2014 (the Indian cropping system is classified into Kharif and Rabi seasons. The Kharif season starts with the onset of monsoon and falls during the month of June/July to October/November, whereas Rabi season falls between November and March), in order to generate relevant biophysical observations and economic data. The two methods were sown simultaneously in the same farmers’ fields for comparison on a pilot basis. The key biophysical observations and cost of cultivation for both methods were collected from 2010 to 2014 and summarized in the following section (refer to Table 1). The villages covered for validating/piloting of DSR through action research are Jonnalagadda village (Guntur rural mandal), Dondapadu village (Narsaraopet madal) and Muppalla village and mandal of Guntur district. The data generated from the pilots were used for the dissemination of DSR practice benefits and up-scaled in other villages/regions of the district.

For analysis, data from 100 farmers were collected, of which 50 were adopting DSR and 50 practising the traditional transplantation method from the non-piloting sites/villages in the district under Krishna Western Delta during 2014. Data were collected from nine villages (Alapadu, Appikatla, Ganapavaram, Jammulamadugu, Lakshmipuram, Modukur, Mulpuru, Narsayapalem and Nidubrolu) randomly based on the village-wise adoption information collected from the Department of Agriculture, Guntur district in 2014. As the upscaling process of DSR adoption in the district was initiated in a successful way, data related to the input, output and perceptions of farmers in adopting DSR were collected. Water application was measured by using Replogle, Bos, Clemmens (RBC) flumes in the validation/piloting sites. An RBC flume of 50 litres per second capacity was designed to use in small field channels for more accurate flow measurements. A calibration chart was developed to correlate the depth of water flow above the crust (placed at the bottom of the flume with a height of 10 cm for free flow conditions) with flow of water in litres per second. Data on the number of irrigations and depth of irrigation applied were collected from surveyed farmers to estimate the water application.

Comparison of various input levels in the traditional transplantation method with DSR was done in the study. This provided cross-sectional data for the two methods in rice. To test the significant difference of input and output mean, paired t-test was used.

Difference of means for the two methods was tested for various inputs including seed (kg/ha), labour (man-days), fertilizer used (kg/ha), pesticide applied (kg/ha), total water used (m³) and yield (kg/ha). Effective rainfall from the block was also measured and added to the irrigation to make the total water application.

The per cent position of each rank was obtained by converting into scores by referring to the table given by Garrett & Woodworth (1971). The respondents were requested to rank the opinions/reasons relevant to them according to the degree of importance. The ranks given by each of the respondents were converted into scores. Then, for each reason, the scores of individual respondents were added together and divided by the total number of respondents. These mean scores for all the reasons were arranged in descending order and ranks were given. By this method, the accuracy in determining the preference was obtained.

In India, the majority of rice-growing farmers follow the traditional rice transplantation method with the onset of monsoon and then depend on the release of canal water for irrigation. With the delayed onset of monsoons and droughts, farmers face problems of water shortage leading to failure of crops sown, or in some cases, they delay sowing or transplanting. The tail-end farmers in the canal commands (especially in Krishna River basin) are increasingly adopting DSR due to better dissemination of available information, water scarcity and less labour availability. The validated field observations of DSR illustrate that the seed and sowing cost reduced by 68–77% when compared to the traditional transplantation method (Table 1). The variable cost was also significantly less with a range from INR 8,000 to 15,000 per hectare. This was mainly due to reduction in land preparation, which raises nursery and transplantation costs. Only two labour man-days were used, on average, in DSR, as compared to 25 man-days in the transplantation method. The water use recorded by RBC flumes from 2010 to 2014 showed that there is an overall reduction of 2,500 m³ (11–23%) in water application. The yield difference between DSR and the transplantation method also ranged from 1 to 6 qt/ha. The WUE, gross margins and cost–benefit ratio over the years also depicted the benefits of DSR (Table 1). The irrigation cost includes both water charges (Rs/ha) and operational cost for irrigation (labour). The operational cost has varied between the two methods and increased the cost in DSR due to the transport of water from nearby sources (ponds/lakes). This is due to the delayed release of canal water from the irrigation project and break in the rainfall trend. The scheduling of irrigation after 45 days of sowing has reduced the water use by 11–23%.

The data collected from surveyed farmers (n = 100) on DSR and traditional transplantation methods were analysed for comparison. The mean age of the sample collected for the study was 50 years with a range of 38 to 70 years. The education level ranged from primary to post-graduation with an average of nine years’ schooling. Farmers’ experience was relevant and had implications for adopting new technologies. In the current study, the average sample experience in farming was about 26 years.

