Fishing for subsistence constitutes a livelihood safety net for populations dependent on aquatic foods around the world – Nature Food

Approach and concepts used to characterize the roles that small-scale fisheries play in supporting livelihoods as safety nets

To evaluate the strategies engaged by individuals and households to reduce food and nutrition insecurity and alleviate poverty through small-scale fishery livelihoods—a term used here to refer to the capabilities, assets and activities required for a means of living or adequate stocks and flows of food and cash to meet basic needs^23,24—two separate measures based on labour statistics were assessed: (1) employment, defined as all persons of working age who, during a short reference period (typically the week before the interview), were engaged in any activity to produce goods or services provided for pay or profit, including both part- and full-time employment to capture seasonal variation (with interviews conducted continuously throughout the year in almost all surveys, including both employed persons ‘at work’, that is, who worked in a job for at least one hour during the reference period, and employed persons ‘not at work’ due to temporary absence from a job or to working-time arrangements, such as shifts in work, flexitime and compensatory leave for overtime); and (2) subsistence, defined as working for one’s own consumption: workers who produce goods or services that are predominantly consumed by their own household, with no transaction occurring in the marketplace²⁵, potentially including pre- and post-harvest activity in fisheries (and so referred to here as ‘subsistence fishing activity’, inclusive of any pre- and post-harvest activity). With these definitions, the term workers is used here to refer both to people who work in fisheries for pay (employment) and to those who work for their own consumption (subsistence), consistent with definitions from scholars of subsistence as sustaining a basic level of livelihood²⁶. Note that these definitions may lead to underestimations of subsistence fishing activity, as many who we classify as employed because they sell a majority of their product may keep some of the catch for their own consumption.

Following Béné⁵, the safety net function of small-scale fisheries has typically been defined as the role they play in providing food production alternatives (that is, subsistence) and additional income opportunities for vulnerable households with low savings, as needed in times of individual or collective hardship, crisis or stress (for example, to cope with a climate-related shock or reduced agricultural production). For this study we broaden the definition of safety net to include the separate but similar poverty alleviation functions of absorbing excess unskilled labour and adding to the portfolio of livelihood options for vulnerable households. While the function has typically been defined more narrowly to focus on buffering vulnerable households in times of crisis⁵, parsing between these functions is unrealistic even at the household level and ultimately unnecessary as all contribute to the role of small-scale fisheries in preventing poverty (welfare mechanism).

As for other primary economic activities such as agriculture, a significant portion of the employment and subsistence in small-scale fisheries is irregular and seasonal, located in rural areas and conducted in parallel with other non-fisheries activities to provide additional income or nutritional support, particularly in inland fisheries to fill ‘livelihood gaps’ during the agriculture off seasons^27,28,29. Acknowledging different usages²⁶, by definition, subsistence fishing activity often provides supplementary or alternative food for households in times of shocks or limited employment³⁰ or forms part of livelihood strategies including multiple occupations—for example, a majority of subsistence fishing households in coastal South Africa were found to be involved in multiple occupations and food insecure³¹; similarly in Savannaket, Laos, occupational diversity was significantly and positively associated with the probability of fishing, largely for subsistence³². While certainly not limited to subsistence fishing activity, the extent and frequency of this activity worldwide, linked with the apparent fish consumption and the estimated nutritional value derived from this consumption, can provide a coarse quantitative proxy indicator of the minimum use of small-scale fisheries as a safety net to help alleviate poverty and malnutrition within the context of total livelihoods supported by these fisheries.

While we consider the presence and frequency of subsistence fishing activity as a coarse proxy for the minimum use of small-scale fisheries as a livelihood safety net, and acknowledging that not all subsistence fishing activity may serve the function of a safety net, the measure may still understate the importance of this function. Beyond subsistence, many who work in small-scale fisheries for employment may also do so in ways that could be characterized as a safety net, engaging in the activity as part of pluralist strategies or in response to shocks. Similarly, some who are counted as employed because they sell a majority of their product may still keep some of the catch for their own consumption.

