Modeling sea cage outputs for data-scarce areas: application to red drum (Sciaenops ocellatus) aquaculture in Mayotte, Indian Ocean

Abstract

Robust and accurate prediction of fish farm waste is a first and crucial step in managing the cause–effect chain that leads to local environmental impacts of aquaculture. Since aquatic production is diversifying with new fish species and extending to new areas for which data can be scarce, it is important to develop parsimonious approaches with fewer data requirements and less scientific complexity. We developed the Farm productIon and Nutrient emiSsions (FINS) model, which simulates fish farm operation and estimates fish biomass, feed inputs, and waste emissions from sea cages using simple modeling approaches and a variety of data sources. We applied FINS to red drum (Sciaenops ocellatus) culture in Mayotte by collecting relevant input data (growth, digestibility) from experimental trials. Three explorative farming scenarios—small, medium, and large—were defined from field survey data to examine and compare emissions of a range of potential commercial culture conditions and production scales (23, 299, and 2079 t year⁻¹, respectively). Comparison of the three scenarios showed that waste emissions per ton of fish harvested during routine operations, and thus environmental impacts, were higher for longer culture cycles (medium farm) because of lower feed conversion efficiency. The FINS model is a simple alternative tool to assess and compare environmental impacts of different farming systems and practices for new aquaculture species and regions. It provides important drivers to assess local environmental impacts of fish farms and can therefore facilitate the process of licensing new farming systems for decision-makers.

Appendices

Appendix 1 Description of the digestibility trial with red drum

The commercial diet tested in this study was composed mainly of soybean, fish meal, wheat, and fish oil and contained 48% protein and 13% lipids. The pellets were first coated with yttrium oxide (yttrium oxide and cod oil at 1% of feed weight each) as an inert tracer to determine apparent digestibility coefficients. The animals were 124-day-old (84 g mean body weight, BW) laboratory farmed red drum, originating from captive broodstock. On day 42 (D₄₂), fish (mean BW 83–85 g) were individually weighed and randomly divided into 3 treatment groups of 30 fish, each group placed in a tank to become acclimated to the experimental environment. Beginning on D₂₀, fish (mean BW 145–156 g) were fed with NUTRImarine 4.5 pellets. On D₀, initial individual fish weighed 206–216 g (mean 211 g), and each group was adjusted to 26 individuals. Each group was reared in a 1-m³ indoor tank supplied with 1 m³ h⁻¹ of filtered seawater in a flow-through system. Water salinity was 37.0. PSU, and oxygen concentration always exceeded 80% saturation. Temperature was 27.5 ± 0.5 °C with artificial lighting of 160 lx at the water surface (12 h:12 h L:D cycle, lights on at 6:00 a.m.).

Feed was manually delivered once a day at 8:30 a.m. to each group until satiation. Feed intake of each group was calculated daily as a percentage of each group’s biomass. A sediment trap (150 l each) located at the outlet of each tank was checked for uneaten pellets, and feed loss was considered nil during the experiment.

Feces were collected twice a day (4:00 p.m. and 8:00 p.m.) in the sediment trap via a siphon system for 9 days (D₇₃–D₇₅, D₇₈–D₈₂, and D₈₅) and frozen. Fish scales were removed from samples, and then feces were concentrated by centrifugation and freeze-dried before analysis.

At D₀, D₂₁, D₄₂, D₆₃, and D₈₅ (last day of the trial), feeding was stopped for 24 h, and then fish were individually weighed. A representative sample of whole fish (n = 6) was withdrawn from each treatment group at D₀ (initial) and D₉₁ (final) and kept frozen (− 20 °C) until analysis of body composition. Whole fish bodies were pooled, ground, and freeze-dried before chemical analysis.

Red drum whole-body samples, feed pellets, and feces were analyzed following standard procedures: dry matter after drying at 105 °C for 24 h, protein (N × 6.25) by the Kjeldahl method after acid hydrolysis, lipids after extraction with petroleum ether by the Soxhlet method, sugar by the Luff–Schoorl method, starch by the Ewers polarimetric method, fiber from fraction analysis by the Van Soest method, and ash by ignition. Yttrium contents were measured in feed and fecal samples by atomic absorption spectrophotometry using a nitrous oxide–acetylene flame, after acid digestion (2% nitric acid and 2 g l⁻¹ KCl).

Appendix 2

Table 4 Daily feeding rate (DFR) and pellet diameter used to calculate feed inputs for red drum (Sciaenops ocellatus) culture in a warm water environment (25–31 °C) in the Farm productIon and Nutrient emiSsions (FINS) model and obtained from surveys of farms feeding using commercial Nutrima® feed (2.2, 3.2, 4.5, 6.0, and 9.0 mm)

Table 5 Proximate compositions of commercial feed (N_{i feed}) by pellet diameter and red drum (Sciaenops ocellatus) (N_{i fish}) used in the Farm productIon and Nutrient emiSsions (FINS) model. Raw values were obtained by analysis and then recalculated for 100% of dry matter assuming that total dry weight (DW) was the sum of protein, lipid, sugar, starch, fiber, and ash fractions

Table 6 Values of key parameters used in the Farm productIon and Nutrient emiSsions (FINS) model for three scenarios of red drum farming: small, medium, and large