Sustainability of Farming Nile Tilapia in Net-Cages in a Reservoir of the Brazilian Semi-Arid Region During an Extended Drought Event

Gustavo Gonzaga Henry-Silva (  gustavo@ufersa.edu.br ) Universidade Federal Rural do Semi-Arido https://orcid.org/0000-0001-5067-225X Júlio César da Silva Cacho Instituto Federal do Rio Grande do Norte: Instituto Federal de Educacao Ciencia e Tecnologia do Rio Grande do Norte Rodrigo Sávio Teixeira Moura Universidade Federal do Maranhao Dallas Lee Flickinger Universidade Estadual Paulista Julio de Mesquita Filho Wagner Cotroni Valenti Universidade Estadual Paulista: Universidade Estadual Paulista Julio de Mesquita Filho


Introduction
The global aquaculture industry was estimated to consist of 19 million participants, most of which are in Asia, the Caribbean, and Central and South American countries (FAO 2020). When considering the objectives described in the 2030 Agenda for Sustainable Development (UN 2019), modern aquaculture must contribute to food security, sustainable use of water, reducing the discharge of e uents and chemicals, promoting inclusive economic growth and establishing other sustainable production standards. The increase in aquaculture production worldwide is due to the rising demand for healthy and nutritionally adequate foods (Boyd et al. 2020;Zhong et al. 2021). Therefore, the sector must adapt to an ever-increasing global sh consumption by generating alternative products and higher yields while improving management of wastes, production techniques, and the e ciency of using natural resources, without negatively affecting the safety of the food products ( In Brazil, aquaculture is estimated to increase by 32% by 2030, with a production of 800 thousand tonnes (FAO 2020). Funding for advancements in Brazilian aquaculture have been due to large investments in the sector over recent years and the creation of public policies that promote participation in the activity (Garcia et al. 2014; Bueno et al. 2015; Padial et al. 2017). The Nile tilapia (Oreochromis niloticus) is currently the main species used for sh farming in Brazil and its production in 2019 was 432,000 t, which was 57% of national production. Tilapia farming in Brazil is mostly carried out in net-cage systems installed in public reservoirs ( (Valenti et al., 2021). Given the importance of understanding the sustainability of sh production in cages and the development of effective analytical methods, the present study described the environmental, social and economic dimensions of sustainability of a net-cage tilapia production system implemented by a commercial cooperative in a reservoir in the semi-arid northeast region of Brazil during an extended drought event.

Materials And Methods
The present study was carried out by accompanying a Nile tilapia farming cooperative in the Umari reservoir, located in the Brazilian semi-arid northeast region (5 °42'13"S and 37 °15'18"W) (Fig. 1). The sh were reared in net-cages and managed by 13 entrepreneurs that formed the cooperative, which had a workforce of 15 employees. This production model implemented in Umari is the most used in Brazil, including the creation of large aquaculture parks in public and hydroelectric reservoirs. This sh farming cooperative showed an annual production of ~281 tonnes of Nile tilapia. The production is sold as whole sh (98%) and eviscerated sh (2%). The reservoir has an area of 2,923 ha and a maximum capacity of 293 million m³ (ANA 2007). The local climate is of the BSw´h´ (Köppen) type, characterized by a hot and semi-arid climate and with the rainy season occurring in late autumn. Between 2012 and 2017 there was a gradual reduction in the volume of the reservoir, due to the suprasonal drought that affected most of the Brazilian semiarid region (Henry-Silva et al. 2019). During the observation period, the percentage of reservoir volume in relation to its maximum volume ranged from 19-21% (Fig. 2).
The cooperative of the Umari reservoir was observed each month from October 2015 to January 2016 during a tilapia production cycle, from stocking to harvest. Tilapia were obtained with an average individual biomass of 5.3 grams and were reared in an intermediate grow-out phase for two months in the Umari reservoir. Tilapia juveniles (118.5 ± 5 g) were obtained after rearing in the intermediate grow-out phase and were restocked in 6 m³ net-cages (2×2×1.5 m) for another two months at 100 and 125 sh/m 3 (600 and 750 sh per cage, respectively) for a production cycle of four months (two months of intermediate grow-out and two months of nal grow-out, which occurred simultaneously at the cooperative). The structure of the net-cages consisted of steel rods to support a galvanized wire mesh enclosure and closed drums were used to maintain the cages oating. The net-cages were set at approximately eight meters from the reservoir bottom.
