Fish farming, or aquaculture, is the controlled raising of fish through a managed cycle: juveniles are produced in hatcheries, grown in ponds, tanks, raceways, or net pens, fed a formulated diet, protected from disease and predators, and harvested at market size. Most farmed species are raised in freshwater or coastal enclosures, and roughly half of all fish eaten worldwide now comes from farms rather than wild capture, according to the Food and Agriculture Organization (FAO). The specific methods vary widely by species, water type, and scale.

The word "farming" hides enormous variation. A tilapia pond in Thailand and a salmon net pen in Norway share the same biological logic but almost nothing else in engineering. This explainer follows the actual production chain, quantifies where numbers are stable, flags where they are not, and reconciles the thresholds farmers and regulators use. It covers what happens at each stage, why systems differ, and where the real limits sit.

Hatchery and juvenile production

Every farmed fish starts as an egg or a juvenile, and the hatchery is where biology stops being wild. Broodstock are held under controlled temperature and light so they spawn on schedule rather than seasonally. Eggs incubate in trays or upwelling jars with continuous water flow, and once larvae hatch they consume their yolk sac before switching to live feed or formulated microdiets.

This early stage is the most fragile part of the entire cycle. Larvae are typically a few millimeters long, and survival from egg to juvenile can range from under 10 percent to over 80 percent depending on species, water quality, and handling. Hatcheries therefore measure success in survival percentages and deformity rates, not tonnage. A single degree of temperature error during incubation can shift hatch timing by a day or more and synchronize or scramble a whole batch.

The output of the hatchery is a uniform juvenile, often called a fingerling or smolt depending on species and life stage. Uniformity matters commercially: fish that grow at different rates create grading problems and size variation at harvest. Many hatcheries grade juveniles by size before sale, and some use photoperiod manipulation to time the smoltification window in salmonids.

Grow-out systems: ponds, tanks, raceways, and net pens

Grow-out is where fish spend most of their lives, and the system chosen determines almost everything about cost, risk, and environmental footprint. The four dominant categories differ in how water is contained and exchanged.

System Containment Water exchange Typical species Key constraint
Earthen pond Lined or unlined soil basin Low, often static Tilapia, carp, catfish Oxygen depletion, algal crashes
Raceway Concrete or lined channel High, continuous flow Trout, sturgeon Requires large clean water source
Recirculating tank Tanks plus filtration loop Very low, 90 to 99 percent recirculated Salmon smolt, barramundi, shrimp Energy cost, biofilter management
Net pen Mesh cage in open water Open exchange with lake or sea Atlantic salmon, sea bass, cobia Escapes, waste dispersal, parasites
Integrated system Pond plus crops or livestock Low to moderate Carp, tilapia Nutrient balance, land use

Each row reflects a different trade-off. Ponds are cheap to build but depend on natural oxygen production and are vulnerable to night-time oxygen crashes. Raceways need enormous water volumes but keep fish dense and easy to harvest. Recirculating aquaculture systems (RAS) recycle most of their water through mechanical and biological filtration, which makes them land-based and location-independent but energy-intensive. Net pens use the sea as both water supply and waste sink, which keeps capital costs low but couples the farm to open-water conditions.

The choice is rarely aesthetic. A farmer with reliable groundwater and cheap land builds ponds. A farmer near cold, fast rivers builds raceways. A company needing year-round smolt supply builds RAS. And a coastal operator with a sheltered bay builds net pens.

Feeding, growth, and feed conversion

Feed is the largest operating cost in most finfish farms, and feed conversion is the metric that governs profitability. The feed conversion ratio (FCR) is kilograms of feed per kilogram of fish gained. Values around 1.0 to 1.5 are typical for modern salmon and trout diets, meaning roughly 1 to 1.5 kg of feed produces 1 kg of fish. Poorly managed pond systems can exceed 2.0.

That number deserves scrutiny because it is often misread. An FCR of 1.1 does not mean the fish converts 1.1 kg of wild fish into 1 kg of farmed fish. Modern feeds include plant proteins, soy, pea protein, and rendering products, and the fishmeal and fish oil fraction has fallen substantially over the past two decades. The relevant question is not just total feed but the wild-fish dependency of that feed, which varies by species and formulation.

Feeding is also a behavior problem, not just a nutrition problem. Fish that are overfed waste feed, which sinks and degrades water quality. Farmers use satiation feeding, demand feeders, and underwater cameras to match ration to appetite. In RAS, uneaten feed overloads the biofilter. In net pens, it adds to benthic loading beneath the cage. The professional caveat here is that FCR is a system-level number: it depends on feed formulation, water temperature, fish size, and mortality, so comparing FCRs across farms without those details is misleading.

Disease, parasites, and mortality

Density is the defining risk factor in fish farming. Crowded fish stress, and stressed fish get sick. The three recurring problems are bacterial infections, viral outbreaks, and external parasites such as sea lice in marine salmon farming.

