91av

THE FISHING INDUSTRY has come a long way since our ancestors first speared
and trapped fish. But, in one essential aspect, fishing remains unchanged.
No matter how sophisticated the fish-finding sonar, or how powerful the
winch that hauls in the catch, fishing boats still hunt wild prey.

Most of the fish we eat have not been selected for higher yields and
more efficient production, as grain and cattle have. The few exceptions
live on fish farms. Aquaculture, the farming of fish, has expanded enormously
in the past few decades. Yet most fish farms are still at the stage our
forebears had reached when they planted wild grass in fields so that they
could collect the grain more easily.

A few fish farmers have bred a better fish. Norway has the biggest programme
aimed at improving fish, and has begun to make some headway. Norwegian fish
farms have achieved an increase in growth rates of 12 per cent in each of
three generations of salmon bred since the the programme began in 1975.
Yet Norway’s success stands out in a history of failure. Biologists are
now debating whether Norway’s approach to breeding fish can be extended
to traits besides growth rate. More important, perhaps, can techniques designed
for efficient and easily controlled fish farms in Europe work in developing
countries, where better aquaculture is sorely needed? In some cases, breeders
must devise entirely new strategies for the Third World, and to do that
they need to know far more about the basic biology, and even sociology,
of fish.

To illustrate the frustrations of breeding fish, biologists cite a classic
case of how breeding from the ‘best fish’ failed to produce a better one.
The scientist takes the biggest individuals in a group, and breeds them.
If size is controlled genetically, the offspring should be bigger than the
average size of the population their parents came from. As often as not,
however, they show the same range of sizes as the fish in the initial population.

Roger Doyle, of Dalhousie University in Canada, believes that such experiments
fail because bigger fish do not necessarily have genes that reliably control
the physiology of growth. Other genes that control size react to social
and environmental factors. The breeder must design experiments taking into
account social and environmental variations that can influence the size
and other characteristics of fish as profoundly as their genes. In some
species, for example, the biggest fish in the pond are the most aggressive;
they are big because they take most of the food. Breeding from the biggest
fish produces offspring that are more aggressive, but not necessarily bigger.
The most aggressive offspring will still grow to be the largest because
they win most of the food, and less aggressive fish will still be smaller.
If this pond full of aggressive fish has the same amount of food to fight
for as before, the breeder could end up with the same proportion of large
and small individuals as the population from which their parents were selected.
The fish will never be uniform. Such subtle differences as which fish hatches
first, by as little as a few minutes, establish differences that determine
how a fish interacts with its siblings and pondmates.

The outcome also depends on how the farmer manages the fish. If food
is distributed so that each fish eats as much as it wants, without competing,
the docile fish, which spend their time eating rather than fighting, will
grow largest. If, on the other hand, there is less food and the fish have
to fight over it, the aggressive fish will be biggest. If the breeder simply
selects big fish to breed from, and access to food changes subtly during
the selection process, the results of the experiment will be hopelessly
confused. According to Doyle, inattention to such environmental and social
factors has ruined many breeding programmes.

This does not mean that fish breeders are wasting their time. It is
possible to improve traits that are influenced by genetic factors – the
problem is in identifying the traits. The more genes that are involved,
the harder the task. Scientists have charted enormous variability in susceptibility
to disease among trout and salmon, for example, and genetics seems to be
involved. The big problem is how to design experiments that select characteristics
that are largely the result of just a few genes and that breed true in the
offspring. The breeder must also take into account environmental influences
on how the genes ‘perform’, otherwise the improved fish may not live up
to their promise when sent to live outside the laboratory.

Graham Gall, of the University of California at Davis, says few fish
farmers take advantage of genetic variation among fish because they often
select the wrong animals. Often, breeding stock consists of the fish left
over at the end of a season, usually the slowest growers. A more systematic
approach is needed. Gall and the Norwegian researchers believe that conditions
in breeding tanks should be strictly controlled, so that farmers can select
the genetically superior fish. Doyle disagrees. He thinks that breeders
should devise selection procedures that take into account the effects of
different types of management. Doyle bases his strategy on his experience
in Southeast Asia, where it is difficult to keep such a tight control on
conditions on fish farms.

Despite the different approaches, fish biologists agree on one point:
that it is essential to keep pedigrees for fish. This has been a bone of
contention among fish biologists. The argument stems from work in Israel
in the 1950s, in which breeders achieved little improvement in carp selected
and bred for five generations. The Israeli researchers decided that picking
individual, good-looking progeny from the mass of fish, and breeding from
them, was a better approach than keeping detailed pedigrees of families
in attempts to trace patterns of inheritance.