Fish from four net-cages of each density were randomly selected and analyzed. A sample of 5% of sh in each of the selected net-cages were weighed weekly and the mean weight calculated to estimate the biomass in each cage and the mean individual weight of the sh. The biomass values were calculated by multiplying the mean individual weight by the estimated total number of sh in each cage, which was based on the difference between the initial sh population and the daily monitoring of sh mortality. Fish were counted and weighed at the harvest to determine the nal average mass, survival, and apparent feed conversion ratio. The sh were fed with a pelleted commercial feed (40% crude protein) until the rst thirty days of cultivation and, subsequently, extruded feed was provided with 32% crude protein. The feed was offered twice daily according to the total biomass of the stocked sh, which was estimated weekly to adjust the feed quantity. The initial feeding rate was 4.5% of the stocked biomass. This amount decreased weekly until reaching 3.1% of the stocked biomass at the end of the experiment. The environmental, social and economic sustainability were evaluated based on three production cycles per year.
A set of seven indicators was used to evaluate the economic sustainability. The indicators represented main aspects of economic sustainability: e ciency in the use of nancial resources and the ability to generate capital for reinvestment (Valenti et al. 2018). The economic analyses were based on data provided by the cooperative and on-site observations. The nancial movement consisted of information for the years 2012 and 2015. All equipment, utensils, inputs and management used during the production were surveyed. Cost-return and cash ow analyses were performed (Engle 2010). The initial investment included net-cages, canoe, electronic balance to weigh biomass, management platform, sheds and other items of low cost. Revenues were reported by the cooperative based on sales in 2015 and estimated for the production carried out by the 13 farmers. All monetary amounts were converted from Brazilian Reals to US Dollars and were based on the average trading price from November 2015 to January 2016 (US$ 1.00 = R$ 4.02). Net revenue was calculated considering productivity and an average selling price of US$ 1.24/kg. Pro t was calculated as the difference between net revenue and production costs, including taxes.
A set of seven indicators was used to evaluate environmental sustainability ( Table 2). These indicators re ect the use of natural resources, the e ciency in using resources, the release of pollutants, and the risk of damaging genetic diversity and biodiversity (Valenti et al. 2018). The nutrients generated from the aquaculture activity were quanti ed each month by placing sedimentation chambers (in duplicate) on the reservoir bottom and under each net-cage for 24 hours. Each sediment sample was weighed and the concentrations of particulate and organic matter were calculated according to Buffon et al. (2009). The natural sedimentation rate of the reservoir was estimated by placing sedimentation chambers in an area located about 200 m from the net-cages (control). The concentration of total phosphorus (Koroleff 1983) of the sedimentation was calculated by subtracting the natural sedimentation concentration (control) of this nutrient from the sedimentation concentration recorded below the net-cages.
Social sustainability was evaluated using a set of 17 indicators (Table 1). These indicators represent four major aspects of the social dimension: social equity, income distribution, equal opportunities, and the generation of jobs and bene ts for local communities (Valenti et al. 2018). Data were acquired by interviewing employees of the Umari cooperative through the use of a semi-structured questionnaire (Valenti et al. 2018). In addition, other data were acquired from a survey carried out with the public agencies of the Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geogra a e Estatistica -IBGE) and the Ministry of Fisheries and Aquaculture (Ministério da Pesca e Aquicultura -MPA). Table 1 Indicators of economic, environmental and social sustainability. Space and resources to produce the feed were excluded from analyses of these indicators.

Results
Survival was 90% and the Apparent Feed Conversion Ratio was 1.5 for both stocking densities. Mean nal weight was 316.9 grams and 358.8 grams for the stocking densities of 125 and 100 sh/m 3 , respectively. Fish farmers anticipated harvesting by a few months due to the reduction in the volume of the Umari reservoir. This early harvest re ected negatively on economic indicators. All investments to implement the net-cage production system were carried out by the 13 members of the Umari reservoir cooperative. The initial investments included 37 net-cages for each member (total of 481 net tanks), canoes, management platform made of wood, shed for production and employee support, electronic balance for weighing biomass and other items of low cost, totaling US$ 93,445. No reinvestments were made for new net-cages since the environmental licenses permitted a maximum of 37 cages per producer. The compensation of the entrepreneur was considered zero since members of the cooperative work on other activities and decide individually where they will apply the pro ts from the tilapia production. The economic sustainability indicators showed that the production system carried out with a stocking density of 125 sh/m 3 was not economically feasible because the Internal Rate of Return (7%) was lower than the Minimum Attractive Rate of Return (10%). The negative Net Present Value (US$ -26619.70) also showed that the system was economically unfeasible. The production system with 100 sh/m 3 generated an Internal Rate of Return (5%) and a negative Net Present Value (US$ -47821.30), indicating that this system was not economically feasible as well ( Table 2). The relationship between average annual income and investment was relatively low (<US$ 1.00) for both stocking densities. The producers used the pro t to compensate for sh mortality, administration of the production, control of inputs and sales.