Treatment options are narrower than in terrestrial livestock. Vaccination is widely used in salmon farming and has reduced antibiotic use substantially in that sector. Antibiotics remain important in some pond and tank systems, but their use is regulated and monitored because of resistance concerns. The FDA approves specific drugs for specific fish species and requires withdrawal periods before harvest, meaning treated fish cannot be sold until the drug clears their system (U.S. Food and Drug Administration).

Sea lice illustrate the edge-case problem. A single louse is trivial; a heavy infestation damages skin, opens secondary infections, and can kill smolts. Farms monitor lice counts per fish and must treat above regulatory thresholds that vary by jurisdiction. Treatment itself has costs: chemical baths stress fish, and lice populations have developed resistance to some compounds. This is a genuine limitation, not a solved problem.

Water quality failure (low dissolved oxygen, ammonia spikes) can kill an entire pond or tank in hours. Escapes from net pens can introduce farmed genetics and disease into wild populations. Antiparasitic resistance can make standard treatments ineffective within a few production cycles. Feed contamination or misformulation can cause whole-batch mortality before symptoms are obvious. Regulatory withdrawal-period violations can make fish unsalable and trigger recalls. Overstocking beyond system capacity is the most common root cause of cascading losses.

Environmental footprint and the wild-fish question

The most contested claim about fish farming is that it is either the solution to overfishing or a cause of it. The evidence supports neither extreme. Fed aquaculture (salmon, shrimp, marine finfish) consumes wild-caught fish in feed, so it adds pressure to forage fisheries. Extractive aquaculture (filter feeders such as oysters and mussels, and some carp) requires no feed at all and can improve water quality.

The distinction matters for how you read any sustainability claim. A fed species with a low fishmeal inclusion rate may have a lower wild-fish dependency than its reputation suggests. An unfed species has near-zero dependency but a different set of siting concerns. There is no single sustainability score for "farmed fish."

Waste is the second footprint issue. Net pens release dissolved nutrients and settle organic matter onto the seabed, and the zone of impact depends on current speed and depth. Land-based RAS concentrates waste into a sludge stream that must be treated or disposed of, which shifts the problem from the water column to the waste-handling system. Neither is free. The honest framing is that every system moves the environmental cost somewhere, and the question is whether the receiving environment can absorb it.

Economics, regulation, and practical thresholds

Fish farming is capital-intensive at the hatchery and RAS end and working-capital-intensive at the grow-out end. Feed, juveniles, and labor are recurring costs; a net pen or pond is a fixed asset that produces for years. Prices for farmed species fluctuate with global supply, feed commodity markets, and disease events, so quoting a current farmgate price is unreliable without a dated source. Treat any specific price as a snapshot rather than a stable fact.

Regulation differs sharply by country and system. In the United States, the FDA regulates drugs and feed additives for aquaculture, while the EPA administers discharge permits under the Clean Water Act for facilities that discharge to waters of the United States. Coastal net-pen farming in federal waters has been debated for years and remains limited. Norway, Chile, and Scotland have their own licensing regimes with biomass caps and lice thresholds.

For anyone evaluating a farm or a product, the practical thresholds are concrete:

Ask what species and what system, because the answer determines almost everything else. Ask the feed conversion ratio and the fishmeal and fish oil inclusion rate, not just "sustainably sourced." Ask about mortality rate over the last full cycle; a figure above roughly 20 percent signals a management problem worth investigating. Ask whether the farm is certified and by whom, and check what the standard actually covers. Ask about disease and treatment history, including withdrawal compliance. Ask where waste goes, because that is the footprint that rarely appears on a label.

Reconciling thresholds across sources is necessary because they are expressed in different units. Lice thresholds are counts per fish. Oxygen thresholds are milligrams per liter. Feed inclusion rates are percentages of diet. Mortality is a percentage of stock per cycle. None of these convert into each other, and a farm can pass one threshold while failing another. A single number never captures a farm.

What fish farming cannot do

The limits of aquaculture are biological and thermodynamic, not just regulatory. Fish grow faster in warm water up to a species-specific optimum, after which growth slows and oxygen demand rises. That ceiling cannot be engineered away with feed. Cold-water species such as trout and salmon have narrow thermal windows, which is why their farming is geographically constrained.

Disease control has a hard limit too. Vaccines work for some bacterial and viral targets but not all, and parasites evolve resistance faster than new treatments are approved. The result is that biosecurity, siting, and stocking density do more work than drugs in a well-run farm. A farm that relies on treatment rather than prevention is structurally fragile.

Finally, the feed bottleneck is real. Global forage fish supply is finite, and while plant-based proteins have displaced much of the fishmeal in salmon diets, some species still require marine ingredients for optimal growth and health. This is the honest caveat that thin explainers skip: fish farming is neither a free lunch nor a villain, and the production cycle is a chain of trade-offs that the numbers can describe but never fully resolve.