3.1 Farmers’ socioeconomic characteristics
The summary of farmers' socioeconomic characteristics have been supplemented in Table 1 (supplementary sheet). Almost two-thirds of the participants were under middle-aged to old-aged group while 34 percent could be categorized into young age group in this study. The rural youth's paradigm shift is clearly articulated in terms other than agriculture (Rekha & Ambujam, 2010). Education is the process by which desired changes in human behavior takes place. It is primarily supposed that a higher level of education should influence farmers to be aware of and critically evaluate the consequences of As contaminated groundwater irrigation. Two-thirds of the respondents (66 percent) and slightly over fifty percent of their family members had primary and low to medium education, respectively, while 26 percent of participants passed secondary to above secondary classes. It could be seen that only 8 percent of respondents and 22 percent of the family members were illiterate. Less than half (42 percent) of participants had small families, while 31 percent had large families. On the other hand, The knowledge status of the respondents showed that no less than 50 percent of farmers lack adequate knowledge of As and its impact on rice and vegetable cultivation with contaminated groundwater, while 34% possess high knowledge. All the participants in the study area had basic knowledge regarding the groundwater contamination with As used for drinking water due to substantial awareness-building circulation from government and non-government organizations in the past decades. However, the knowledge differences were created with the advanced aspect regarding the crop contamination due to As elevated groundwater irrigation. The family size also influences the farmers' perception of groundwater irrigation. More than half (58 percent) of the farmers had small, 29 percent had medium, and only 4 percent possessed large (3.01-6.00 ha) farm holdings, which are the collective possession from own and others land in borga. Farmers with larger farms are predicted to be more eager to convert their land to irrigated fields to minimize their loss rather than keeping the land barren (Rekha and Ambujam, 2010). The result also revealed that the farm size largely determined the annual income of the participants. Nearly 60 percent of the respondents had very low to medium-income mainly derived from agriculture, particularly rice and vegetables. Of the rest, 19 percent had high, and 20 percent had very high annual income from some business in addition to agriculture.
Cosmopoliteness influences farmers' perception since it enables them to be introduced to the latest technologies by exploring neighboring localities, towns, and abroad. Nearly half (48 percent) of the participants had low cosmopoliteness, followed by 32 percent with high cosmopoliteness. Similar to the cosmopoliteness, distribution of the farmers based on the information sources exposure showed less than half (48 percent) of the participants had a low level of information sources exposure, followed by 52 percent had medium to a high level to get the latest agriculture information. The farmers' educational status would have influenced the exposure to information sources. In addition, the information technology revolution had a profound impact on the farming communities.
All the farmers in this study had active participation in the agricultural and farm management activities; however, they were categorized based on their extent of involvement. More direct participation in farming enhances the actual field-based knowledge and experience and increases farm productivity due to the close observation and management possibility. Over half (57 percent) of the participants had medium to high direct participation in farming in their crop production, and the rest required some support from others for cultivation activities. Opinionatedness allows a farmer to exercise leadership capacity for the fellow crop growers regarding several decision-making processes, including crop variety selection, irrigation management, and intercultural operations. Nearly 50% of participants had low opinionatedness, 27 percent had medium, and 24 percent had high opinionatedness. Regarding agricultural credit use, mostly half (49 percent) of the farmers did not use any credits; only 7 percent had low use, while 22 percent received medium and high credits for rice and vegetable production. Different banks, NGOs, cooperative organizations, and businessmen provide the credits. Although presumed as the financial support for the initial period, the higher interest finally captures them into the trap for most cases.
The organizational participation based farmers’ distribution depicts that approximately half (49 percent) of the participants had low, one-third had high, and 18 percent had medium participation with different organizations. Organizational participation facilitates social networks to promote the information flow, which stimulates farmers' perceptions and decision-making on agricultural management (Bouma et al. 2008; Kilelu, 2004; Owusu et al., 2012). Farmers' innovativeness in the adoption of As mitigation irrigation management and other practices in the study area was evaluated. It elucidates that almost half (49%) of the respondents have no innovativeness, followed by 26 percent have medium level, and 25 percent have high innovativeness. The ownership of agricultural machinery largely determines the freedom of production management, especially the irrigation practice with a specific strategy. The respondents mainly had similar agricultural machinery where 43% and 31% possessed a medium and higher number of irrigation management tools. Table A.1 (supplementary sheet) also demonstrates that almost one-third of farmers had individual low, medium, and high-risk orientations. Those who had higher educational status, information sources used, and high organizational participation had a higher level of risk orientation (Rekha and Ambujam, 2010). In addition to the above, this study revealed that higher ownership of FPM also influences farmers' risk orientation. However, this psychological character influenced farmers' perception and the adoption of the As mitigating strategy.
