Characteristics and variability of rainfall
Figure 6a indicates the variability of annual rainfall at the stations located in each zone of Burma from 1901 to 1939. The annual rainfall in the Coastal Zone was comparatively higher (more than 4000 mm at most of the stations) than in other areas. The annual rainfall amount mostly ranges from 2000 to 3000 mm in the Delta Zone, from 700 to 1300 mm in the Dry Zone, and from 1500 to 2200 mm in the Hilly Zone. Figure 6b indicates the average monthly rainfall from 1901 to 1939 for each zone. The monthly rainfall amount was also comparatively high in the Coastal Zone than in other areas, and less in the Dry Zone. The monthly average rainfall in Dry Zone was below 200 mm in the months. Figure 6b indicates that the rainfall mainly occurred from May to October in all zones.
Figure 7 indicates spatial pattern of change trend of seasonal local and basin average rainfall. The changing trends in seasonal local and basin average rainfall for the selected districts in the period from 1901 to 1939 are presented in Additional file 1: Figs. S3 and S4, respectively. The detailed statistics summary of change trend analysis of rainfall for all districts is presented in Additional file 1: Table S6. There was a clear interannual variation in the rainfall in each district. SLR positively varies over time at 25 districts out of 37 districts, while the SBR positively varies over time at 32 districts out of 37 districts, i.e., most of the districts experienced an increase in seasonal rainfall for the period 1901–1939. However, the increasing trend of local rainfall over time was statistically significant with a significant level of 5% (two-tailed test) only in six districts, while the increasing trend of basin average rainfall was statistically significant with a significance level of 5% or 10% (two-tailed test) in 14 districts (Additional file 1: Table S6), which were located in the Delta and Dry Zones.
Variations in rice yield and production
Figure 8a indicates the spatial pattern of the average value of rice yield from 1901 to 1939. The rice yield was highest in the Delta Zone than in other zones because this zone had favorable soil and rainfall conditions for rice crop cultivation than in other areas. As expected, this zone had an average rice yield that was > 1500 kg/ha in all the districts where Gleysols were widely distributed, except in Insein, and > 1800 kg/ha rice yields in Myaungmya (2000 kg/ha), Henzada (1869 kg/ha), and Tharrawaddy (1889 kg/ha) Districts. The lowest rice yield was obtained in Insein District (1424 kg/ha). Furthermore, while the average value of district-level rice yield in the Coastal Zone varied from 1360 kg/ha in the Arakan District to 1710 kg/ha in the Sandoway District, the rice yield was > 1500 kg/ha in the Sandoway (areas with Cambisols and Fluvisols), Akyab (areas with Gleysols and Cambisols), and Mergui (areas with Acrisols and Nitosols) Districts. It was also observed that although the average rice yield was much higher in the Mergui District, low fertile soils (Acrisols and Nitosols) are widely distributed in the district. Along the seacoast in the Amherst and Thaton Districts of the Coastal Zone were high-yielding areas; however, crop damage from flooding of the Salween River led to low yields in many of the inland areas, reducing the average value of rice yield for these districts (Siok-Hwa 2012). Alternatively, rice yield was lowest in the Dry Zone. A study observed that while there were high variations in rice yield of the Dry Zone because some areas in this zone had irrigation systems (large-scale irrigation systems developed by the Burmese King and small-scale irrigation systems constructed by local people, such as small diversions, wells, and ponds), other areas received little rainfall (Win 1991). The district-level rice yield in the Dry Zone varied from 702 kg/ha in the Myingyan District to 1828 kg/ha in the Minbu District. Additionally, the highest yield in the Dry Zone was obtained in the Minbu District, where the soil types were Cambisols, Gleysols, and Luvisols, and most of the crops were cultivated under an irrigation system. Moreover, the average value of rice yield in the Dry Zone was higher in Mandalay (1608 kg/ha) and Kyaukse (1726 kg/ha) Districts, where Vertisols, Luvisols, and Acrisols were distributed, and most of the crops were cultivated under an irrigation system. In contrast, the average value of rice yield was 950 kg/ha in the Meiktila District, even though the crops were cultivated under irrigation and in good-fertile soils (Gleysols, Luvisols, and Vertisols). One possible reason for such lower rice yield in the Meiktila District was crop damage due to flooding in the areas. We also observed that the average values of rice yield in Shwebo (areas with mostly Vertisols, Gleysols, and Luvisols) and Yamethin (areas with Gleysols, Luvisols, and Nitosols), where half of the crops were cultivated under irrigation, were 1258 and 1313 kg/ha, respectively. In contrast, the average value of rice yield in the Hilly Zone varied from 1243 kg/ha in the Salween District (areas with Acrisols) to 1758 kg/ha in the Myitkyina District (areas with Gleysols and Acrisols). Nevertheless, the rice yields in Upper Burma were considerably lower than that in Lower Burma, mainly due to insufficient water supply in Upper Burma (Siok-Hwa 2012).