Sources of data

For this study, three different types of large-scale and standardized household survey instrument provided a previously unused source of data on small-scale fisheries livelihoods: population censuses, labour force surveys and household income and expenditure surveys conducted by governments’ national statistics agencies (Supplementary Information and https://www.ilo.org/surveyLib/index.php/home). Survey data from at least one of these instruments were available for 78 countries over the period 2008 to 2018 (with 54 of the 78 surveys conducted in 2014 or subsequently), representing almost 79% of the world population in 2016 (labour force surveys for 33 countries, household income and expenditure surveys employment modules for 44 countries and a PC for one country). Data collected in the surveys were classified according to common standards that allowed for cross-country comparison based on the type of activity undertaken as defined by the International Classification of Economic Activity and the International Standard Classification of Occupations standards. Questions on employment were asked about the activity conducted within the previous week as the reference period, while for subsistence the reference period was typically once within the previous year, but in both cases interviews were conducted throughout the year to reflect seasonal variation and generate annual average participation (additionally, many national surveys in Asia and Latin America collect quarterly data via panel samples or sub-samples interviewed four times during the year). To be representative, the surveys aimed for national coverage with samples randomly selected from a listing of households (‘master sampling frame’) typically created on the basis of the most recent population census and stratified by geography (‘enumeration areas’). These data are generally available to the public upon request from national statistical offices.

Data collection, organization and gap filling

The questions included in the household surveys provided the data for the analysis. For questions related to employment, respondents were first asked (yes/no) if they are employed (that is, if they ‘did any work for a wage, salary or commission’, ‘ran a family (farm or non-farm) business’, ‘helped in the household (farm or non-farm) business or ‘had a job or business s/he will return to’ for at least one hour during the reference period, typically the previous week), and for those who answer yes, a series of subsequent questions were asked, including their occupation and economic sector of employment. For questions related to subsistence activity, respondents identified as employed were asked ‘are the products obtained from this activity for sale/exchange or for family use’ and were able to select one of the four following options: (1) ‘only for sale/barter’, (2) ‘mainly for sale/barter’, (3) ‘mainly for family use’ or (4) ‘only for family use’ (with responses indicating (3) or (4) categorized here as engaged in subsistence activities).

Responses to these questions were categorized according to the type of activity performed by gender. Those persons engaged in activities related to fisheries were further classified based on their type of work, which allowed for identification of those employed in fisheries (and subsequently in small-scale fisheries) and those engaged in fisheries activities for subsistence, in both cases by gender. In addition to the category specifically labelled for ‘fishing activities’ and related to the harvesting stage, other categories of activity were cross checked and included to identify employment in pre- and post-harvest activities as different stages of the production process (Supplementary Table 5; for example, ‘fish processing’, ‘wholesale of fishery products’ and so on).

Identification of employment in small-scale fisheries

In the absence of a universal definition of small-scale fisheries³, we follow the practice of the International Classification of Status in Employment³³ to characterize operations in different sectors as small-scale based on employment classifications. Following this practice, employment in fisheries was classified in the surveys as either paid or self-employed, with the latter sub-divided into ‘employers’, ‘own-account workers’ and ‘contributing family workers’. To disaggregate this employment between small- and large-scale fisheries, those persons classified as ‘own-account workers’ and ‘contributing family workers’ were assumed to participate in small-scale fisheries. Of the remainder, those persons employed in enterprises whose total number of workers was less than two-thirds of the 90th percentile number of workers engaged in all fisheries-related enterprises within a given country were assumed to participate in small-scale fisheries. This same operational criteria—based on the context-dependent threshold and the status in employment—was also applied to those who engage in pre- and post-harvest activities connected to fisheries as a proxy for operations linked to small-scale harvesting, acknowledging that in some cases, large enterprises may process fish caught by small-scale harvesters and vice versa. The cut-off of two-thirds of 90th percentile was meant to capture the average number of workers employed in the top 10% of the largest business operations in fisheries. The choice of the 90th percentile, rather than a crude absolute threshold that is applied without distinction across countries, can better capture differences in the size of business across countries. In this regard, depending on the country-specific distribution of the number of employees and its corresponding relative average number of workers in the largest 10% of fishing operations, small-scale fishing operations in one country can be classified as large-scale in another country. The average number of workers in the largest 10% of fishing operations (90th percentile) was further adjusted by two-thirds to also include in large-scale fisheries those employees working in the intersection of small- and large-scale fishing business.