Farmers in the sample had varied farm size ranging from marginal to large (0.2 to 32 hectares) with black clay soils. The results from the paired t-value (Table 2) illustrate that there is a significant reduction in the means of seed rate, labour and irrigation, reducing the cost of input application for the farmers in DSR. The labour requirement was reduced by 20%, and was found to be 45% in the case of Punjab (Pathak et al. 2013).

Fertilizer and pesticide application did not vary at a significant level with DSR adoption. However, increase in yields was noticed with the reduced number of irrigations during the tillering/vegetative phase. The total water application reduced from 2,000 to 2,500 m³ (Tables 1 and 2) in DSR due to the early sowings without puddling.

Production function fitted estimated the changes in rice production due to the adoption of DSR. Table 3 presents the production elasticities of rice. The yields of rice for DSR and traditional transplantation method were regressed against the predominant factors of production. The R² value of 63% indicates that the rice yield variation among the total of farmers was explained by the selected independent variables in the study area (Table 3). The production elasticities for labour, fertilizer and technology adaptation are positive but significant for labour and DSR adaptation. Negative sign observed for seed, pesticide and irrigation water use indicated the need for reduction in the use of the respective resources. The irrigation water can be reduced further by integrating the AWD manner instead of continuous wetting during the vegetative phase.

Farmers’ perceptions on DSR adoption in the study area were also analysed by Garrett ranking. Positive responses were given concerning the reduced costs by avoiding transplantation, increased tiller length, resistance to lodging, free tillering with a greater number of tillers, lower occurrence of pests and diseases due to less humidity, a deeper root system that absorbed nutrients effectively from deeper layers and lower water demand in the initial crop growth stages (Table 4). Response was also given on the increase of nutrient use efficiency assuming methane emissions are reduced due to the controlled irrigation and spacing in the DSR method. However, exact methane emissions were not measured due to the limitations of the study, but need to be addressed in future research studies.

Farmers also responded to the constraints in DSR adoption. It is understood that the soil leveling is important in DSR with controlled irrigation practice (Table 5). The weed problem is greater in the initial phase of the crop (Mahajan et al. 2013). Tillage serves only as a temporary means of weed control because the soil contains many ungerminated weed seeds. Ploughing may bury weed seeds at a depth that prevents germination but may also expose other, once deeply buried seeds to conditions favourable for germination (Stoskopf 1985; Kumar & Ladha, 2011).

Heavy rains immediately after seed sowing will wash out the seed from the field, therefore care has to be taken by considering forecast information. Farmers also responded on the dropping of panicles during manual labour harvest due to the increased number of tillers and panicles. Draining of water in the initial phase of the crop during heavy rains is also a problem in the poor drainage areas.

The constraints may seriously impact the agricultural technology development and extension. Hence, there is a need for an emphasis on institutional change and technology pathways that are more effective (Hall et al. 2003; Watts et al. 2003). In the last 20 years, there has been a change in the thinking on the nature of the agricultural technology development and promotion process (Rasheed & Hall 2002). In the development of a technology or modified practice adoption and dissemination, stakeholder participation in the process is very important. Active stakeholder participation from farmers to the policy level is needed for the success of any technology adoption. Within the existing institutional arrangements working towards specified goals of adoption is challenging. There is a need for more awareness and capacity building programmes targeted towards farmers, government agencies and policy-makers.

The stakeholder mobilization in the present study followed an integrated participatory framework (Figure 2). The key steps included were the identification of water management practices/interventions for validation followed by technology assessment and refinement in the farmers’ fields. The interventions were developed as technology packages of practices. Data recording procedures were developed to establish a common database for comparison and analysis. Capacity building programmes were also developed for farmers and the Department of Agriculture (DoA), and the Irrigation & Command Area Development (I&CAD) officials in the region. Dissemination materials were produced in local languages and coupled with stakeholder workshops. The institutionalization of the interventions and innovations and resource mobilization and management were achieved.

In the process, farmers, scientists, agricultural department, irrigation department and media were brought together which helped in building trust. The pros and cons of the practice were discussed in the workshops, as listed in Tables 1 and 2. Television programmes such as Annadatha (television programme broadcast for farmers) and daily newspapers were used for disseminating information about DSR to a wider audience. The Agricultural Technology Management Agency (ATMA), which is part of the Department of Agriculture also published the validated results for dissemination purposes. Most of the stakeholders involved in the upscaling and disseminating of the practice were not involved through a formal channel. Framing all the activities into the extension cycle, we can see a step-wise pattern of invention, innovation and institutionalization.

The area under the DSR system in 2010 was limited only to 100 ha, but it increased to 103,210 ha by 2015 in Guntur district due to the efforts made to upscale with technical backstopping (Table 6). The adopted area has increased to 64% of the actual rice sown area in the district.

