Production without medicalisation: Risk practices and disease in Bangladesh aquaculture
Abstract
Improved biosecurity and livestock disease control measures in low resource settings are often regarded as beneficial for agricultural productivity, rural incomes, global health, and sustainability. In this paper we present data from a study of shrimp and prawn aquaculture in Bangladesh to argue that this relationship is not as straightforward as it would seem. Analysing quantitative and qualitative data from a multi-method field study involving 300 “missing middle” farmers, we demonstrate the importance of socio-economic and ecological conditions to any disease management strategy. We describe how a technical programme to introduce “disease-free” seed faltered partly as a result of the farmers' tendency to offset disease and livelihood risks by frequently re-stocking their ponds. Changes to seed provision were accompanied by calls to alter farmers' livestock production practices. Paradoxically, these changes exposed farmers to more intense risks, potentially locking them into unsustainable disease management practices. The analysis emphasises that vernacular farming practices should be considered as key assets rather than barriers to disease management strategies, and that closer attention be paid to value chain and other risks as drivers of unsustainable practices.
Graphical Abstract
Disease burdens threaten future food production, particularly so in the case of aquaculture. Reducing disease burden needs to consider not only the incidence of disease but also the socio-economic effects of changing farming practices. Employing a multi-method approach to farms in SW Bangladesh, we argue that understanding disease risk practices are central to any attempt to reduce disease burdens and unsustainable treatments of disease.
1 INTRODUCTION
Producing enough food of the desired quality and nutritional value during a period of rapid environmental change is dependent upon the management of an increasing array of emerging and re-emerging diseases that affect crops, livestock and people (Perry & Grace, 2009). In Asia, the issues may be particularly apposite; “The region is home to dynamic systems in which biological, social, ecological, and technological processes interconnect in ways that enable microbes to exploit new ecological niches” (Coker et al., 2011, p. 9). Finding sustainable solutions to these challenges is pressing. Currently, well over half of the global consumption of antimicrobials occurs in the food and farming sector (Van Boeckel et al., 2015), with lasting effects on human, animal, and environmental health (Laxminarayan et al., 2013). Finding alternatives to this growing reliance upon unsustainable treatments will depend in part upon reducing disease burdens. In this paper, we focus on a particular food production sector and location (shrimp and prawn aquaculture production in southwest Bangladesh) to argue that lessening this burden is more than a matter of simply reducing disease incidence; it also requires understanding the nature of the burden, and the livelihood effects associated with those diseases. We take a specific technical programme to introduce “disease-free” seed to farmers as our case and develop a multiple regression model comprising data from 300 aquaculture farms in order to analyse key drivers of farm outcomes. Data from longer interviews and workshops with farmers, hatchery workers, officials, and suppliers enrich our interpretation of the resulting models and underline the importance of farmer experience and practices in any disease management approach. We demonstrate that the attempt to reduce disease incidence through improved biosecurity not only needs to work with, rather than against, the grain of farmer practices; it also benefits from assessing how disease risks are managed within farming households. Only by understanding the latter, we argue, will the intended effects of reducing both disease and reliance on unsustainable treatments be realised.
2 AQUACULTURE IN BANGLADESH
Aquaculture, or the farming of aquatic organisms, is the world's fastest growing food and protein producing sector, displacing capture fisheries as the principal global source of aquatic protein, accounting for 64% of total fisheries sales (FAO, 2018; Lim & Neo, 2014). In the last few decades, the overwhelming majority (90%) of this growth has been focused in Asia, and has been associated with numerous benefits, including export-led rural development and increased access to dietary protein. To varying extents, expansion has involved environmental degradation and biodiversity loss (particularly involving mangrove wetlands (Hamilton, 2013)), salinisation, downstream water quality issues, and marine stock depletion as a consequence of global feedstock supply (Jayanthi et al., 2018; Tacon et al., 2006). Social issues have included structural inequities across international value chains (Asche et al., 2015); inequalities and injustices in terms of land tenure and labour (Paprocki & Cons, 2014); and uneven access to nutritional benefits within production countries (Islam, 2014). Nevertheless, recent evidence suggests that the benefits of aquaculture may be becoming more evenly shared within and between states than previously thought (Belton et al., 2018), particularly as trade is re-configured from bi-polar, North–South to multi-polar, domestic and increasingly South–South patterns (Bush et al., 2019), and as One Health approaches offer the potential for sector wide improvement (Stentiford et al., 2020).