Identification of persons engaged in subsistence fishing activity

Data on subsistence fishing activity (defined as working for ‘own consumption’, including potentially pre- and post-harvest activity) were available in 32 of the 78 national surveys used, with those engaged in subsistence fishing activity in these countries equivalent to 81% of the total number of people employed in fisheries in the same countries. For the majority of these surveys, data were only available to estimate persons engaged in subsistence fishing activity at a frequency of at least once during the previous year, based on the reference period. For nine surveys (representing 25% of the total number of persons estimated globally to be engaged in subsistence fishing activity), the data permits to estimate subsistence fishing activity at a higher frequency of at least one hour during the previous month, or in some cases during the previous week. Additionally, the surveys in Cambodia and Laos (0.96 million and 1.06 million estimated people participating in subsistence fishing activity respectively) were conducted only once per year but confirmed that the activity occurred within the previous week (that is, the recall period). Finally, the surveys from seven countries (representing 23% of the estimated global total of persons engaged in subsistence fishing activity) provide data showing that on average those engaged in subsistence fishing activity spent 4.2 hours doing so per week reported, suggesting a significant investment of time in the activity (and justifying an assumption that the global total of persons estimated to be engaged in subsistence fishing activity did so more frequently than once per year).

Organization of the data and estimates to fill gaps

The microdata from the 78 national household-based surveys were processed, harmonized and reported at the national level for each country. To allow for comparison, the results in the 78 national datasets were adjusted to the study year 2016 (the study year chosen for the Illuminating Hidden Harvests assessment, based on more recent middle year of the period for which data was collected: 2013–2018) by taking the ratio of employment for the survey to the International Labour Organization (ILO) data on the total population employed in agriculture, forestry and fishery in that country and applying it to the ILO data for that population in 2016 (Supplementary Box 3). Where data were missing within the 78 national datasets, they were estimated by calculating and applying ratios from the mean of available data from other countries within the same geographic archetype. Geographic archetypes were specified at the lowest possible regional grouping, according to regional groupings used by the ILO. The most common gaps and the ratios applied to fill them are described in the Supplementary Information and for missing gender-disaggregated data Supplementary Box 4.

Global extrapolation from the 78 national datasets

The results from the 78 national datasets were extrapolated to the regional level using the geographic archetypes from ILO and subsequently to the global level. To correct for non-response bias in countries not included in the national datasets (which were selected based on the availability of information and not randomly), a weighted regression analysis based on independent variables considered as predictors was used, following recommendations by the ILO³⁴. Weights of the different predictor variables were calculated as the inverse probability of selection (or inverse propensity score) to account for differences between the 78 countries for which data were collected and the world’s remaining countries to which the results were extrapolated (Supplementary Table 8). Using these weights to correct for non-response bias, the weighted regression analysis was conducted, essentially generating estimates based on assumed relationships between employment, subsistence and livelihood dependency variables and a set of predictor variables. The predictor variables used were chosen based on (1) strong correlation with the outcome variables (measured by the R squared) (Supplementary Table 9) and (2) availability worldwide. These included (1); Employment in agriculture, forestry and fishery; (2) Employment in industry and employment in services; (3) Total population; (4) gross domestic product (GDP) per capita (purchasing power parity); (5) GDP growth; (6) Value added in agriculture, forestry and fisheries. For marine small-scale fisheries, the additional predictor ‘Length of coastline (km)’ was included, whereas for inland small-scale fisheries, it was ‘Area of inland water bodies’ (Supplementary Table 9 provides predictor variables used).

As a cross check, the resulting estimates were compared to: (1) data compiled on small- and large-scale fisheries employment in 58 national-level case studies and (2) government responses to a survey conducted by the United Nations Food and Agriculture Organization (FAO) with all member states during 2018 and 2019, conducted as part of the Illuminating Hidden Harvests global assessment of small-scale fisheries (Supplementary Fig. 7). Additionally, results were compared to publicly available datasets on aggregated employment in fisheries: (1) the International Labour Organization labor statistics database (ILOSTAT) data on employment in either fishing or aquaculture (aggregated) and (2) FAO data on employment in fisheries (aggregated between small and large-scale fisheries) (Supplementary Fig. 8). For countries where significant differences emerged, experts were consulted to help provide further explanations and eventually adjust estimates from the weighted regression analysis. A final check was comparison to a dataset of the global marine fishing effort, disaggregated between small and large-scale fishing, to identify any countries where zero small- or large-scale fishing effort occurred, but the estimates from surveys suggested a non-zero employment in fish harvesting³⁵. Finally, the results are comparable to previous studies, such as World Bank¹² (Supplementary Information provides more detail on the process of cross checks).

