Alternate wetting and drying irrigation maintained rice yields despite half the irrigation volume, but is unlikely to be adopted by small hold

Introduction

Concerns are increasing about producing enough food for the growing global population due to unreliable water supplies and lower overall water availability.
Irrigation helped with yield gains during the Green Revolution but relied on unsustainable water withdrawals.
Rice production is a concern because it's a staple crop for over 3 billion people and requires water-intensive irrigation or high rainfall.
Alternate wetting and drying (AWD) irrigation can maintain rice yields with half the irrigation volume, but smallholder lowland rice farmers in Nepal are unlikely to adopt it.

Abstract

Alternate wetting and drying (AWD) irrigation can save water while maintaining rice yields, but adoption is limited in some countries.
Knowledge gaps include AWD's effect on early vegetative vigor, its relationship with yield, its effects on local cultivars' yield and water use efficiency, and the socio-economic factors influencing irrigation scheduling.
An on-farm field trial compared two local cultivars (Hardinath-1 and CH-45) under AWD and continuous flooding (CF) in Agyauli, Nepal, along with social research methods.
AWD plots received 57% less irrigation water than CF plots, but yields didn't significantly differ, indicating AWD could enhance crop water use efficiency.
In CH-45, there were no treatment differences in yield components, while in Hardinath-1, an 11% decrease in filled grain number was compensated by a 14% increase in effective tillers per hill.
Tillering and green fraction were significantly higher under AWD.
Most local farmers already use a local adaptation of AWD, but have few incentives to reduce overall water use under current water governance, making formal AWD adoption unlikely.

Limits to AWD Adoption in Nepal

Water-saving irrigation techniques have been developed to allow farmers to reduce water use and costs.
Alternate wetting and drying (AWD) irrigation, developed by IRRI in the 1990s, allows the soil to dry out partially before re-irrigating.
Most lowland rice varieties can tolerate a 30% reduction in total irrigation volume without significantly decreasing yield.
IRRI’s “safe AWD” recommendations involve drying the soil until water depth reaches 15 cm below the surface and re-irrigating to a standing water depth of around 5 cm.
Crops may be continuously flooded during and after flowering or AWD irrigation may be continued through this period.
AWD significantly reduced water use while maintaining or increasing yields compared to continuous flooding (CF) in field experiments and on-farm trials.
AWD has been linked to improved root growth, altered plant hormone signaling, and enhanced grain filling rate.
Formal adoption of AWD by farmers in many rice-producing regions has been limited due to varying effects on yield based on soils, climates, cultivars, and management practices.
AWD may be unsuitable in sandy soils or heavy clays with shallow water tables.
Socio-economic, cultural, and political reasons may also contribute to farmers' reluctance to adopt new irrigation technologies.
Farmers' rationales for current irrigation systems need consideration, especially for smallholders in less developed countries where land and labor may be more limiting than water or yield.
A triangulation of scientific and social scientific methods was used to determine reasons for limited AWD adoption in Nepal.
An on-farm field trial investigated local perceptions of AWD and the physiological effects of AWD on early plant growth and yields of two locally grown rice cultivars.
It was hypothesized that AWD would allow significant water savings without decreasing crop yields.

Materials and Methods

Four replications of two irrigation treatments (CF and AWD) and two cultivars (CH-45 and Hardinath-1) were laid out in a randomized arrangement of 4 m x 4 m square plots.
The field was prepared using a buffalo-pulled plow and puddler, and farmyard manure and 46 kg \ ha^{-1} each of diammonium phosphate and urea were applied before transplanting.
Transplanting was carried out using local planting densities and spacings (mean number of seedlings per hill = 5, mean spacing between hills = 160 mm).
The crop was top-dressed with a further 60 kg \ ha^{-1} urea and a stemborer infestation treated with chemical insecticide at 36 days after transplanting.
Heading occurred in CH-45 and Hardinath-1 from 32 and 51 days after transplanting, respectively, and both cultivars were harvested on Day 88.
Mean maximum and minimum temperatures were 37°C and 26°C, respectively; mean relative humidity was 72%.
Rainfall was infrequent during the first 60 days after transplanting and more frequent thereafter as the monsoon arrived.
CF plots were irrigated to maintain a mean ponded water depth of 20–30 mm, an average of twice every three days.
AWD was imposed from 10 to 66 days after transplanting in Hardinath-1, and from 10 to 40 days in CH-45.
The maximum water depth following rewatering of AWD plots was 78 mm above the soil surface at 30 days after transplanting, while the minimum water depth before rewatering of AWD plots was 150 mm below the soil surface at 20 days after transplanting.
All plots were continuously flooded from approximately one week before, to one week after, flowering.
IRRI recommendations for “safe AWD” were adapted according to the local context.
Plots were flooded when water reached, or was just above, the plow pan (approximately 15 cm below the soil surface).
Soil water tubes (radius 52 mm) were used to record daily measurements of ponded water depth in CF plots and soil water depth in AWD plots.
Following lodging, plots were drained, and soil moisture maintained, to prevent grain germination.
AWD plots received on average less than half the irrigation volume of CF plots.