A persistent issue, in terms of aquaculture's consolidation and expansion, is disease. Within shrimp production, for example, an estimated 40% of global production is lost each year to disease (Lundin, 1996). Partly as a result, and partly as a consequence of an expanding aquaculture industry supply market, pond and animal treatments are on the rise. For example, antimicrobial use is considered to be significant and growing across the sector (Thornber et al., 2019), a trend that is of critical concern in terms of the growing risks and incidence of antimicrobial resistance. Aquatic systems are regarded as particularly favourable for the emergence, persistence, and transmission of resistance traits, and may be a key “gateway” in terms of a growing environmental resistome (Cabello et al., 2016; Gillings, 2017; Taylor et al., 2011; Watts et al., 2017, p. 2). A critical question for the sector is the extent to which productivity can be sustained, while reducing disease burdens and minimising use of antimicrobial treatments.
A key focus for disease management and inappropriate uses of treatments is what Belton et al. (2018) refer to as “squeezed” or “missing middle” farms. Missing middle farms “enjoy none of the benefits of investments in biosecurity or pathogen control characteristic of intensive systems nor the low input/low risk/low output typical of extensive systems” (Little et al., 2018, p. 345). They are nevertheless commercial farms of varying sizes that make up the vast majority of global aquaculture food production, and are part of what Kakkar et al. (2018) refer to as the “invisible cohort” of farmers responsible for a majority of the world's food production. In Asian aquaculture, they include those farms that survived the initial wave of production diseases in the 1990s by maintaining or reverting to less capital intensive and low risk farming methods, but who are currently responding to growing food market demands to increase food production by turning to commercial feed and other farm inputs (Bush et al., 2010; Edwards, 2015; Little et al., 2018). The extent to which these farms can adjust to meet growing food demands will be dependent on the ways in which diseases and production issues are managed in the future. In Bangladesh, where shrimp and prawn production is second only to garments in terms of export value, these “missing middle” farms can be characterised as follows: they are relatively loosely tied into long and complicated value chains (selling produce to faria, or market intermediaries, and then on to depots, and so having little direct contact with retailers or exporters); they are weakly regulated in terms of planning, operational, and veterinary oversight; and they are likely to be in the process of adopting commercial feeding practices and other farm improvements (including greater use of treatments and other inputs; Karim et al., 2012).
In southwest Bangladesh, the modification of tidal, brackish, and freshwater environments using embanked fields (known as gher) to culture shrimp, prawn, and finfish by adding wild or hatchery-raised seed (postlarvae or fry) can be broadly summarised as a form of “salvage accumulation” (Tsing, 2015). This term refers to the conversion of materials and processes, many of which have their own spatial and temporal relations, into capitalist wealth. The majority of this wealth is amassed by lead firms (processors, exporters, retail multipliers, and branded operators) “without controlling the conditions under which commodities are produced” (Tsing, 2015, p. 63). With over 200,000 shrimp and prawn farmers in Bangladesh, the system is relatively resilient at the level of lead firms (who can source product from a large selection of farms), while leaving farmers exposed to market dynamics or other checks to livelihood. A system of pooling produce in relatively informal wholesale markets via faria or market intermediaries means that regulation of standards and produce is difficult to implement (Høg et al., 2019). While governance of the “chain” is attempted in terms of buyer-led and/or NGO-mediated accreditation and regulation of farm practices and food safety procedures (Islam, 2014), the salvage accumulation system and the role of market intermediaries render traceability and compliance challenging.