3.2 Farmers’ perception
According to McGraw-Hill (2004), perception is the process by which sensory stimuli are registered as meaningful experiences, while Epstein et al. (2018) understand perception as the way of dispersing stimulation through structured experiences. Perceptions are more sophisticated constructs made up of simple pieces connected by association and are therefore more susceptible to the influence of learning. Though the senses of taste, hearing, touch, and smell have all been investigated, the vision has garnered the most interest. Perception is the process of becoming aware of or comprehending sensory information in psychology, philosophy, and cognitive science (McGraw-Hill, 2004). Table 2 demonstrates that 25 percent of the farmers possess high perception in the study area regarding As contamination in rice and vegetables due to contaminated groundwater irrigation, drivers of irrigating As elevated groundwater, it's possible mitigation strategies and health impact.
Table 2
Farmers’ perception on arsenic contaminated groundwater irrigation for rice and vegetables production (N = 200)
Category | Percent | Mean | Standard Deviation |
Low perception (129–136) | 39 | | |
Medium perception (137–155) | 36 | 146.6 | 14.16 |
High perception (157–178) | 25 | | |
Total | 100 | | |
On the other hand, 36 percent of them have a medium, and 39 percent have low perception levels. After a comprehensive assessment of farmers’ awareness regarding As in drinking water and foods, Mishra et al. (2021) reported that Bangladeshi farmers have comparatively high awareness regarding As in drinking water rather than in the foods they consume. A total of 43 statements under seven groups were administered to get a detailed understanding of farmers' perceptions (Table 3 in supplementary sheet). All the farmers responded to each of the statements from their learned experiences. A brief overview has been presented under seven subsections below.
3.2.1 Perception on As-contaminated or groundwater (AsW) or As free water (AsFW) use
Nearly two-thirds (62 percent) of the respondents strongly agree, and one-fourth agrees that no AsW means no rice/vegetable cultivation (Table 3 in supplementary sheet). They opined that AsW is available throughout the year for crop cultivation in their locality while AsFW is seasonal. Apart from this, an overwhelming (89 percent) of respondents still debated not using the AsFW in their fields. This might be because although they are aware of the drinking water As contamination, the majority of them still lack proper knowledge regarding the possible crop contamination with As. On the other hand, only 19 percent of farmers believe in the possibility of rice and vegetable cultivation with AsFW. The explanation for such a stance is that they possess comparatively larger farm holdings with adequate irrigation management tools.
3.2.2 Drivers for irrigating AsW
Easy accessibility is the prime cause for AsW use, is unequivocally declared by all the participants in this study. Nearly 98 percent of the respondents claimed that they prefer irrigating their crop fields with some shareholders to reduce the production cost. This prevalent scenario of field irrigation practice threatens the choosy irrigation management in this study area. The scarcity of the AsFW (e.g., surface water), particularly during the winter season, compels them to go for groundwater irrigation. Another reason for using AsW is the saving purpose of the AsFW for household use, as reported by 26 percent of the respondents. Only 3 percent of the farmers are self-sufficient to irrigate with their own pump and manage irrigation as per their choice.
3.2.3 Effect of AsW irrigation on crop fields
While demonstrating the impact of AsW irrigation on crop fields from their experiences, two-thirds of the farmers remained undecided whether the AsW led to add additional As in their crop fields or not, although the rest one-third believed in As addition. Similarly, four-fifth of the farmers were undecided regarding the fertility loss of their crop fields with As incorporation due to groundwater irrigation. On the other hand, slightly over 50 percent of the participants observed their irrigation channel became red, 40 percent reported yield loss near the channel, and land became hard.
3.2.4 Effect of AsW irrigation on rice & vegetables
Only 19 percent of the respondents believe in the As accumulation in rice & vegetables upon As contaminated groundwater application. The level of education, organizational participation, information source exposure, and cosmopoliteness enhanced their knowledge regarding this issue and influenced their perception. More than 95 percent of farmers were undecided about the other parameters such as the impact on tillering, influence on plants' height, uniformity of flowering, plant growth and grains maturity, grains filling percentage, or yield reduction. However, only 2–4 percent of participants agree with those advanced symptoms.
