We compiled marine fisheries employment information in 144 maritime countries. Data on global fisheries employment were sourced from technical reports published by institutions such as the FAO and the ILO, peer-reviewed publications, as well as fisheries and agriculture departments of individual countries. We segregated countries by their UN Human Development Index status (High, Medium and Low) and by geographical location according to FAO’s six regions (1-Europe, 2-Asia, 3-Africa, 4-South America, 5-Oceania and 6-North and Central America). The Low HDI category was represented by countries from Africa only; therefore, in total, we had 13 HDI-Regional groups categorized from H1 (High HDI countries in Europe), H2, H3… to L3 (Low HDI countries in Africa). This classification system enabled us to more fully capture the socioeconomic and geographical differences within and between each of the two feature classes.
For each country, we first searched the literature for the number of fishers in the direct and indirect sectors. Usually, the FAO’s Fishery Country Profiles provided a starting point. If available and assessed as reliable, then the data were used. Data on the number of fishers in the direct sector were commonly found under categories such as ‘primary sector employment’, or number of jobs in ‘coastal’, ‘marine’, or ‘industrial’ sectors. If no estimate was available, or an estimate was available but assessed as being not reliable, then government websites and other sources were investigated for a better estimate. Estimates were assessed as being non-reliable if they (i) were more than 10 years old, i.e. 1999 and older, or (ii) explicitly stated that no data had been collected recently, or similar statements to that effect. If an estimate could not be found from FAO or other sources, then we used a proportional transfer approach to fill in the gap.
To conduct a proportional transfer, we computed the average ratio of reported fishers to country population for each of the 13 HDI-Regional groups. If country a was missing fisher employment data, we multiplied the population of country a with the fisher to population ratio for the HDI-Regional group to which country a belonged, and generated an estimate of the number of fishers in country a. For example, in 2003, the population of Singapore was 3.8 million (United Nations 2009). Singapore is a high HDI country in Asia; the average proportion of reported primary sector employment to total population in H2 countries was 0.37%. We then took the product of these two numbers to arrive at 14 000 primary sector fishers in Singapore.
We used marine fisheries employment statistics from 2003 wherever possible. If unavailable, the next closest year’s data were used and then scaled to the base year under the assumption that changes in fisher employment mirrored the general population growth rate. Thus, employment statistics from years other than 2003 were scaled using population change rates from the United Nations statistics division (United Nations 2009). Employment numbers generated in this article do not differentiate between categories of workers. Rather, they show the sum of people who are engaged either full-time or part-time in fishing. This approach was taken because the definitions of full-time and part-time employment vary by country and are not necessarily comparable. A similar approach was used in a comprehensive assessment of the number of fishers in the European Union in 2006 (Salz et al. 2006).
Finally, we disaggregated aquaculture (fish farming) and inland sector employment from total fisheries employment, such that we report only the portion derived from marine fisheries. When a country breakdown was not provided, but aquaculture and/or inland sectors exist and were included in total fisheries employment statistic, we used an approximation based on total catch and production levels to extract employment in the marine sector. We calculated the proportion of each country’s catch from marine fisheries and production from aquaculture and then assumed that the same proportion applied to employment statistics. This approach was used for nine countries – Kuwait, Liberia, New Zealand, Philippines, Russia, South Africa, Syria, Tanzania and Turkey. Although we recognize that employment is not proportional to catch, this is the closest approximation we could do in the absence of other data.
Small-scale fisher estimate
We paid particular attention to quantifying gaps in the number of small-scale and/or unlicensed fishers globally. We estimated missing data, or adjusted existing data, for every country following a set sequence (Fig. 1).
To determine whether small-scale fishing occurred, we assessed the fisheries characteristics of 144 maritime countries (Table S1). As per the approach above, we first searched FAO Fishery Country Profiles for indications of small-scale fishing, followed by peer-reviewed literature and grey literature as necessary. If these sources contained insufficient information for determining the existence of small-scale fishing, we contacted relevant government and/or research agencies.
For all those countries that had small-scale fishing, we again used the same data collection procedure to find the number of small-scale fishers in each country. Small-scale fishers were variously reported as ‘artisanal’, ‘traditional’ or ‘subsistence’ fishers. Where small-scale fishing existed but no data were available, we estimated the number of small-scale fishers in country a (SSFa) using the formula:
where Ca is the rural coastal population of country a and Pa is the proportion of country a’s population that carries out small-scale fishing. The determination of these variables is described below.
Geographically, fishing communities occur along coastal zones and tend to be away from urban city centres. Small-scale fishing is generally prevalent in poorer areas of tropical developing countries where there are few alternative employment opportunities or where famine or war has uprooted people to coastal areas where they take up fishing. We thus defined the small-scale fisher population as being individuals living in rural areas in the coastal zone (FAO 2005). For all countries except small island developing states (SIDS), Ca were obtained from the low-elevation coastal zone series available through the Socioeconomic Data and Application Centre (SEDAC), which used digital elevation maps to estimate rural and urban populations on all land contiguous with the coast that was 10 m or less in elevation (McGranahan et al. 2007). Given that some fishing communities may be located close to urban areas or above the 10 m elevation limit, Ca is a conservative estimate. Low-elevation rural population data tended to underestimate the coastal population of SIDS. Instead, the ‘population within 100 km of the coast’ data series from SEDAC tended to provide more representative estimates of coastal population in most SIDS. Therefore, we used the 100-km data series as input for Ca in SIDS.