Selection and use of national datasets for more in-depth case studies on the safety net function of small-scale fisheries

National datasets were used for more in-depth analysis of the safety net function where information was available. In 14 of the datasets (Bangladesh, Brazil, Cambodia, Chile, Egypt, the Gambia, India, Indonesia, Mexico, Peru, Senegal, Sierra Leone, Tunisia and Yemen), sufficient information was available to assess the role of small-scale fisheries employment in total employment at a sub-national level. Additionally, in seven datasets, more detailed information was available on subsistence fishing activity (Fig. 4), including on how these workers allocated their labour (Kiribati, Lao PDR, Senegal, Cambodia, South Africa, Bangladesh and Myanmar). For the ten countries shown in Fig. 3 to have the highest ratio of the number of people engaging in subsistence fishing at some point during the year to the number of people employed part or full time in small-scale fisheries (Lao PDR, Kiribati, Bangladesh, Cambodia, Myanmar, Senegal, Sierra Leone, South Africa, Indonesia, the Gambia), this list was calculated based on national datasets with available observations, rather than the global dataset including estimates. The list of the top ten countries worldwide would change if it included the countries with data based on estimates.

Estimates of the nutritional safety net function of small-scale fisheries

From the Illuminating Hidden Harvests assessment, data were compiled on a wide range of indicators of the contributions of small-scale fisheries to society, including the volume of catch landed, by species, in 58 country case studies. For each of these country case studies, researchers either compiled available data on the use of the catch (the percentage of the volume landed that was used for commercial sale and destined for human consumption domestically, commercial export, subsistence or for non-human consumption), by species, or used expert judgement to estimate¹⁵. For 14 of the 58 countries, there were sufficient data on catch use to estimate the average total volume of catch for subsistence by species for the period from 2013 through 2017 (multiplying the volume of catch landed for each species by the percentage reported or estimated by experts to be used for subsistence).

For each of the 14 countries, the estimated volume of subsistence catch was assumed to be divided evenly between the persons engaged in small-scale fisheries and their household members to generate an estimated per capita consumption of subsistence catch by species (mg per day) for a total number of persons. Drawing upon nutrient data compiled from peer-reviewed publications and existing databases for over 500 marine and inland fish species, the Illuminating Hidden Harvests assessment developed a model for predicting the species’ composition of six nutrients commonly deficient and driving poor health outcomes: iron, zinc, selenium, calcium, vitamin A and omega-3 fatty acids¹⁵. Using this model, the volume of subsistence catch by species in each of the 14 countries was converted into volumes of each of the six nutrients assumed to be consumed by the population of subsistence fishers and their households. A significant volume of the catch in each of the 14 countries was not identified by species (‘not elsewhere included’) (Supplementary Table 11), and the nutrient composition imputed based on the mean nutrient values of species caught by marine or inland small-scale fisheries within the same country (Supplementary Information). Finally, the volume of the six nutrients from the subsistence catch consumed by a given population in each of the 14 countries was compared to the minimum recommended nutrition intake (RNI) for each.

Statistics and reproducibility

The study was designed as an analysis of existing data from 78 standardized national household-based surveys, conducted by governments’ national statistics agencies from 2008 to 2015. No available surveys and data were excluded from the analysis. To fill in any gaps in data in surveys and to estimate results for any countries without surveys, we used standard econometric modelling (weighted regression analysis). For the case studies, the subsistence catch volume data were used from country case studies conducted in the global Illuminating Hidden Harvests study, and a predictive modelling approach developed for that study was used to predict the nutrient composition of the catch. Data gaps were filled using the mean nutrient values of species belonging to the same country, same sector or detailed functional group.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.