3.2.5 Impact of fertilizers and pesticides on As addition
Application of pesticides (Campos, 2002) and fertilizers, especially Phosphate fertilizer, (Jayasumana et al., 2015) may escalate As levels in the crop fields. Almost all the respondents were undecided since they did not get such information from any media or social networking.
3.2.6 Health impact
From their knowledge of groundwater As contamination and knowledge about the As related health impact from the drinking water exposure, 7 percent agreed, and 35 percent of the farmers highly agreed with the possible As transfer to the human body due to As elevated rice and vegetables consumption. However, more than fifty percent of the respondents remained undecided. Similarly, 45 percent of the participants perceive As may cause cancers, while 39 percent agreed on the development of skin lesions.
3.2.7 Farmers' practiced As mitigation strategy
Nearly one-third of farmers perceive that alternate wetting and drying (AWD) and surface water irrigation can reduce As accumulation in rice and vegetables. Seven percent of the participants believe that raised bed rice cultivation would limit As loading in rice grains. A very insignificant part (1–2 percent) of the participants perceive fertilizer management, such as supplementing with more urea, MoP, gypsum, zinc sulphate, cow dung, and intercultural operations such as mulching in vegetable fields or spreading Ash would limit As accumulation.
3.3 Correlations
Correlation coefficients between the independent and dependent variables has been estimated (Table 4 in supplementary sheet); and Table 5 (in supplementary sheet) shows the correlation matrix representing the overall interaction between the variables. According to Table 4 (supplementary sheet), among the socioeconomic characteristics, farmers' age, annual income, family education, family size, farm size, and agricultural credit use were non-significant. In contrast, farmers' age and family size were negatively correlated with their perception of As elevated groundwater irrigation for rice and vegetable production. The study of Alam (2001) and Kabir (2002) revealed a negative correlation of family size with perception, while Majlish (2007) reported a non-significant correlation. Afique (2006), Pal (2009), and Adeola (2012) revealed that farm size had no discernible effect on farmers' perceptions. Friedler et al. (2006) argued that there was no correlation between the age or income of farmers and their perceptions. Islam (2000) observed no association between farmers' utilization of credit and their perception.
On the other hand, farmers' education, knowledge, information sources, direct participation in farming, cosmopoliteness, opinionatedness, innovativeness, risk orientation, farm power and machinery (FPM), and organizational participation were positively significant with perception at a 1% significance level (shown in Table 4 in supplementary sheet). Pal (2009) revealed that farmers' education positively correlates with their perception. Kabir & Rainis (2012) and Adeola (2012) also found that education significantly affects farmers' perceptions in Bangladesh and Nigeria. Individuals with higher education levels usually perceive risks and understand mitigation necessity in a very advanced way (Dosman et al., 2001). In their survey in Gujarat province in India, Kumar & Popat (2010) exposed that knowledge, a psychological characteristic of the farmer, had a significant positive association with their perception. The study of Adeola (2012) reported similar findings in Nigeria. The farmers' information sources can play a crucial role in building positive or negative perceptions of any phenomenon. Rezaei et al. (2017) claimed a significant relationship between farmers' exposure to the information sources and their perception. Farmers engaged in farming activities helps determining their decision-making capacity in any circumstance (Larsen et al., 2002; Rahaman et al., 2018). Therefore, direct farming participation had a significant relationship with farmers' perceptions (Rokonuzzaman, 2016). Islam (2000) revealed a significant positive correlation between farmers' perception and annual income.
Regarding the association between farmers' ownership of FPM and their perception, through their study in the water markets in Bangladesh, Mottaleb et al. (2019) demonstrate that irrigation pump ownership largely determines farmers' perception. However, they concluded that since the irrigation system in Bangladesh is mainly based on pumping underground water, pump ownership significantly influences the structure and choice of irrigation practices. Regarding the relationship between organizational participation and perception, Keshavarz and Karami (2013) reported that membership in social organizations positively influences farmers' perceptions. Membership in formal or informal organizations helps the farmers get benefits and social support (Fuller-Iglesias et al., 2009). Segnestam (2009) argued that organizational participation helps disseminate innovations and develop mutual trust among the farmers, which eventually shapes farmers' perceptions. While studying cosmopoliteness, Alam (2001) noted a significant positive association between farmers' cosmopoliteness and their perception. According to Hamid (1995), there is a significant relationship between cosmopoliteness and farmers' use of the recommended level of plant protection practices. Farmers' opinionatedness and perception were found to have a significant positive association in the study of Islam (2000). Londhe et al. (2018) discovered a substantial positive relationship between perception and participants' risk orientation and innovativeness. The study of Rekha & Ambujam (2010) in Tamil Nadu, India, about the farmers' perception of contaminated water irrigation revealed a significant positive correlation between farmers' perception and their educational status, information sources, annual income, farm size, risk orientation, and innovativeness.