The economic indicator Risk Ratio consisted of 11 factors that increase the risks of negative impacts on aquaculture. In the present study, this indicator was of 18% for both stocking densities. Only two of the 11 factors were identi ed: (i) lack of a business plan during the planning stage and (ii) institutional instability due to changes in regulations and in agencies of promotion and inspection. The cooperative generated only two types of products, which were whole sh (~98% of total production) and llets (~2%).
The cooperative focused on the three markets of intermediaries, slaughterhouses of the state of Rio Grande do Norte, and small merchants in the city of Upanema/RN and the surrounding region near the Umari reservoir. The capital generated for reinvestment in the activity was considered zero since the environmental licenses permitted a maximum of 37 net-cages per producer (Table 2). 9. Invested Capital Generated in the Activity 0.00 0.00 The work required to carry out the cultivation in the Umari reservoir was 2 man-hours-year per square meter (MHY/m²) for both stocking densities, with no variation because the number of workers, hours of work per day and the size of the production area were the same ( Table 3). The required work per unit of production was 0.13 and 0.15 man-hours per kg of sh produced (MH/kg) for the densities of 125 and 100 sh/m 3 , respectively. The pay equity was 80% for both tilapia densities. Values of racial inclusion (75%), gender inclusion (52%) and age inclusion (41%) were equal for both densities as well when considering that the composition of the employees was the same between the two densities. The indicator income xation in the local economy showed that only 2 and 3% of the acquisitions of goods were made in the municipality for the densities of 125 and 100 sh/m 3 , respectively, whereas the other acquisitions were made in other municipalities or states. Local consumption showed that only 10% of the sh was commercialized and consumed in the municipality through small merchants.
Employees of the cooperative received health bene ts through a public federal health system known as the Uni ed Health System, which is available in most Brazilian municipalities. The management model for this system is decentralized, with the federal government, states and municipalities working together to ensure free health care for all citizens. This health service is maintained by tax revenues from private health programs. The education level showed that only 7% of the employees were in school while the other 93% had complete or incomplete basic education. In addition, permanence in the enterprise was three years for each employee. Regarding the participation in community activities, 100% of employees showed association with the rural workers union of the city of Upanema/RN. Among the work safety items, the enterprise showed 87% of equipment and actions necessary to carry out the activity with proper safety. Only the use of pigmented gloves and the use of equipment by quali ed professionals were not identi ed.   (Table 4). Phosphorus accumulation was 0.0021 kg and 0.0020 kg of phosphorus per kilogram of sh produced for the stocking densities of 125 and 100 sh/m 3 , respectively. Pollution from herbicides, pesticides and hormones was zero since none of these products were used in the net-cage system in the present study. The production system generated 106 kg and 101 kg of particulate material per tonne of sh produced for the densities of 125 and 100 sh/m 3 , respectively. Approximately 90% of this emission consisted of organic matter for both densities (Table 4).  . Nearly all (98%) of the sh production in the Umari reservoir was sold as whole sh with a xed value to intermediaries, slaughterhouses and small merchants in the region. In addition, the production was with taxes and without government subsidies, resulting in a higher production cost when combined with paying employee salaries and other costs.

Discussion
The relationship between net income and initial investment showed that the amount invested in the Umari cooperative was not effectively transformed into income. This was perhaps due to the low revenue generated from the production, resulting in a low return over the life of the project. The indicator of risk was 18% for both stocking densities. Of the 11 items used to determine this indicator, only the absence of a business plan during the planning phase for implementation and institutional instability were observed, since instability may occur due to changes in environmental laws. However, this indicator also showed that some of the cooperative members had specialized training, which leads to greater security of the production given that the correct management is carried out daily and sh diseases are treated with more con dence. The cooperative also showed no con icts with the local community or non-governmental organizations. The number of products and the available markets were considered adequate for the activity with both stocking densities, showing no reduction in sustainability since all production was sold and met local demand. The capital generated from the activity was zero for both systems, showing no reinvestments for the purchase of new cages or any other equipment during the analyzed period because the environmental licenses granted to the cooperative permitted the use of up to 37 cages per member.
The entire workforce participated directly in production and employees were local residents. The pay equity was 80% for both stocking densities. These values were considered satisfactory as most employees worked daily to manage the production and received equal salaries, and two employees worked additional night shifts as security and received higher salaries. Furthermore, participation in external activities showed that all employees were members of the rural workers union. On the other hand, social sustainability was reduced for both densities because of the low generation of job opportunities. Hence, the overall number of employees was low given the amount of work per area and per production for the net-cage tilapia culture system in Umari reservoir.