Many scientists think that this attitude slowed attempts to breed better
fish. By analysing families separately, they say, it is easier to see when
one family shows more improvements than another. As the breeders know which
fish are the parents, they can repeat the cross. This should produce improvements
faster than breeding randomly from good-looking individuals, whose advantages
might not be something they can pass on in their genes. Analysis of the
Israeli experience is helpful, however, in designing better selection procedures.
Trygve Gjedrem, of the Norwegian Institute of Aquaculture Research, suggests
that the Israeli researchers did achieve some genetic improvement but failed
to recognise it because it was masked by environmental effects.

The Israeli work also shows the extent to which social interactions
can confuse the results of breeding experiments, especially with fish. Isolating
fish in an attempt to unravel environmental and genetic effects fails because
an important part of a fish’s environment is other fish. An isolated fish
would be so abnormal that its characteristics would be meaningless for the
breeder.

One problem the Israeli researchers encountered stemmed from the fact
that fish born on different days grow in very different ways. The interaction
between the fry drives them apart into groups with different characteristics.
The fish were all analysed together, using the classic statistical model
developed at the University of Edinburgh by the geneticist Douglas Falconer.
Falconer’s model specifies the crosses and backcrosses a breeder must make
to increase the frequency of a desired trait in a population of animals.
The model assumes that interactions between animals make no difference to
the expression of their genes. This assumption allows all the offspring
of a cross to be counted together in assessing the frequency of the trait.
It is a useful simplification for breeding animals such as cattle.

But cattle are not fish. The differences between fish that result from
their interactions, for example, between fish born at different times, mean
that this assumption does not hold for them. When the Israeli scientists
analysed different age-classes of fish together, they saw no statistical
increase in the trait they were trying to select for. If they had looked
at age-classes individually, however, they might have found an increase,
which would have shown that the trait was indeed inherited.

Statisticians must find ways to take account of interactions between
fish, and other factors that make fish different from cattle, when designing
breeding experiments. For example, many fish die during these trials. Breeders
treat the deaths as random, but they might not be. Experiments aimed at
isolating a strain of fast-growing fish, for example, may only find one
with a high survival rate, for reasons that have nothing to do with genetics.
Such a strain would not breed true for high growth rate.

Doyle believes that behavioural differences may account for 70 per cent
of the variation among fish. For example, if fish are fed freely, the fast-growing
fish may be the ones which are hungrier, because they metabolise at a greater
rate and use food less efficiently. If food becomes limiting, they may suddenly
become the slowest growers. Many breeding experiments, says Doyle, are too
small to allow for such differences.

Gall is worried that scientists and fish farmers have no rational approach
to improving stocks. Some biologists have reported hybrid vigour in the
offspring after breeding from two inbred lines of fish. They suggest that
it would be more useful to exploit this phenomenon than to select fish with
particular genetic characteristics.

Pick of the parents

Gall disagrees. This strategy would mean that aquaculture would then
be based on stocks with little genetic variation, and little capacity to
adapt to changes in their environment. Such an approach led to disastrous
crop failures when applied to corn in the US. But fish farmers are entrepreneurs.
‘They have to develop the patience for long-term genetic improvements,’
said Gall. Just the habit of systematically recording the characteristics
of fish on pedigree charts has a way of improving results, he said. Farmers
pay more attention to variations in their fish, and with pedigrees, they
can try various crosses, check which ones work, then repeat the successes.
This applies equally to a salmon breeder in Scotland and a fish farmer in
Thailand.

The principle of charting pedigrees is simple. The practice of tracing
which fish is related to which is more difficult. In industrialised countries,
much research is focused on ways of marking individual fish in detailed
breeding programmes. One approach is to analyse the genetic characteristics
of fish retrospectively, by screening DNA from the mitochondria, the bodies
in cells that produce energy. Because mitochondria are inherited from eggs,
but not sperm, the DNA contained in mitochondria provides a record of maternal
lineage. More direct techniques include making notches on fins, dye marking
and branding fish with cold metal. Gall uses a fairly sophisticated method
of tagging fish. He injects a tiny plastic transponder into a fish’s abdomen.
The transponder emits a coded electrical signal when stimulated by a magnetic
wand. The 10-digit code can be used to distinguish hundreds of fish. Transponders
are now cheap enough to use in large breeding programmes.

Christophe Herbinger, of the St Andrews marine station in Canada, discovered
why some Atlantic salmon mature too early by analysing families. Farmers
want fish that mature late, investing their energy in growth rather than
reproduction. Herbinger found that the fish in some families fattened during
winter, while others continued to grow longer. Fish in the families that
grow least in winter, mature later. Herbinger found that families bred true
for the trait of fattening – so farmers selecting from these fish might
ensure that they have fish that mature late.