Within this production system, a number of initiatives have been trialled in recent years as a means to reduce disease and improve productivity (Rahman et al., 2018). They included extension activities aimed at improving pond preparation, better-quality feed regimes and programmes to encourage improved seed use. In particular, delivering “disease-free” postlarvae aimed to disrupt disease transmission between hatcheries, wild populations, and grow-out farms. Shrimp hatcheries had been encouraged to test and accredit their output as disease-free (using polymerase chain reaction [PCR] testing facilities administered by WorldFish and the Department of Fisheries) and, in one case, adopted specified pathogen free (SPF) production methods. Whether these seed were appropriate to the open ponds and conditions on the majority of Bangladeshi farms was a key question. Farmers tended to prefer locally caught (though illegal) wild postlarvae, arguing that they were better acclimatised to the grow-out ponds. Questions around uptake, adaptation, and the extent to which the accredited postlarvae would reduce disease burden were key to this investigation.
Food production is clearly more than a matter of technical specification of inputs. As geographers within a political ecology tradition have demonstrated, the products, labour, landscapes, and environmental effects of food provision are conditioned by values, policies, and priorities that relate to state and non-state actors and institutions (Moragues-Faus & Marsden, 2017; Watts & Peet, 2004). Researchers have utilised Marxist, actor-network, and feminist science studies approaches to trace how threats to food production are generated through the spatial interplay of sites, their economic and ecological situations (Hinchliffe et al., 2016; Wallace & Kock, 2012). In contradistinction to the technical and behavioural slant of many policy documents, these relational approaches emphasise the ways in which human and nonhuman labour (Barua, 2019), land, and production practices (Mitchell, 1996), or everyday geographies of production (Belton & Bush, 2014) are constituted within and occasionally re-constitute broader political cultures and economies of food production (Bair & Mahutga, 2016).
In adopting this broadly relational approach to food production practices, we conceptualised livestock disease and treatments as materially embedded risk practices. These risk practices incorporate both the risks of a disease event and the ways in which diseases are experienced as threats to livelihood. Risk can in this sense refer to the relatively formal probabilities of calculable events like disease outbreaks, as well as to the ways in which people manage or cope with a range of threats (on reflexive and social concepts of risk, see for example Beck, 1992; Douglas, 1992; Liverani et al., 2013; Power, 2014). In most technically and epidemiologically focused work on disease risk, there is a tendency to understand risk as a set of dichotomous events (incidence of disease outbreaks) and, following this, to seek to reduce disease probabilities as a means to improve production and drive down the need for unsustainable livestock treatments. Commonly, in food and farming, this focus on the risk of disease drives solutions that involve enhanced pathogen control, food sector modernisation, and biosecurity (Hinchliffe et al., 2016). Moreover, there is a tendency to assume that the “disease-free paradigm” that is common (and not altogether unproblematic) within the global North (Wallace, 2009, 2016) can be translated to food production settings within Lower and Middle Income Coutnries (LMICs) and the global South. In following this colonial logic, a number of issues are downplayed. The specific and often rapidly changing conditions of production within tropical and sub-tropical environments tend to be ignored (Cole & Desphande, 2019), as are the ways in which risks are experienced within socio-economic and ecological situations. As we will demonstrate, this tendency to assume a one-size-fits-all approach to disease risks (Kakkar et al., 2018) is not only inappropriate to the setting but may also miss the opportunities that present themselves once a “risk practices approach” is taken.
3 METHODS AND MATERIALS
In order to investigate farmer risk practices, we developed a multi-method approach (Brewer & Hunter, 1989). We generated a survey involving 300 farmers in order to investigate seed use, its effects on production and farm outcomes (including productivity, mortality, disease, and profits). We interviewed hatchery owners (n = 25), farmers and officials (n = 10), and ran three workshops (n = 120) in which all relevant groups produced qualitative models of disease, treatments, and antimicrobial resistance risk maps. Initial interviews and farm visits were carried out prior to the survey in order to develop a working premise on key issues and production types. The workshops were held after the survey and were used to relay and debate interim results. The final farmer interviews (n = 46) were carried out to develop more understanding of issues arising from the survey, especially in this case the role and structure of finance (and micro credit) in shaping farmer decisions on seed purchase and use. In this paper, the focus is on reporting the analysis of the survey, utilising the qualitative data to aid interpretation of those results.