Figure 8b indicates the changing trend of average rice yield over the period from 1901 to 1939 for each zone, and rice yields in all zones show the long-term reducing trend during the study period. The long-term reducing trend of rice yield in all areas was statistically significant at 0.1% significance level of two-tailed test.
Figure 9 indicates the variation of the rice crop planted area, rice-harvested area, and rice yield for selected districts from each zone for the study period. The crop areas were rapidly expanded in the Delta and Coastal Zones in the Lower Burma than in other areas where the annual or seasonal rainfall amount was sufficiently higher than the water requirement for the rice crop cultivation. Rice production and yield fluctuated depending on the behavior of monsoon rainfall and its distribution pattern. Particularly in the years 1919 and 1923, there was a large reduction in rice yield in some districts located in the Coastal, Delta, and Dry Zones. The rice-harvested areas were also lower than the rice crop planted areas in the years 1919 and 1923 in the zones. The difference between planted and harvested rice areas likely depends on the zone. In the districts located in the Dry Zone except for Prome District and some districts of the Hilly Zone, the rice-harvested area was usually lower than the rice crop planted area, and a reason for such lowering in rice-harvested area was insufficient amount of rainfall in the zone for rice cultivation. The occurrence of continuous rainfall during the early growth period of the rice plant often caused floods in the areas, which reduced rice yield and production (Win 1991). In some years, the reduction in rice yield was caused by the occurrence of less rainfall in a particular area.
Figure 10 indicates the interannual variations and trends of rice yield over the period from 1901 to 1939 for the selected districts, and summary statistics of the trend analysis are indicated in Additional file 1: Table S7. Figure 11 indicates the spatial pattern of change trend of rice yield with a statistically significant level (0.1% or 1% or 5% or 10%) (two-tailed test). The changing trend was statistically significant with a significance level of 0.1% or 1% or 5% or 10% at 31 districts. The results of trend analysis indicate that 33 districts out of 37 districts experienced a decrease in rice yield for 1901–1939. The rice yield increased in trend only in four districts (Arakan, Hanthawaddy, Pyapon, and Salween). The rice yield mainly depended on the fertility of the land and the amount and distribution of rainfall in a particular area. The reducing trend of rice yield over the years in Burma during the colonial period was attributed to several reasons, such as variability of seasonal and monthly rainfall and other environmental factors during the crop growth period, the estimation of rice yield covering wider areas representing various types of soil and climate with the years, and reaching the minimum fertility rate of the land after several years of rice cultivation. During this period, cropland areas for rice cultivation were expanded by clearing swamps, grasslands, and forest areas, and the fertility of such lands possibly became the minimum level after many years of cultivation. Even though rice yield tended to decrease during the study period, rice production remarkably increased due to the rapid expansion of cropland areas for rice cultivation.
Figure 12 shows the cross-plot between rainfall and rice yield anomalies with their linear tendency. The linear regression analysis statistics are also indicated in Additional file 1: Table S8. The results showed that rice yield anomaly positively varied with rainfall anomaly in all dry and hilly zone districts, except Salween District in the Hilly Zone. We also observed that rice yield anomaly negatively varied with rainfall anomaly in four (Amherst, Thaton, Sandoway, and Akyab) of eight districts in the Coastal Zone and seven (Rangoon, Pyapon, Myaungmya, Bassein, Maubin, Insein, and Pegu) of 11 districts in the Delta Zone. There were also some rice yield variation tendencies due to rainfall deviation, such as a reduction in rice yield when rainfall is lower than the average rainfall, particularly in most dry and hilly zone districts or higher than the average rainfall in more than half of the districts located in the Delta and Coastal Zones. However, the linear relationship between rice yield and rainfall anomalies was statistically significant, with a significance level (p-value using two-tailed test) of 5% or 10% only in the Sandoway District (areas with mostly Cambisols) in the Coastal Zone; Insein District (areas with Gleysols and Nitosols) in the Delta Zone; Prome (areas with Gleysols and Nitosols), Thayetmyo (areas with mostly Luvisols and Cambisols), Meiktila (areas with Gleysols, Luvisols, and Vertisols), and Myingyan (areas with Luvisols) Districts in the Dry Zone; and Bhamo District (areas with Gleysols and Acrisols) in the Hilly Zone. The annual variation in rainfall and rice yield anomalies with their 5-year moving average values is presented in Additional file 1: Figure S5. The characteristics of rainfall and rice yield anomalies were completely different. The rainfall anomaly widely differs from period to period; however, results of rice yield anomaly clearly show the distinction of the period with higher or lower rice yield than average rice yield. Rice yield tended to decline after 1910 in most districts (Win 1991; Okamoto 1998). Mostly higher than average rice yield was observed before 1910 in the Coastal Zone and before 1918 in other zones, while mostly lower than average rice yield was observed after 1910 in the Coastal Zone and after 1918 in other zones. Moreover, during the latter part of the study period, the reduction in rice yield was possibly due to the variability of various environmental factors and the expansion of cropland areas with less favorable environments.