Small island developing states in this study are Antigua and Barbuda, Bahamas, Bahrain, Barbados, Belize, Cape Verde, Comoros, Cuba, Cyprus, Dominica, Dominican Republic, Fiji, Grenada, Guinea-Bissau, Guyana, Haiti, Jamaica, Kiribati, Maldives, Marshall Islands, Mauritius, Micronesia, Nauru, Palau, Papua New Guinea, Saint Kitts and Nevis, Samoa, Sao Tome and Principe, Seychelles, Solomon Islands, St. Vincent, St. Lucia, Suriname, Tonga, Trinidad and Tobago, and Vanuatu.
Pa is a value between Pmax and Pmin. The proportion of country a’s population that carries out small-scale fishing was estimated by dividing available estimates of the number of small-scale fishers in country a by that country’s rural coastal population. Small-scale fisher data were sourced from Chuenpagdee et al. (2006). We chose to use rural coastal population rather than total country population because Chuenpagdee et al. (2006) assumed that small-scale fishers usually ‘performed day trips (a few hours sailing, a few hour fishing, and a few hours sailing back)’, which implied that fishers lived near the coast. Small-scale fisher estimates were available for 85 of 144 maritime countries and for all HDI classes and geographical regions except for mid-HDI European countries. Hence, we used primary sector employment sourced from the FAO for mid-HDI European countries. The calculated proportions of fishing population were then grouped by HDI-Region. Within each HDI-Regional group, zeroes and outliers, which we defined as values >20% or lying outside 2 standard deviations from the average, were discarded, and then the highest and lowest of the remaining values, Pmax and Pmin, were defined (Table 1).
Table 1. Minimum (Pmin) and maximum (Pmax) input values of the percentage of coastal population that fish in each HDI-Region.
|HDI-Region||Pmin (%)||Pmax (%)|
We then used a simple Monte Carlo algorithm to estimate the number of small-scale fishers in countries where they occurred. Monte Carlo technique uses repeated random sampling to compute results where there is uncertainty and gaps in knowledge of the system being modelled, in this case the number of small-scale fishers. The simulation involved sampling 5000 random draws from a designated range of the input variable Pa, assuming a triangular distribution within each range. The margin of error from running 5000 samples was <1%. The range of Pa was defined by the upper and lower observed values of the per capita small-scale fishing population of all countries within a HDI-Regional group, Pmax and Pmin.
We illustrate the Monte Carlo simulation using the example of the Philippines. In Equation (1), rural coastal population (CPhils) is a known variable (6.9 million people in 2003), while the proportion of people who fish (PPhils) is uncertain. Monte Carlo fills the uncertainty by assigning PPhils a value between 1.1 and 9.6% at each of the 5000 iterations, where 1.1 and 9.6% are the minimum and maximum values of the percentage of coastal population that fish in medium HDI countries in Asia (Table 1). The number of small-scale fishers in the Philippines (SSFPhils) is calculated 5000 times, and the final number is the averaged value of the estimates from these 5000 iterations.
To avoid double counting, we subtracted officially recorded numbers of subsistence and artisanal fishers (where available) from SSFa as both those classes of fishers were likely to come from the same rural coastal population. Officially recorded artisanal fishers in some countries outnumbered small-scale fisher estimates generated by the model. In this case, we assumed that the small-scale sector was adequately captured by official documentation (i.e. Monte Carlo small-scale fisher estimate equalled zero). Finally, total direct employment (Da) was calculated by adding the small-scale component (SSFa) to reported direct sector jobs (i.e. jobs in the ‘primary’, ‘coastal’, ‘marine’ or ‘industrial’ sectors), which we grouped collectively under the term RDa (Equation 2).
Secondary sector estimate
The secondary sector as reported in the FAO Country Profiles ranged widely to include activities from processing and manufacturing, to boat building, wholesale and vending, to transport and administration. At the same time, many countries did not define what constituted the secondary sector. To remain on the conservative side, we merged ancillary services with the secondary sector such that our ‘secondary sector’ employment estimate covered jobs in manufacturing and processing, as well as typically ancillary activities such as marketing and equipment repair.
Estimating secondary (i.e. indirect) sector employment was a two-step procedure. First, we collected data on secondary sector employment using the same procedure as for small-scale fishing. Countries without secondary sector employment data were assigned a value using the proportional transfer approach, as described in the data collection section. We then calculated the ratio of employment in the secondary to primary sector for each country a. The ratios were then grouped by HDI-Region, and outliers, defined as those lying outside 2 standard deviations from the average, were discarded. We then identified the maximum and minimum ratios of secondary to primary sector, RSPmax and RSPmin, for each HDI-Region (Table 2). The second step involved calculating secondary sector employment that accounts for small-scale fishers (Sa). This was carried out by applying Equation (3) in a Monte Carlo simulation consisting of 5000 iterations, where RSPa is a value between RSPmax and RSPmin.
Table 2. Minimum (RSPmin) and maximum (RSPmax) ratios of secondary to primary sector employment in each HDI-Region.