A WorldFish census of 1,394 farms was used to structure the survey farm sample into categories based on previously expressed seed preferences. Farms were subsequently selected randomly from each category in the Bagherhat, Khulna, and Satkhira districts of the southwest coastal region (see Figure 1), the main area for shrimp and prawn production in Bangladesh, and covering a range of environmental conditions. Farms close to the mangrove wetlands (the Sundarbans) were saline throughout the year and predominantly raised black tiger shrimp (Peneaus monodon, BD Bagda), while further inland farmers also cultured freshwater tiger prawn (Macrobrachium rosenbergii, BD Golda). Most farmers beyond the highly salinised areas integrated aquaculture with rice agriculture.
The survey was designed to obtain information on farm characteristics, pond management and outcomes, translated into Bangla, uploaded onto a digital field survey tool,1 and piloted on 20 farms before its roll out by the team with trained NGO field assistance.2 Lead farmers (those recognised in their community as experienced and knowledgeable) assisted in locating farmers on the sample list. Statistical analyses of key dependent variables (stock mortality, productivity, morbidity, and profitability) involved initial selection of 50 explanatory variables that related to environmental, socio-economic, and management variations across the sample. The number of variables was reduced using univariate regression, excluding those with higher p values (or probability of chance occurrence, in this case the cut-off used was p > 0.3). Multivariate analyses were carried out for the four dependent variables using the remaining variables and employing general linear models with designated probability distributions. A maximal model for each response variable was created by sequentially removing explanatory variables from the model in a stepwise fashion using log likelihood ratio tests, with a designated significance of p < 0.05, until the most parsimonious final model was found for each dependent variable. Coefficients (productivity and profit) and odds ratios (mortality and disease) were calculated for explanatory variables remaining in the final models. These remaining variables were interpreted as indicators of the main drivers shaping farm outcomes, with interpretations triangulated through use of the contextual knowledge generated in the qualitative inquiries conducted before and after the survey.
4 RESULTS
The majority of respondents were male,3 educated to at least school level, with two thirds owning and the remainder renting or sharing their ponds. As well as shrimp and prawn, all the farmers used their ponds for producing lower value crops, including finfish, rice, mud crabs, and livestock, depending on pond and soil salinity. Pond water was sourced from the river, tidal waters, or from canals and shared dikes, with variable levels of water exchange depending on location relative to those sources. Ponds ranged in size, with some extensive and large gher in the southern saline districts, which were over 1 ha in area, and much smaller, freshwater ponds in the north of the sample area (<0.2 ha). Most were shallow, with a mean depth of 0.8 m. The majority of the farms could be classified as “small” (0.2–1.0 ha, 173 farms, 56%) or “medium” (1–3 ha, 79 farms, 25%; Jahan et al., 2015). The mean stocking density4 for prawn and shrimp was 1.8 animals per square metre. Only 10% of shrimp farmers followed a single stocking protocol (stocking once in a production cycle). This may be an over-representation in terms of the general picture in Bangladesh, as we sampled purposively to include farmers who were using certified seed, and who would have been encouraged to observe best practice. The remaining farmers re-stocked ponds more frequently, generally several times in a cycle. Despite our attempts to structure the sample by seed preference, almost all farms used a mix of postlarvae types, with only a handful exclusively using tested or SPF seed. Farmers explained their failure to access their preferred seed in terms of limited availability and price differentials (both of which were pronounced for SPF and wild seed; see Figure 2).
Annual mean productivity was 374 kg/ha (SD 311) for combined shrimp and prawn, though half of the farms struggled to produce more than 300 kg/ha. Production was augmented on many farms with a variety of species of finfish, which raised total mean pond productivity to over 1,000 kg/ha. Health problems were common, with 85% of shrimp farmers and 50% of prawn farmers reporting disease.5 Farmers commonly listed other hazards (cyclones, theft), water quality issues (fluctuations of temperature and salinity), and water management problems (water exchange and treatment). The impact was large, as the calculated mean mortality rate was 80% shrimp and 65% prawn lost prior to harvest.
