Fig 1.
Map of the study area in south-central coastal Bangladesh.
The districts covered by surveys are displaying salinity differences as well as the systems of polders.
Table 1.
Independent variables included in this study1.
Fig 2.
Correlation (corr) matrix of variables collected through smallholder household survey.
Table 2.
Factors representing key differences in rice-based smallholder farming systems of coastal Bangladesh.
Table 3.
Multi-level model summary showing the effects of year and embankment status on each factor (latent variable).
Fig 3.
Change in factors over time during the study period (error bars indicate 95% confidence intervals).
Factors are independent and represent key differences in smallholder farming systems for households within and outside polders in the study area.
Fig 4.
Changes over time in most influential variables contributing to each latent factor during the study period (2005–2015) (error bars indicate 95% confidence intervals).
The associated factor from Fig 3 is displayed in each panel. (F1 = Farming challenges, F2 = Economic status, F3 = Crop Management practices, F4 = Asset endowment, F5 = Farm characteristics, and ‘+’ = Positive loading, ‘-’ = Negative loading, HH = Household).
Fig 5.
Relationship between influential factors and farm-level crop productivity for households within and outside polders (at 95% confidence level).
A) Model predictions showing the effect of the three most important factors on household total annual crop production in terms of rice equivalent yield (REY), B) Model predictions showing the effect of the three most important factors on household crop productivity per unit area (REY per ha).
Fig 6.
Household poverty trajectory during the study period.
Diagram illustrates the movement of households into different categories of income. Note, categories refer to Per capita income per day in USD, A = ≥ 2, B = 1 to 1.99, C = 0.5 to 0.99, and D = < 0.5.