2.1 Sustainability of the Adaptation Strategies Water is essential for agricultural farming, and the increasing use of irrigation water, especially groundwater, has contributed significantly to crop productivity in Bangladesh. Rice yield, for example, increased from 1 MT/ha in 1971/72 to 2.88 MT/ha in 2010/11. Much of this increase in yield was due to an increase in the share of irrigated rice, especially during the dry season: Boro rice farming increased from 11% in 1972/73 to over 65% in 2010/11 (BBS, 2012). The contribution of surface and groundwater sources to the total irrigated agriculture has changed significantly over the years – groundwater has increased from 41% in 1982/83 to 68.5% in 1996/97 to over 80% in 2010/11, while surface water declined from 59% to less than 20% over the same period (BADC, 2011; Planning Commission, 1997). Groundwater usage has increased over the years due to an increase in farming intensity, the government’s massive investment in irrigation development and the shrinking of surface water resources. Delayed monsoons and less and/or an uneven distribution of rainfall under the impact of climate change have aggravated water availability and increased water scarcity. Over the last three decades the area under irrigation has expanded significantly throughout the country in order to increase food production, mainly through a rise in the number of shallow tube wells (STWs). The area under surface water irrigation has declined since the 1980s, while the area under STW irrigation has increased by a factor of ten (BADC, 2011). The shallow aquifer has become highly contaminated by arsenic in many parts of Bangladesh, including in the study area. This situation poses a serious threat to human health and livelihood.
2.2 Study Area and Data Sources Rajshahi is in the heart of the drought-prone northwestern region of Bangladesh (24.40°N 88.50°E) (Fig. 1). The district has an area of 2407 km2 with a population of 2.4 million people (population density of 997/km2), making it the largest district of the Barind Tract (33% of the region). Because of its predominant dependence on crop agriculture, the district is referred to as the ‘bread basket’ of the country. Rice farming is very sensitive to changes in weather and climatic conditions. The climate of the region is semi-arid and is characterized by low rainfall and high fluctuation of precipitation. The annual rainfall is about 1400– 1600 mm, while most parts of the country receive at least 2000 mm of rain per year. However, seasonal and inter-annual variability of rainfall is high in the district where a disproportionate amount of rain (90%) falls during short spells (June–October), even though the total amount of rain does not change much (BMD, 2013; Rashid et al., 2013; Shahid, 2011). As a result, agricultural production is constantly threatened due to the high likelihood of drought and recurring water shortages. Traditionally, the rural economy and livelihood depend heavily on rain-fed agriculture. However, over the years the share of the irrigated rice farming area to the net cultivated area in the region has increased considerably. Water scarcity is expected to exacerbate this vulnerability through changes in precipitation patterns, including heavier and more erratic rainfall during monsoons (June–October) and a lack of rainfall in some areas during the winter season (October–March).
A cross-sectional survey to collect data from farming households in the Rajshahi District was adopted for this study. Household data were collected from 12 randomly selected villages in the district during October–December 2013. A multi-stage random sampling technique was employed to select the Upazillas (sub-districts), villages and households. At the first stage, random sampling was used to select two Upazillas. At the second stage, six villages were selected from each of the selected Upazillas, making a total of 12 villages.
As the number of farming households within each village varies considerably, a predetermined number of 15% households from each village were selected for the survey which gives a sample size of 550 for the 12 villages surveyed. This is considered to be sufficient: Bartlett et al. (2001) considered 5% to be adequate for cross-sectional household surveys. Furthermore, rural farming communities in the study area make up a mostly homogeneous group which also validates the use of a small sample (Blaikie, 2010).
The unit of analysis was the farm households, and these were selected by simple random sampling using the list of farming households collected from the Department of Agricultural Extension. Households were approached until the required number of surveys for a particular village was completed. Finally, heads of the households were selected as survey participants because they usually have the decision-making power for farming and the household’s resources. Interviews were conducted by trained interviewers under the supervision of the researcher, either at participants’ homes or a suitable place agreed upon (e.g., farms and community meeting places). Four observations were dropped due to data inconsistencies (e.g., interviewees were found not to be the head of households) which resulted in 546 observations for data analysis. Additionally, focus group discussions, key informant interviews and secondary sources of data augmented the survey to get a holistic view of the drought adaptation strategies practiced by the participating communities. A structured questionnaire was administered in person to elicit data regarding several aspects of adaptation strategies practiced by farming households and their socio-economic characteristics, institutional access, farm characteristics and perception of climate change.