Two thirds of farmers used commercial feeds to supplement homemade and natural food sources. Pre-stocking and post-stocking treatments were applied on most farms, with a majority of farmers using lime and a minority using disinfectants, pesticides, and piscicides prior to stocking. Locally produced fertilisers (including cow dung, urea, and mustard oil cake) as well as triple superphosphate were used to stimulate phytoplankton growth. Treatments added after stocking included lime, probiotics, and occasionally, antibiotics (oxytetracycline, ciprofloxacin, and florfenicol). For sources of advice on treatments and disease management practices, lead farmers and farm supply shops were key in helping farmers to identify health issues, diseases, and select treatments. Supply shops were often run as concessions, selling branded products and medicines to farmers with minimal regulatory or veterinary oversight. A common sequence was for farmers to report a health problem to a farm supply shop and for a product to be sold (often on credit) without farmers necessarily being told or being sure what the product contained. In workshops and interviews with farmers and supply shop owners, participants reported that treatments and pharmaceutical sales were increasing, with farm shops under pressure from feed and pharmaceutical company agents to recommend products. While records of antibiotic residues had only rarely been picked up by regulatory laboratories in importing countries, farmers and shop owners told us that antibiotics were widely available and would be used, albeit infrequently, as “a last resort” in an attempt to rescue stock.
Farms made on average 385,000 Bangladeshi taka (BDT) gross, or 260,000 BDT (£2,470, US$3,100) net per pond in 2017/18 (approximately 492,000 BDT per ha (£4,700, US$5,800)) from shrimp and/or prawn. The majority of farms achieved net incomes of less than 200,000 BDT, with a handful of farms recording losses. On average, farms earned 65% of their income from shrimp and prawn cash crops, with additional income sources predominantly from rice, finfish, and other sources of employment. The largest farm expenditure item was prawn and/or shrimp seed, which on average accounted for over 50% of running costs.
Table 1 presents the key explanatory variables (excluding location) driving mortality, productivity, disease and profits derived from maximal models. Green rows indicate a positive effect, orange a negative effect of the variable. Deviance is an indication of the variable’s overall fit to the maximal model, while the odd’s ratios (Mortality and Disease) and Coefficients (Productivity and Profit) indicate the direction and size of the variable’s effects.
| Farm outcomes | Key drivers | Deviance of maximal model | Odds ratio or coefficient (95% CI) |
|---|---|---|---|
| Mortality |
Stocking frequency Salinity problems (reported) Use of untested shrimp seed |
Shrimp 15, prawn 34 Prawn 16 Shrimp 10 |
1.05, 1.32 2.4 1.4 |
| Productivity |
Stocking frequency Water exchange Pre-stocking lime (kg/1,000 m2) |
Shrimp 38, prawn 113 Prawn 18 Prawn 11 |
16, 75 163 3.1 |
|
Pond depth Organic fertiliser |
Prawn 11 Shrimp 7 |
-139 -84.1 |
|
| Disease |
Commercial feed use Use of SPF postlarvae |
10 10 |
0.5 0.1 |
| (white spot disease in shrimp) |
Hired labour Salinity problems Use of untested shrimp seed Water source (river, tide, canal) Use of PCR tested seed |
22 17 12 11 10 |
5.8 3.7 4.6 4 1.8 |
| Farm profit | Stocking density | 34 | 23.1 |
|
Rent payments Stocking frequency |
7 4 |
-20.7 -1.6 |
- Abbreviations: PCR, polymerase chain reaction; SPF, specified pathogen free.
While location (at sub-district level) was a key driver of farm outcomes, several farm management and environmental variables were significant in the final models (Table 1). Negative farm outcomes included increased mortalities associated with frequent stocking, excess salinity in prawn production, and the use of untested (hatchery and wild-caught) seed. Disease was reported more often on those farms that used hired labour, had salinity problems, and used untested and PCR tested seed. Conversely, farm outcomes improved with commercial feed use, using SPF postlarvae, and, for prawn aquaculture, ensuring water exchange and liming prior to stocking. In a number of cases there were clear examples where positive effects on one farm outcome were offset with negative effects on others. So, while water exchange was beneficial for prawn (as a means to regulate salinity), it was associated with increased disease transmission if the water source was open and communal. Similarly, addition of organic fertiliser had a marginal effect on prawn production, but was associated with a reduction in shrimp productivity.