More basic research like this is needed to sort out the complex relationships
between genes and the final characteristics of fish before breeding programmes
will succeed. In Norway, farmers take part in the research. They have invested
38 million kroner, about Pounds sterling 4 million, in the cooperative running
of one of the breeding stations operated by the Institute for Aquaculture
Research. Farmers record the characteristics of offspring from parents with
pedigrees. About two-thirds of the eggs sold to fish farms in Norway last
year had pedigrees.

Expensive facilities, and a large network of farmers to record the results,
enable the Norwegians to set up large, controlled breeding experiments.
Researchers cross males from each family with females from the 260 other
families on the stations, producing a large number of genetic combinations
for testing. Norwegian trout and salmon breed only once a year, which eliminates
variations among fish that hatch at different times of the season.

The Norwegian programme is so large that the statisticians who analyse
the results can sort out the environmental from the genetic effects. Environmental
differences between different pedigrees are likely to be random; as the
number of groups included in the analysis increases, the differences start
to cancel each other out. Genetic differences between fish with different
pedigrees then become visible against the background of environmental variation.
The Norwegian programme has many advantages – but its approach might not
work in developing countries. In the controlled confines of Norwegian experiments,
environmental variations are kept to a minimum. In Thailand, fish breed
several times a year, which means that there are very large environmental
variations between fish that are similar genetically. Fish farmers may keep
records of a few tens of families, but farms are small, and do not pool
their records effectively. As a result, it is impossible to design large-scale
programmes. There are few batches of fish breeding at any one time, too
few to allow breeders to sort out the random environmental differences between
them and distinguish the genetic ones.

Researchers and fish farmers working in these sort of conditions must
adopt a different approach. Thai scientists are experimenting with new strategies.
They use fish with known characteristics as reference strains. These fish
live alongside each experimentally crossed family, under the same environmental
conditions, and serve as a yardstick for improvements in selected characteristics
in the experimental fish. The scientists then select the best fish within
families and breed from them to continue to select the desired trait, rather
than attempting crosses with other families. Fish within the same family
live in the same pond and experience the same environment, so improvements
in subsequent generations are more likely to be the result of genetic factors.

The scientists do not try to sort out environmental from genetic differences
between families until they have enough fish from each family to do a large
number of crosses between families at one time, enough to minimise environmental
differences. In the meantime, they do not have to be so strict about controlling
the fishes’ environment. Doyle believes that the approach has led to improvements
in stock in a shorter time than it would take to set up large-scale experiments.

* * *

WHAT IS IT THAT MAKES A GOOD FISH BETTER?

AQUACULTURE produced 4.7 million tonnes of fish in 1985, some 13 per
cent of the world’s fisheries. According to the UN’s Food and Agriculture
Organisation, the industry is growing at 5.5 per cent per year, taking the
pressure off some of the hard-pressed stocks of wild fish. Many developing
countries are promoting aquaculture enthusiastically. In the Philippines,
fish and shrimp farms produce a large proportion of the country’s protein.
In China, fish farms account for a third of all aquatic produce.

In industrialised countries, farming valuable species such as trout,
salmon and some shellfish, has become an important industry. Output in Norway
and Scotland, both big producers of farmed fish, is expected to treble in
the next decade. As the industry grows, so do some of its problems. Controlling
disease in the confined space of a fish farm is one of the biggest problems.
‘The trouble is that fish live in water,’ said one biologist. That is not
as facetious as it sounds. If a cow becomes sick, you can cut it out of
the herd and isolate it for treatment. Sick fish are harder to spot, and
by the time disease is obvious, it has usually spread.

Losses to viruses and bacteria are large and treatment not very effective.
The easiest way to administer a drug is to add it to the water. The treatment
sometimes encourages other pathogens, by killing competing organisms. It
can also lead to the development of strains of pathogens that are resistant
to drugs. To overcome the risks of disease, which can spread like wildfire
through crowded cages, farmers would like to breed fish that are themselves
resistant to disease.

Biologists know little about the mechanism of resistance in fish, however.
Most efforts to isolate resistant stocks of fish have simply involved exposing
fish to a pathogen and then breeding from the survivors in the hope that
their offspring will be more resistant to infection than the average of
the original population. The offspring, as a rule, do not. The reason might
be that environmental factors, and not genes alone, determine a fish’s ability
to resist a pathogen, and environmental factors can change subtly during
an experiment. The breeder’s job is to design experiments that tease out
the genetic factors and select for them.

Bernard Chevassus, of the National Institute for Agricultural Research
in France, would like to see selection based on new types of screening procedures
that pinpoint resistant individuals without putting them to the ultimate
test of exposure to disease. One such test might be based on the presence
of proteins called HLA antigens that are found on the surface of cells.
The inheritance of these antigens is strictly governed by genes. There is
evidence that links the presence of some of these antigens with resistance
to disease. If that is confirmed, families of resistant fish could be selected
by screening for antigens, rather than observing how groups of fish respond
to infection.