Exploration of relationship between rainfall and rice yield
The summary results of multiple regression analysis with slope coefficient, constant, significance level, and t-stat values of each rainfall index are indicated in Additional file 1: Table S9 for all districts. The t-stat values with bold in Additional file 1: Table S9 are statistically significant variables (at 10% or 5% or 1% or 0.1% significance level of two-tailed test). The results of multiple regression analysis indicate that the rice yield varied negatively with SLR, NDD, CDD, CWD, and SBR at most of the districts located in the Coastal Zone. In contrast, rice yield varied positively with HRD and 5DR. In Delta and Hilly Zones, rice yield varied negatively with all rainfall indices at most of the districts. In the Dry Zone, rice yield correlated positively with SLR and SBR and negatively with NDD, CDD, CWD, HRD, and 5DR at most of the districts, respectively. The results show that the flood-related indices, such as SBR, HRD, and 5DR, had both positive and negative effects on rice production, i.e., rice yield varied negatively with those indices at some districts and positively varied in some other districts. The rice yield negatively or positively varied with rainfall indices. We observed that the relationship between rice yield and each rainfall index was statistically significant, with a significance level of 0.1% or 1% or 5% or 10% (two-tailed test) in the districts presented in Table 3. The SLR was statistically significant, with a significance level of 10% only in the Bhamo District in the Hilly Zone. Similarly, the NDD was statistically significant, with a significance level of 5% or 10% in the Tavoy District of the Coastal Zone; Hanthawaddy, Tharrawaddy, and Toungoo Districts of the Delta Zone; Meiktila and Mandalay Districts of the Dry Zone; and the Katha District of the Hilly Zone. Furthermore, while CDD was statistically significant, with a significance level of 1% or 10% in the Pegu and Toungoo Districts of the Delta Zone and the Bhamo District of the Hilly Zone, the CWD was statistically significant, with a significance level of 10% only in the Rangoon District in the Delta Zone. Results also showed that the HRD was statistically significant, with a significance level of 5% or 10% in Sandoway and Kyaukpyu Districts of the Coastal Zone; Magwe District of the Dry Zone; and then Salween and Katha Districts of the Hilly Zone. Likewise, 5DR was also statistically significant, with a significance level of 5% or 10% in Amherst and Kyaukpyu Districts of the Coastal Zone; Bassein District in the Delta Zone; and Bhamo District in the Hilly Zone (see Table 3 and Additional file 1: Table S9).
The relationship between rice yield and annual rainfall variability with various rainfall indices was analyzed using multiple regression analysis, and the effect of annual rainfall variability on rice production was explored. However, the monthly variability of rainfall during the crop growth period in the years may significantly impact on rice yield because the amount and distribution of rainfall in a particular year determine the success or failure of rice crop production. Rice crop yields will boost significantly if the precipitation occurs at the right times and in the right amounts. To explore the monthly variability of rainfall in the years, the monthly rainfall during monsoon for rice yield failure or success years was analyzed. Figure 13 indicates the variation of monthly rainfall for crop failure years for the selected districts. (Blue bar diagram in the figure is the average of the period 1901–1939.) From the results, we observed that the monthly rainfall that resulted in years of crop failure was usually either comparatively higher during the crop growth period (particularly in July and August) than the average monthly rainfall of period 1901–1939 or less during the early or latter half of crop growth (Fig. 13). For example, there was a higher rainfall between July and August in most districts of the Coastal and Delta Zones in 1923 (e.g., Bassein, Maubin, Henzada, Pegu, Sandoway, Akyab, and Prome (near delta area) Districts in Fig. 13). Contrastively, in 1919, less rainfall was observed during the latter growth stage of rice crops in most districts. Investigations also showed that some districts experienced rice crop failure due to less rainfall during the early or middle stage of crop growth in some years (e.g., Maubin, Pegu, Meiktila, Shwebo, and Katha Districts in 1918 or 1920). However, other districts had higher rainfall during the early growth stage and less rainfall during the latter growth stage of the rice crop in 1919 (e.g., Bassein, Maubin, Sandoway, Akyab, and Shwebo Districts in Fig. 13). Additionally, while higher rainfall during crop growth often triggered flooding, which reduced production as a result, lesser rainfall during crop growth also reduced rice production. The monthly distribution of rainfall for crop success years (years with higher yields than the average value of rice yield from 1901 to 1939) is indicated in Additional file 1: Figure S6. Furthermore, we observed that the amount of monthly rainfall during the crop growth period for crop success years was usually closer to the average value of 1901–1939 in most districts located in the Delta and Coastal Zones and some districts in the Hilly Zone. These results confirmed that the monthly distribution of rainfall within a particular year determines the influence of rainfall on rice production. Thus, it is vital to analyze the impact of monthly rainfall on rice yield and production. Previous researches reported that continuous rain during the early growth period of the rice plant often caused floods, reducing rice production in Burma (Win 1991). The reduction in rainfall amount could cause negative consequences on rice production, particularly in the Dry and Hilly Zones of Burma.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.