The relationship between stocking density, stocking frequency, and farm outcomes was particularly important in understanding farmer risk practices. Stocking practices were significant for the mortality, productivity, and profitability multivariate models, though not so for disease (Figure 3). Mortality was higher, at any given density, on those farms that stocked ponds more frequently than the mean, with the interaction of stocking density and frequency accounting for the highest proportion of model fit after location. The productivity models were also associated with this stocking variable, which accounted for the highest proportion of the model goodness of fit. In this case, though, better outcomes were associated with increased frequency of stocking (see Figure 3b,c). While this variable was not significant for any of the disease models, it was significant in reducing profitability, though with a lower proportion of model deviance than for mortality and productivity. Interpreting these relationships will be developed in the discussion.
5 DISCUSSION
The majority of “missing middle” farms in Bangladesh were operating on tight biological and economic margins. Most farmers managed to produce small profits across a range of pond types and environments, at the same time as incurring and suffering high rates of animal morbidity and mortality (Hoque et al., 2017). Environmental conditions clearly contributed to farm performance, with salinity levels a key driver of culture species and their relative success (Hasan et al., 2020). Seasonal and sometimes diurnal fluctuations in pond temperature and salinity would increase animal stress and were more pronounced in the relatively shallow ponds used by the farmers. Pond depth was often a compromise between the requirements of the organisms farmed, the conditions needed to grow rice, and the lack of available investment to improve embankment height.
Within these physical and economic constraints, the evidence from the survey and analysis suggests that farms benefited from improved seed, additional feed, and better management techniques (Ali et al., 2018). Untested seed sources were associated with increased mortality, while use of SPF and tested seed reduced incidence of shrimp diseases compared with untested sources. These differences in seed performance may have been a result of variation in postlarvae quality, and a result of closer NGO involvement with those farmers adopting the SPF seed (though notably NGOs' advice and other expected management variables did not appear significant in the final maximal models). Reducing physical movements associated with shared labour, water exchange, and stocking were all significant in terms of reducing shrimp disease incidence and mortality (Karim et al., 2012; Lightner, 2005). Sharing labour and equipment, utilising communal water resources, and frequent stocking could all be expected to increase the risk of transmission of biofilms, fomites, and animals, and, in many cases, relatively simple improvements in pond management and biosecurity (for ponds, people, and equipment) would be expected to moderate this effect. In this sense, the results support continuing farm extension activities and improvements in biosecurity as means to reduce disease incidence in grow-out farms.
Beyond this relatively straightforward message regarding improvements in seed, feed, and management as means to improve margins and animal health, a more intricate issue arose regarding farmer stocking practices. Those farmers who re-stocked more frequently had higher mortality figures. This might be explained as a simple consequence of farmers replenishing lost stock. It may also reveal a risk with re-stocking, which would be expected to increase mortality risk as a result of higher stock densities, disease transmission, and predation (especially cannibalistic predation of juveniles). A single stocking system would therefore seem preferable as a means to improve stock management and so reduce mortality rates. It was nevertheless clear that those farmers who re-stocked frequently also managed to safeguard and even improve the productivity of the pond (Figure 3b,c). Indeed, in workshops, it became clear that farmers routinely stocked their ponds even without experiencing losses of stock. Continual stocking, often with seed from mixed provenance, seemed to ensure a reasonable harvest, over an extended period, even if this practice was suboptimal in terms of disease, profit, and wastage of postlarvae (or high mortality). The explanation for this relates in part to the dynamics of ponds and markets. In an open pond and salvage accumulation system, it may make sense to stock more than once in order to achieve one or more of the following: adapt to changing water conditions (with reductions in salinity and temperature after the monsoon for example making it more feasible to culture prawn); reduce the likelihood of disease or environmental changes wiping out an entire season's crop; and respond to fluctuations in wholesale prices of both seed and produce (staggering stocking and subsequent harvests allowed farmers to avoid peak stocking prices for seed and seasonal surpluses of produce). In other words, farmers who stocked more frequently were able to generate livelihood across the growing season, spreading their expenditure and sales, while becoming more able to cope with frequent hazards that they and their livestock experienced. Conversely, single stocking was associated with high initial costs and the risk that an entire crop could be lost if pond conditions changed or disease struck. In sum, while disease incidence might be reduced with a single stocking regimen, it would also raise the stakes of a disease or other event, placing livelihoods at greater risk.