3.4 Regression results
Predictor variables (independent variables) that explain farmers' perceptions (the dependent variable) were determined using a stepwise multiple regression analysis. Table 6 illustrates the findings of stepwise regression. The total variance explained by the five independent variables is 0.884 (R = .889, R2 = 0. 884), as seen in this table. Of the total variance, participants' knowledge explained 74.6%, direct participation in farming 8.2%, information sources 4.5%, participant education 0.7%, and organizational participation 0.8%. The F value for participants' knowledge, direct participation in farming, and information sources are significant at 0.1% level, while for participants' education and organizational participation are significant at 5% level. This means that the five recognized predictor variables account for 88 percent of the variance in the dependent variables.
Table 6
Regression of the estimated perception on the independent variables (N = 200)
Variables | R | R Square | Adjusted R Square | Std. Error of the Estimate | R Square Change | F Change | Sig. F Change |
Participants knowledge | .865 | .748 | .746 | 7.140 | .748 | 291.373 | .000 |
Direct participation in farming | .911 | .830 | .826 | 5.899 | .082 | 46.587 | .000 |
Information sources | .935 | .875 | .871 | 5.089 | .045 | 34.297 | .000 |
Participant education | .939 | .882 | .877 | 4.973 | .007 | 5.533 | .021 |
Organizational participation | .943 | .889 | .884 | 4.832 | .008 | 6.638 | .012 |
3.5 Path analysis
With the path analysis, the total effects are broken down into indirect and direct effects on certain independent variables. Direct participation in farming presents the highest positive total effect (0.855) and direct effect (0.503), whereas information sources show the highest positive indirect effect (0.624) (Fig. 1). Organizational participation (0.796, 0.226) and participant education (0.716, 0.196) represent the second and third highest total and positive direct effect, respectively, both with positive impact (Table 7 in supplementary sheet). Risk orientation (0.593) ranked second and organizational participation (0.570) ranked third in terms of positive indirect effect. Out of the eight independent variables, four variables [participant education (X1), knowledge (X2), information sources (X3), and cosmopoliteness (X5)], each of these, has a highest indirect effect on farmers' perceptions directing
towards transformation through direct participation in farming and organizational participation. On the other hand, another three [innovativeness (X6), risk orientation (X7), organizational participation (X8)] have the highest indirect effect through direct participation in farming and participant education which are depicted in Table 7 (supplementary sheet). However, path analysis revealed that just a few variables directly impacted farmers' perception levels. However, interconnected variables were principally involved for the effect of several variables on farmers' perceptions.
3.6 Arsenic content in collected samples
The study revealed As content in irrigation water (ranges from 0.108–0.356, 0.111–0.338, 0.110–0.371, 0.041–0.364, and 0.065–0.356), soils (ranges from 15.645–30.675, 14.325–29.612, 16.327–32.1, 11.895–32.667, and 11.375–32.262), vegetables (ranges from 0.83–3.56, 0.26–3.25, 0.45–3.8, 0.21–3.9, and 0.23–3.84), rice grains (ranges from 0.192–0.75, 0.22–0.69, 0.18–0.89, 0.09–0.86, and 0.117–0.74), and farmers’ scalp hair (ranges from 0.34–2.21, 0.4–2.36, 0.42–2.38, 0.32–2.44, and 0.35–2.17) for Sadar, Faridganj, Matlab north, Kachua, and Hajiganj, respectively. Table 8 demonstrates the probability level of As content in all five items. Arsenic in irrigation water is significantly different at a 1% probability level in the study sites.