The Department of Agriculture has also taken serious measures to promote DSR by conducting a large number of campaigns, group discussions and result demonstrations at village level. The timely supply of seed drills (253 seed drills supplied during 2011–2014) on a subsidy basis by the Department of Agriculture has also increased adoption in the district.

The impact of adaptation was seen three-fold, namely, additional benefits/revenue, savings on fuel consumption, and intensification of agricultural cropped area. The additional benefits generated by the adoption of DSR were estimated by considering the gross margin difference with the transplantation method and area adopted in the district. The benefits have varied over the years due to the climatic conditions and market price (Figure 3). Farmers/trainers experienced in DSR have been asked for help by other new farmers, paying contractual wages, which is generating additional income to them (case of Jonnalagadda village).

Land preparation and puddling of rice crops need machine/tractor power. The transplanting method requires about 72.25 litres of diesel/ha from land preparation to harvest of the crop. In the dry method of DSR, 47 L/ha is sufficient. DSR does not need puddling which helps in the saving of energy used. Considering the diesel savings of 25.16 L/ha, the total diesel savings were 2.5 million litres when extrapolated to the entire area under DSR (Figure 4). This is equivalent to the reduction of 6,788 metric tons of CO₂ emissions (2.62 kg of CO₂ per litre).

As shown in Table 7, the total methane emissions from traditional rice transplanted fields were estimated to be 315 kg CH₄ per hectare (Joshi et al. 2013). In contrast for DSR, the estimated methane emissions were 220 kg CH₄ per hectare during Kharif season. Referring to these values, adaptation of DSR in Guntur district has reduced about 9,785 tons of CH₄ emissions during 2015. DSR adaptation is not only saving costs, labour, and time but also reducing greenhouse gas, particularly methane that causes global warming. DSR is reducing the methane emissions by 30–38% in India (Joshi et al. 2013; Pathak et al. 2013). The reduced emissions of these gases helps in climate change adaptation and mitigation.

On the other hand, water application is also reduced with adoption of DSR. Agricultural intensification in the normal rainfall season can lead to additional benefits to the region/district. Adoption of DSR in the district has reduced the water demand by 480 Mm³. Interpreting it at project level, NSP provides 7,465 Mm³ water on average to a command of 0.89 Mha. Paddy, cotton, chillies and irrigated dry crops are grown in the command area (GoAP 2009). Paddy is grown in 0.26 Mha to 0.40 Mha depending on water availability. The transplantation of rice requires about 13,510 m³/ha and DSR 11,038 m³/ha (average of 2010–2014 data). Adapting to DSR under the command area by 50% of the minimum area (i.e., 0.13 Mha) can further increase the area by 29,000 ha. Palanisami et al. (2014) also analysed the optimal allocation of land and water for NSP by using multi-goal linear programming to maximize the paddy production, farmers’ income and minimize water use with technology adoption. The authors also concluded that DSR reduces water use (20%) without any variation in production. Adoption of DSR is expected to increase during water stress years at the mid- (2021–2050) and end-century (2071–2098) periods.

According to the study, stakeholder integration in upscaling the adaptation practices is needed. Dissemination of the scientific information through different actors and extension methods was greatly experienced in upscaling. The cluster level or system level adoption of the DSR in the canal command areas can help in regulating the water resources and reduce the scarcity.

DSR is efficient in terms of reducing climate change impacts, water use, improving yield (1–6 qt/ha) and better returns to the farmer. Land levelling, weed control and heavy rains during sowing are the major constraints in adoption of DSR. Nonetheless, adoption of DSR has increased from 0.04 to 64% of the rice area in the study district. As a result, additional returns of INR 1,663 million were achieved during 2015 by reducing fuel consumption (6,788 metric tons of CO₂ emissions) and methane emissions in the region. Farmers in Guntur district, especially in the tail-end of the canals are increasingly adopting DSR. This clearly shows farmers are adapting to uncertain monsoon conditions, shortage of labour, and irregular irrigation and water availability. Stakeholder participation in the study showed that it provided an opportunity to the members to contribute knowledge in their respective fields, work towards a common goal with clearly defined roles and responsibilities, articulating problems, finding solutions, exploiting the synergies of working in groups, and sharing the lessons learnt. DSR can be recommended in many parts of Andhra Pradesh and India suffering from delayed and uncertain monsoons and during reduced water release from irrigation projects. Promotion of the practice as an alternative to traditional rice at a wider scale requires more training, capacity building and awareness of farmers. Close cooperation between the scientific community, line departments and farmers also helps in upscaling the technology.