Single stocking and a single harvest were also financially difficult for most farmers. Farmers had limited funds to make significant changes to their ponds or to purchase all their seed in one go. In our follow-up interviews (n = 46), all but six farmers depended on micro-finance in the form of small loans that needed to be paid back to banks or NGOs through weekly instalments (Bateman, 2010). This financial situation meant that most farmers had little choice but to purchase seed at relatively frequent intervals and to make sure they had enough regular income by harvesting stock over an extended period in order to service their loans. In other words, while overall profitability may be somewhat suboptimal as a result of multiple stocking practices, the strictures of financial availability dictated otherwise.
The relevance of these findings in terms of adoption of disease-free seed and associated practices, and the consequences in terms of use of pond and animal treatments, relate to these contextual and practical experiences of risk. For farmers, working in environments that were open to a number of diseases and other challenges, purchasing disease-free seed was of limited value as it removed only one of any number of disease transmission pathways. Farmers were well aware of the range of threats to pond and animal health, and viewed hatchery seed with some scepticism (something seemingly confirmed by the experience of disease incidence with tested postlarvae). A more relevant issue for farmers was the robustness or health status of the seed. They asked for seed that could survive in ponds that were subject to daily and seasonal variations in salinity, temperature, and oxygen levels. This meant that wild caught seed, despite being illegal, was preferred by many farmers as there was an assumption that the postlarvae and resulting adult animals were adapted to local conditions.
In discussion with farmers, there was a clear sense that changes to production methods that left them more exposed to livelihood loss (i.e., single stocking and single species systems) were also more likely to lead to what they described as indiscriminate use of medicines. Farmers said that they would turn to the “desperate measures” of antibiotic use even though they knew of the dangers and that the treatments often failed (Denyer Willis & Chandler, 2019). One group agreed that “if everything dies, we have nothing.” The interesting issue is that for many farmers, this last resort was less likely in ponds that were used for multiple crops and species. Encouraging farmers to adopt disease-free seed, as well as shifting farming practice in order to maintain relative biosecurity (and disease freedom) had the paradoxical effect of changing livelihood risks. The latter would be out of step with available finance, market relations, and ecological conditions. In other words, reducing disease occurrence while increasing farmer exposure to the consequences of those risks was a problem. All of the tactics and coping strategies that had been developed and used in the last few decades to ensure livelihood in an open and frequently challenged landscape would be undermined.
The implication for antibiotic use was that, paradoxically, farmers were less likely to use antibiotics in a more frequently diseased system. As farmers shouldered all the risks of production within a salvage accumulation process, the stress of losing a crop could lead to emergency measures. Best practice management could in this sense intensify the market and economic relations that lead to suboptimal outcomes. Pursuing “disease-free” farming within this socio-economic arrangement positioned farmers into relations of production which made acts of desperation more rather than less likely. In contrast, and without underestimating the difficulties and room for improvement in food production practices, there is clear potential of polycultures and agro-ecological forms of farming to reduce medicalisation by modulating risk experience and providing “insurance” in precarious physical and economic environments (Ali et al., 2018; Bush et al., 2010; Hanh & Boonstra, 2018), as well as generating environmental and nutritional gains (Shepon et al., 2020). Rather than being seen as barriers to disease management, these practices may be part of a raft of approaches to producing food in ways that are resilient and require low treatment protocols.
6 CONCLUSIONS
Changing food production practice is an adaptive challenge (Heifetz, 1994). Technologies like improved or disease-free seed need to be fitted to the social, economic, and ecological conditions of production. Rather than barriers to adoption of new technology, these conditions need to be seen as the bases for improved practices. Indeed, in this study we have demonstrated, first, that an apparently beneficial technology was somewhat ill-fitted to the farms. Farmers were more interested in the health and resilience of their livestock rather than the more abstract and ultimately less meaningful “disease-free” label. Second, and just as importantly, disease-free seed demanded a shift in farmer practices in order to be demonstrably beneficial in the field. As we have demonstrated in our analysis of farm outcomes, the difficulty was that this shift to fewer stocking events involved a concentration of risk and financial investment that was unfamiliar, unavailable, and unattractive to the farmers.
