Breeding and Genetics

Quantitative genetics rests on a simple decomposition: the performance an animal displays is the sum of the genes it carries, the environment it experienced, and the interaction between the two. Selection can only act on the genetic part, and the practical work of a breeding programme is the separation of that part from everything else in a set of observations that mix them together.

Aquatic species offer advantages that terrestrial livestock breeders do not have. Fecundity is high, so selection intensity can be extreme. Generation intervals are short in many species. External fertilisation permits controlled crosses and, in some species, the production of full-sibling and half-sibling families for estimating genetic parameters. Against this stands the difficulty of identifying individual animals, the fact that families must often be reared separately before tagging, and the environmental variance introduced by that separate rearing.

Genetic gain is permanent and cumulative, which distinguishes it from every other improvement a farm can make. Better feed must be bought again next cycle; better aeration must be maintained and powered; a genetic gain, once fixed in a population, is inherited without further cost. Over many generations this compounds into differences in growth rate, survival and product quality that no husbandry change can match.

The opposite is equally true and rather more common. A hatchery that repeatedly spawns from a small number of readily available parents, or that unknowingly breeds relatives, accumulates inbreeding whose effects appear as reduced survival, deformities and declining growth several generations later. By the time the pattern is visible, the population that would have been needed to correct it has often been lost.

Heritability

Heritability in the narrow sense is the proportion of the observed variation in a trait that is attributable to additive genetic effects — the part that is transmitted predictably from parent to offspring. It is not a property of a trait in the abstract but of a specific population measured in a specific environment. The same trait can show useful heritability in one setting and almost none in another, usually because environmental variation in the second setting swamps the genetic signal.

Breeding value

An animal's breeding value is the effect its genes are expected to have on its offspring, expressed as a deviation from the population mean. It is what selection should be based on, and it is not directly observable. It is estimated from the animal's own record and the records of its relatives, weighted by the relationship between them and by the heritability of the trait.

Selection differential and response

The selection differential is how far the chosen parents exceed the population mean for the trait. The response is how far their offspring exceed it. The ratio between them is heritability, which is why selecting hard on a trait with low heritability produces disappointment: the parents were exceptional, and most of the reason was environmental.

Genetic correlation

Traits are not independent. Selecting for growth may change body shape, age at maturity, disease resistance or flesh composition, because the same genes influence several traits. Where the correlation is unfavourable, single-trait selection produces a gain accompanied by a loss that may not be noticed for several generations. This is the principal argument for multi-trait selection with an explicit index.

Effective population size

Effective population size describes how many individuals are genuinely contributing genes, as distinct from how many are present. In hatcheries it is frequently far smaller than the broodstock count, because a few dominant spawners contribute disproportionate numbers of offspring and because family survival to tagging varies. It is the single number that governs the rate at which inbreeding accumulates.

Traits worth selecting for

Growth rate has historically been the primary objective because it is easy to measure, moderately heritable in most species and directly connected to cycle length. Survival and disease resistance have become at least as important where a specific pathogen constrains the sector, and resistance to several agents has shown sufficient heritability to respond to selection. Beyond these sit feed efficiency, which is difficult to measure on individuals held communally, and quality traits including flesh composition, colour and yield.

Genotype by environment interaction

A line improved under one set of conditions may not retain its advantage under another. Where the production environments served by a programme differ substantially in temperature, salinity or system type, the interaction has to be measured rather than assumed, and it may justify separate selection lines.

Most farms and most hatcheries will never operate a selection programme, and should not attempt one. What they should do is manage their broodstock so that the genetic quality they purchased is not eroded, which is a much smaller task with most of the downside risk attached to it.

1. Record the origin, date and supplier of every broodstock group, and keep those records for the working life of the animals.
2. Use as many parents as the facility can practically spawn, and balance the numbers of males and females contributing.
3. Rotate spawners rather than repeatedly using the individuals that are most convenient or most responsive.
4. Avoid selecting broodstock from within your own production output unless the pedigree is known, since that is the fastest route to unintended inbreeding.
5. Introduce unrelated animals from a documented source at planned intervals to maintain variation.
6. Keep families or spawning batches identified through the nursery stage wherever the facility allows it.
7. Evaluate purchased seed by keeping a comparison group, so that a supplier's performance is a measurement rather than a reputation.
8. Record what was crossed with what, even when no selection is intended, because the record is what makes a later correction possible.

Broodstock are the most valuable and most demanding animals on a site. They require space, water quality at least as good as any production unit, a diet formulated for reproduction rather than growth, and often the controlled photoperiod and temperature regimes that condition them to spawn on schedule. Under-resourcing broodstock holding while investing in selection is a common inconsistency, and it limits the programme through poor gamete quality rather than through genetics.

Family-based programmes impose specific engineering requirements: many small, independent, individually supplied rearing units for the separate-rearing phase, hydraulic separation between them, and a tagging facility. The number of units required rises with the number of families, and it is normally this, rather than the genetics, that sets the practical scale of the programme.

There are also regulatory and environmental considerations. Movement of improved strains, and any use of genetically modified organisms, is regulated, and escape of farmed animals into waters holding wild populations of the same or related species raises legitimate genetic concerns. Containment is part of the responsibility that comes with holding an improved population.

• Define the breeding objective in writing, with economic weights, before selecting anything.
• Constrain inbreeding explicitly at every generation rather than checking for it afterwards.
• Maintain records that allow any animal's ancestry to be reconstructed; a breeding programme without a pedigree is not one.
• Test candidates in an environment representative of where their offspring will be produced.
• Keep the nucleus population separate from commercial production, both physically and in its management.
• Prefer multi-trait selection with an index over sequential single-trait selection, because correlated responses are otherwise invisible until they are large.
• Measure realised gain against a control or reference group rather than against the memory of previous cycles.

• Begin with strain comparison under your own conditions before contemplating a selection programme, because it is cheap and frequently sufficient.
• Build the base population from several sources to maximise the variation available to select on.
• Track effective population size as a routine number, not as an occasional calculation.
• Maintain a slowly changing reference or control group so that realised gain can be measured rather than assumed.
• Resource broodstock holding to a higher standard than production units, since gamete quality limits everything downstream.
• Document the dissemination path from nucleus to commercial hatchery, and set minimum parent numbers at every tier.
• Where technical capacity is limited, buy improved seed from a documented programme and concentrate effort on not degrading it.

The central tension in any breeding programme is between gain and diversity. Selecting more intensely raises the response in the short term while reducing the number of families contributing, which raises inbreeding and erodes the variation that future response depends on. Formal optimum contribution selection resolves this by maximising genetic merit subject to an explicit constraint on the average relatedness of the selected group, which is a better formulation than selecting on merit and hoping the relatedness is acceptable.

A second theme is the economics of the programme itself. Genetic gain accrues to whoever produces the animals, which is often not whoever paid for the programme, so breeding programmes in aquaculture require a dissemination and business structure as carefully designed as the selection scheme. Many technically sound programmes have failed on this rather than on genetics.

Finally, the objective itself should be revisited. A programme selecting for growth in a sector whose binding constraint has become disease, or feed efficiency, or resilience to warmer water, is optimising a variable that no longer limits anything. Because response is cumulative, a misdirected objective compounds just as reliably as a correct one.