Quantifying Early Vigor and Late Lodging

Weekly manual tiller counts and daily manual measurements of the growing leaf of the main tiller of sample plants were carried out for six randomly selected border plants per plot.
Leaf elongation rate (LER) was calculated as the difference between its lengths on subsequent days.
LER was determined to have finished when its daily elongation rate fell below 10% of its maximum.
Green fraction (GF) per plot was calculated on a weekly basis using BreedPix 2.0 to analyze digital photographs.
Fresh weed weight per plot was calculated 31 days after transplanting.
Visual damage scores were given to CH-45 plots following lodging events at 66, 68, and 82 days after transplanting.

Measuring Yield Components

The proportion of effective tillers was calculated from the number of tillers and panicles per sample hill four days before harvest.
Grain size and weight were measured according to an IRRI protocol.
Three random samples of four hills per plot were selected and cut three days before harvest.
Number of grains and number of filled grains were manually counted for each panicle sample.
The remaining crop was cut at 88 days after transplanting, threshed, and air-dried.
Grain weight was adjusted to 14% moisture.

Statistical Analysis

Data were analyzed in R using quasipoisson models for time series data, and in SPSS with t-tests and nonparametric tests for individual days.
Weed and yield data were analyzed in R with two-way ANOVAs and general linear models, respectively.

Social Research Methods

A household survey was devised and piloted with four local households.
Hundred and one households were then selected at random from a list of all 321 households growing chaite rice in Agyauli VDC, and surveys were carried out during four weeks by the secretary of the village farmers’ committee.
Surveys gathered basic social information, irrigation sources and management, current irrigation scheduling techniques, and irrigation-related challenges.
Respondents were also asked about their perceptions and experiences of AWD and asked to select an irrigation scheduling technique from a choice of two in a series of three choice experiments.
Survey data were anonymized and translated from Nepali into English by local NGO staff.
Qualitative answers were given a numerical code for analysis, and basic descriptive statistics generated using SPSS.
Follow-up interviews were arranged to explore surprising responses about uses and perceptions of AWD.
Participant observation of irrigation management and maintenance was carried out throughout the field trial.

Results

Early Vigor and Yield

Tiller number was signifi cantly higher under AWD than CF from 21 days after transplanting onwards.
Irrigation and cultivar signifi cantly affected leaf elongation rate (LER) over the whole vegetative period and on some individual days.
Green fraction (GF) was signifi cantly affected by irrigation regime, and especially cultivar. Under AWD, CH-45 had signifi cantly higher GF throughout the vegetative period.
Weed fresh weight per plot was signifi cantly higher, but much more variable, under CF than AWD in CH- 45, but there was no signifi cant difference between treatments in Hardinath- 1.
Visual lodging damage scores in CH- 45 plots were signifi cantly higher under AWD following the second lodging event.
Grain weight per plot was not signifi cantly different between treatments or cultivars.
The number of fi lled grains per panicle was signifi cantly higher (by 89%) in Hardinath- 1, and under CF (by 12%).
The percentage of fi lled grains per panicle was signifi cantly higher in CH- 45 (by 18%) and again signifi cantly higher under CF (by 6.7%).
Individual grain weight was also signifi cantly higher in CH- 45 (by 21%), but did not differ between treatments.
Dry straw weight was signifi cantly greater in CH- 45 (by 39%), although in Hardinath- 1 it was slightly higher (by 13%) under AWD than CF.
Harvest index was therefore signifi cantly higher in Hardinath- 1 (by 26%), and slightly but non-signifi cantly higher under CF in both cultivars.
Water use effi ciency (WUE), calculated as moisture-corrected grain weight per plot divided by the total irrigation water received per plot during the whole crop cycle, was signifi cantly higher under AWD (by 133%), particularly in Hardinath- 1 under AWD.
There was a signifi cantly higher percentage of effective tillers (by 2%) under AWD overall. There was a strong, signifi cant interaction with cultivar