The number of individuals that received bene ts was low despite the activity providing direct and indirect employment opportunities for approximately 130 intermediaries and small merchants. However, the generation and distribution of income have more social relevance when considering the possibility that each worker represented a family. It is also noteworthy that the enterprise provided safe working conditions to those involved in the production process as only pigmented gloves and the use of equipment by quali ed professionals were absent among the work safety items. In addition to providing security to cooperative employees, the enterprise was socially inclusive by providing opportunities to people with low education. Only one of the 15 employees studied while the others had incomplete elementary or high school education. Various age groups and ethnic groups were represented among the employees, but the enterprise was operated only by men. The average permanence of employees in the activity was 3.0 years, which is relatively high considering that the cooperative only existed for four years when the present study was carried out. The high permanence in the activity was perhaps due to the social relevance and importance of the activity to the members.
Approximately 10% of the sh were consumed in the region where the cooperative was implemented and all production was consumed in the state of Rio Grande do Norte. All production consumed within the state suggests that the activity improves the local food supply. The ratio of direct and indirect generation of income to the capital invested in the enterprise was low for both stocking densities. This indicator re ected the reality that cultivation systems carried out in net-cages require little management. The compensation of labor relative to the gross production of the enterprise was less than US$ 0.29 per employee per kilogram of production for both densities. Although the employees of the cooperative had no access to private health programs, they all had access to the public federal health system. The proportional cost of work showed that a reasonable share of the production costs was allocated to pay employees.
Income retained in the local community is among the more important indicators of social sustainability. The low retention observed in the present study was due to the purchases of major inputs (including feed and ngerlings) from other municipalities and states, minimizing local purchases for basic maintenance items from local retail, as well as occasional lodging for members, fuel, ice and other items of low impact on the local economy. This is especially relevant since feed and ngerlings usually show the highest proportions of total operating costs in sh farming activities (Barbosa et al. 2020). However, local retention of income becomes 90% on the state level, showing that minimal nancial resources are destined to other states. The retention of resources invested in the activity, especially feed and ngerlings, can become higher if these items were produced in the city of Upanema -RN, where the Umari reservoir is located. Moura et al. (2016) also showed that the pro t generated by the activity was destined toward expenditures outside of the community where the enterprise was inserted, which reduced the capacity of the enterprise to provide local social and economic development. The accumulation of nutrients from suspended solids over time can have signi cant impacts on the environment when considering that most of the waste generated was organic matter, which reduces dissolved oxygen through microbial aerobic decomposition (Flickinger et al. 2020a). In general, the present study showed that environmental sustainability was in uenced by the generation of solid wastes, of which its increase over time reduced the sustainability of the production system carried out with either density. Consideration must also be given to investments for the construction of arti cial ponds to optimize water use, which may provide longer continuity for this activity when compared to the use of reservoirs in the Brazilian semi-arid region. Earthen ponds facilitate the control of wastes and the escape of exotics when compared to reservoirs. Recent studies have shown the technical feasibility of integrating prawns in earthen ponds with sh as free-swimming or in net-cages, and that these IMTA systems may prolong the use of water given the improved conversion of nutrient inputs into harvested biomass . Reduced rainfall has been recorded in the state of Rio Grande do Norte over recent years, more speci cally in the hydrographic basin of the Apodi-Mossoró river, where the Umari reservoir is located. At the end of 2016, approximately 70% of the reservoirs in the state that have a water capacity above 5,000,000 m3 were practically dry (<1% of total volume). In 2016, when the Umari reservoir was 18% of its total volume, the possible turbulent vertical circulation favored the mixing of the entire water column and as a consequence, the reduction of dissolved oxygen and a mortality about 60 tonnes of Nile tilapia (Henry-Silva et al. 2019). As a result, sh farming in net-cages in the Umari reservoir was temporarily suspended and resumed recently on a much smaller scale, with only four producers investing in the activity and producing about 80 tonnes of Nile tilapia annually. In March 2020, the reservoir remained with a reduced volume, with only 32% of its total volume (SEMARH 2020).
The sh farming in net-cages must consider the hydrological characteristics of the reservoirs since variations in their water level can drastically reduce the sustainability of this activity. The majority of the reservoirs in the semi-arid northeastern region of Brazil are subject to high variations in water volume due to prolonged periods of low rainfall (<700 mm/year), of which a reduced water level combined with turbulent vertical circulation may result in mortality of sh reared in net-cages (Henry-Silva et al. 2019).
Thus, the sustainability of net-cage sh farming in the Umari reservoir and other reservoirs in the Brazilian semi-arid region are also related to the volume of these aquatic environments and the length of drought periods.

Declarations
Ethical Approval: This is a research article following the ethical standard of the institution. Location of the Umari reservoir in Brazil (5°42'13"S and 37°15'18"W). The circle indicates the area of the reservoir where the Nile tilapia are farmed in net-cages.