Table 8
Comparison of As concentration in different components collected from five (05) different location of Chandpur districts of Bangladesh
Locations | As in irrigation water (mg/L) (against background value 0.1 mg/L by FAO and 0.01 mg/L by WHO; Chakraborti et al., 2018; WHO, 2004) | As in soil (mg/kg) (against global average 10 and FAO limit 50 mg/kg; FAO, 1992; Rahman et al., 2013) | As in vegetable (mg/kg) (against permissible limit 0.5 to 1.0 mg/kg; Liu et al., 2010; MAFF, 1997) | As in Grain (mg/kg) (against permissible limit 0.37 mg/kg; WHO, 2016) | As in hair (mg/kg) (against background value 0.08–0.250 and toxicity indicator 1.0 mg/kg; Arnold et al., 1990) |
Hajiganj | 0.227ab | 21.90b | 2.03ab | 0.459a | 1.24ab |
Kachua | 0.204bc | 20.69c | 1.82cd | 0.418b | 1.08c |
Matlab north | 0.192c | 21.10b | 1.61d | 0.367c | 1.00c |
Faridganj | 0.234a | 23.00a | 2.21a | 0.472a | 1.28a |
Sadar | 0.217b | 23.08a | 1.93c | 0.399bc | 0.96cd |
LS | ** | ** | *** | * | * |
CV (%) | 6.81 | 8.81 | 5.51 | 6.28 | 6.70 |
SE (±) | 1.17 | 0.93 | 1.24 | 1.15 | 0.96 |
In column, means followed by different letters are significantly different. LS means level of significance, CV means co-efficient of variance, SE means standard error, ***means at 0.1% level of probability, **means at 1% level of probability and * means at 5% level of probability
The lowest As content in irrigation water is revealed from Matlab north while the highest is found in Faridganj. Similarly, As level in the study sites' soil is significantly different (p ≤ 0.01), where Matlab north and Hajiganj's soil contain statistically similar As to Sadar and Faridganj. significantly (p ≤ 0.05) higher As is found in grains from Faridganj and Hajiganj compared with that from Kachua, Matlab north, and Sadar. In contrast, the lowest and highest grain As is recorded in Matlab north and Faridganj, respectively. Vegetables As in all the five study areas differs significantly at a 1% probability level. Vegetables As level from Hajiganj is pretty close to Faridganj, and the same for Sadar is very close to Kachua. Similar to the grain As content, the lowest and highest vegetables As is recorded in Matlab north and Faridganj, respectively. Faridganj and Hajiganj have been found to have significantly ((p ≤ 0.001)) higher and closely resemble hair As concentration. Again, hair As level observed from Matlab north and Kachua is also statistically similar. Hair As content of Sadar is also at immediate proximity to Matlab north and Kachua. At all five locations, the mean As concentration in vegetables and irrigation water is much higher comparing with the permissible limit (Chakraborti et al., 2018; WHO, 2004), while the As level in soil is higher than the As level on the global scale but below the FAO proposed limit for agriculture (FAO, 1992; Rahman et al., 2013). Except for Matlab north, grain As content surpassed the safe limit (WHO, 2016) at all places. On the other hand, scalp hair As is recorded above the toxicity limit for four locations except for Sadar revealed below toxicity level but above the background value. Results suggest significant As transfer from irrigation water to rice and vegetables and subsequent body loading.
3.7 Principal component analysis (PCA)
Figure 2 depicts four unique clusters are produced by the varying lengths of the eigenvectors. Correlations between items are represented by the angle between eigenvectors, and the length of each eigenvector is proportional to the variance of the corresponding data item. Hair As, grain As, water As, soil As, and vegetable As are all examples of factors that fall into one of the five categories denoted by clusters (I), (II), (III), or (IV). Parameters with identical values are observed to cluster together in Fig. 2. This divergence can be explained by the fact that the As in irrigation water and soil (cluster III) contribute to a similar variance, while the As in scalp hair (clusters I), grain
(clusters II), and vegetables (cluster IV) do not. Lengthwise, Cluster II was the lowest, and Cluster IV was the highest, suggesting the lowest and highest variations, respectively. It's clear that there's a strong relationship between categories (I) and (II) among the four options. Table 9 (in the supplementary sheet) displays the PCA results for As concentration of various parameters. Table 9 shows that the first principal component PC has an eigenvalue greater than 1, indicating that it adequately describes the variances. As for irrigation water (0.458), grain (0.448), and soil (0.446), these three factors account for the vast majority (92.5%) of the total variance explained by the first PC (Table 9). Values highlighted in bold in the Table are particularly relevant to understanding the PC, as a higher numerical value denotes a more substantial contribution. Thus, the PC1 loading values were largely influenced by the parameters of irrigation water As, grain As, and soil As.