Social dimensions of irrigation scheduling

Survey respondents derive their irrigation either from a small shared gravity-fed irrigation canal, a kulo (55% of respondents, with kulo s shared between a mean number of 99 households); a shared or private shallow tubewell (36%); or a combination of kulo and tubewell
Most respondents fl ood their rice fi elds every ~2–3 days (55%). Some 15% irrigate daily, and 21% weekly, while others irrigate depending on soil type and elevation relative to water sources.
An overwhelming 93% of respondents use the same irrigation scheduling every year. Crucial times for the crop to have access to water were stated as during tillering (88% of respondents), a few days after transplanting (25%) and after weeding (14%).
More than half (56%) of respondents said they can always get access to enough irrigation water for chaite rice, with 5% qualifying that they can always access water from their tubewell but not the kulo.
When asked whether they felt their ability to irrigate chaite rice was restricted, the majority of respondents (63%) said that insuffi cient water was a problem, although 35% said it was not restricted.
For respondents who cannot always access enough irrigation water, ways of preventing yield reductions comprise technical strategies, such as using an electric or diesel motor pump or earth dams in the kulo to ensure water reaches the ﬁeld; using a tubewell to supplement their water supply and social strategies, such as co- operating with and borrowing turns from other farmers and speeding up the rotation system.
Given the limited formal promotion of AWD in Nepal, 91% of respondents had heard of AWD, and 90% said they had tried it on their farm.
The overwhelming majority of respondents to these questions preferred the CF option, or fl ooding every 2–3 days as in local practice, over AWD with fl ooding every 7–10 days (95:5 and 97:3 percent respectively), whilst opinion was more divided between the former two (57:43).

Discussion

Early vigor is not related to ﬁnal grain yield

Despite applying over 50% less water, the AWD treatment maintained a similar crop yield to CF plots, in spite of numerous physiological changes that occurred during the growing season.
Higher tillering is often associated with higher tiller abortion later in vegetative growth, the higher percentage of effective tillers in Hardinath- 1 grown under AWD is unexpected.
Although AWD delayed heading by 8 to 17 days in previous field trials, there were no differences in heading date between treatments, and plots under AWD had slightly, but not significantly, lower yields than CF plots.
Each cultivar maintained yield under AWD in a slightly different way: with no significant variation in yield components in CH- 45 which was exposed to a shorter duration of AWD, while more effective tillers in Hardinath- 1 compensated for a decrease in grain number.
There were also signifi cant interactions between irrigation treatment and cultivar. Despite much earlier heading of CH- 45 , higher grain weight and fi lling occurred, perhaps related to the slower decline in LER in this cultivar.
AWD has been proposed to increase resilience to lodging, since the crop may develop deeper roots to access soil water. However, in heavy soils such as the clays in Agyauli, drying can lead to particle cementation and soil compaction.

From promising science to a socially viable agronomic technique?

The agronomic results, when combined with the findings that many local farmers experience problems accessing sufficient water to irrigate chaite rice, suggest that AWD shows considerable promise.
AWD, as recommended by IRRI, relies on two key assumptions about the socio-economic context in which it is to be adopted. Firstly, that adopting AWD is a straightforward and rational choice for farmers, as long as water supply is reliable and can be controlled separately for each field or farm. Secondly, that there are strong, immediate incentives for reducing on-farm water consumption, for example if water is paid for per volume per farmer
Research on AWD adoption has recognized and addressed some of these problems, primarily advocating changes to water governance mechanisms like stricter water pricing, and the creation of infrastructure such as storage ponds.
Agricultural anthropologists such as Richards have long called for agricultural and agronomic research to consider farming as a performance: a lived process in which change is determined by dynamic environmental, cultural and social factors as well as economic costs and benefits.

Conclusions

AWD proved to be an extremely effective water-saving irrigation technology in the context of chaite rice production in Agyauli VDC in lowland Nepal, greatly increasing the water use efficiency of locally important varieties without decreasing yields.
However, it is unlikely to be adopted in Agyauli and areas with similar socio- economic conditions unless local water management, and use (or both) is altered.
Future extension work on AWD needs to consider ways of supporting farmers in changing water management and governance, but also ways in which the technology and the science underpinning it could be adapted for these contexts.
Increasing the involvement of farmers in AWD research, through the use of participatory and interdisciplinary methods, could improve uptake of these or similar water-saving techniques.