We have introduced a broad definition of risk (or risk in practice) to demonstrate how, for farmers, disease is more than a matter of dichotomous events (presence or absence of pathogens) to which approximate probabilities are attached. Rather, it is tied up with livelihood practices that are themselves shaped by precarious value chain and environmental relations. The implication of this work is that a technical version of best practice management may reinforce a tendency to increasingly rely on medical and chemical inputs. Rather than treat farming practice as a set of technical and knowledge deficiencies that need to be overcome in order to optimise or modernise production, we have worked from the risk practices and experiences of farmers and their livelihoods to identify how incomes are secured and how such practices can reduce reliance on treatments.
As Little and colleagues note, a common strategy in fresh and brackish water shrimp and prawn aquaculture systems across several Asian settings has been to reduce risk by de-intensifying and diversifying crops (Little et al., 2018, p. 343). Much of this adjustment has been “needs must” and is likely to change rapidly as greater demands are made on the food system. A key question for the invisible cohort of missing middle farmers who are crucial to future food production will be the extent to which effective means of maintaining income and livelihood can be integrated with improvements that can yield greater amounts of better-quality food. How, in other words, can the process of enhanced sustainable production be understood as conditional upon a “mixture of modernisms” that incorporate a range of knowledges and other practices? In this, the role of vernacular practices in systems that do not conform to a disease-free paradigm will be key. Similarly, the exposure of squeezed middle farmers to an inequitable distribution of livelihood risks within domestic and global value chains has a tendency to drive forms of production that are unsustainable. If a key reason for treatment use is the desperate need to make a living within an intrinsically precarious system, then finding ecological as well as socio-economic ways to reduce rather than intensify risk-in-practice is a priority.
ACKNOWLEDGEMENTS
The research was made possible on receipt of an ESRC grant under the cross UK Research Council “Tacking Antimicrobial Resistance” programme (“Production without medicalisation: a pilot intervention in global protein production”; ref ES/P004008/1). MMR is supported by the CGIAR Research Program on Fish Agri-food Systems (FISH). Himangshu Biswas and Siddhwartha Kumar Basak and the WorldFish Khulna team assisted in the field with interviews. Suhada Akter led the Arban team of infopeddlars and carried out the survey. The authors would like to thank the expert referees for their insightful comments on a previous draft of this paper. All errors remain the authors' responsibility.
ENDNOTES
- 1 The survey was delivered using a digital survey form designed by ARBAN (Activity for Reformation of Basic Needs – a development NGO) technical specialists using Survey Lite app technology that offers real-time interaction with end users through instant synchronisation to the Survey Lite survey, retrievable via the Splash Portal interface.
- 2 The survey was carried out by five “infopedlars,” an employment scheme that offered work experience opportunities to female students and young women in Bangladesh. ARBAN and the research team trained the infopedlars in survey delivery and software application, as well as farming and disease issues.
- 3 Despite active attempts to survey more women farmers, only 5% of the final sample were female, reflecting the gendered character of aquaculture production in Bangladesh. Many smaller sized farms were nevertheless run as family concerns, with women contributing to and sometimes leading farm tasks.
- 4 To estimate adjusted stocking density, the reported number of postlarvae (PL) added to the pond was divided by the reported frequency of stocking events on each farm.
- 5 Diseases were identified using photographs of clinical signs, and not subject to laboratory confirmation. Most disease in shrimp was “unknown” and white spot. In prawn dhari jhara and broken antennae were most common. It is probable that many of the reported diseases were signs of the same or related problems and some may relate to physical stress as well as disease.
- 6 Analyses were performed for a range of reported diseases, but for simplicity, we limit the results and discussion in this paper to white spot, which was the most frequently reported disease.
Open Research
DATA AVAILABILITY STATEMENT
Survey and interview data from this research are available http://reshare.ukdataservice.ac.uk/853865/, and http://reshare.ukdataservice.ac.uk/853866/