Other possible tests might involve the activity, in the test tube, of
a fish’s slime or serum against disease organisms. Efforts to pin down which
of these factors make a fish more resistant to disease are just beginning.
Scientists still have much to learn about the basic mechanisms of infection
and resistance to disease in fish.

Fish breeders have many goals other than a fast-growing, disease-free
fish. They would like to produce fish that tolerate low levels of oxygen
in the water, that use food more efficiently and that are more docile and
easier to harvest. The biology of these characteristics is likely to be
very complex, and unlikely to be amenable to the sort of genetic screening
that may be possible with disease resistance. For such complex characteristics,
large-scale, statistically designed breeding experiments may be the only
answer.

* * *

JUST A GLEAM IN THE EYE OF A GENETIC ENGINEER

IF fish farmers cannot breed the perfect fish, perhaps genetic engineers
could do better by direct manipulation of a fish’s genetic material. Fish
are difficult subjects for breeders, but they have many advantages for the
genetic engineer. Unlike the eggs of mammals, fish eggs are easy to handle
and to inject with the tiny glass needles that are used to transfer genetic
material. The eggs are large; there are lots of them and they grow easily
in culture. It is surprising, then, that genetic engineers have achieved
so little.

Scientists are just beginning to overcome some of the early difficulties.
One of the most fundamental problems of genetic manipulation of fish is
that their eggs have a very small nucleus. The scientist must inject genes
into the cytoplasm of the egg rather than into the nucleus, which may then
fail to take up the genes.

In some experiments, offspring of injected fish had none of the injected
gene, showing that it was not incorporated. In some cases, the injected
genes are incorporated into the genome of only one of the first few cells
of the fish. As a result, they are expressed in some, but not all of the
fish’s tissues. The fish is called a chimera. The gene may not reach the
particular tissue where it must exert its effect. If it is not the fish’s
egg or sperm, the gene will not be passed to offspring.

After many attempts with different preparations of genes and injection
techniques, however, scientists have successfully added genes to trout,
salmon, catfish, carp and other farmed species. No implanted gene has had
any observable effect on the fish so far. Scientists at the Chinese Academy
of Sciences report that when they insert the genes for growth hormone into
carp, the carp grow bigger. In a similar experiment at Auburn University
in Alabama, Rex Dunham inserted genes for growth hormone from trout into
carp. He reports that the engineered carp grew 20 per cent faster than normal.
In both experiments, the bigger fish were within the normal range of sizes,
so it was not clear if the gene had had a real effect. The genes may not
have much effect because they are inserted with promoters (the sequence
of DNA attached to an injected gene that causes it to be turned on and expressed)
from rats or mice. The search is now on for fish promoters.

Genetic engineering raises the spectre of unnatural ‘altered’ species.
Perhaps optimistically, scientists expect fewer legal and ethical objections
from the public to the manipulation of fish than they have faced with mammals.
Some of the objections concern the release of altered organisms to the environment.
Most countries have strict regulations governing experiments out of doors.
But some fish are certain to escape from fish farms. Once loose, they can
pass their added genes on to their wild relatives through interbreeding
– with unpredictable results.

Some fish can be engineered so that this does not happen. Trout can
be made triploid, containing three instead of two copies of their genes,
with heat shocks or chemical treatments to eggs during their first divis-
ion. Triploid trout are sterile. The technique is also possible with grass
carp, but the eggs of other farmed fish have proved resistant to efforts
to induce triploidy.

The main problem now facing prospective fish engineers is to identify
the genes that fish farmers want in their stock. One overall characteristic
they want is increased size, which may mean increased growth rate, or more
efficient use of food. Growth hormone promotes both. Extra genes for growth
hormone may not mean simple increases in growth, however. The extra hormone
could have unpredictable effects on the growth and eating quality of fish.
Scientists are toying with ideas for several genetic improvements in fish
that might be more straightforward. One involves metallothionein proteins,
which bind heavy metals such as cadmium and mercury. These metals are increasingly
polluting both fresh and sea water. A fish with a greater ability to bind
heavy metals and neutralise them would tolerate higher levels of pollution.

Another target is the gene for ‘antifreeze’ proteins, such as those
made by winter flounder. Like antifreeze in cars, these proteins lower the
freezing point of blood. Such a protein could prevent Atlantic salmon from
freezing to death, as they often do in cages on farms. Scientists have identified
the antifreeze gene but so far have tested it only in fruit flies.

Much of the work of developing libraries of gene sequences for other
experimental animals must be repeated for fish before genetic manipulation
will be profitable. Some biologists argue that it would be better to use
classical techniques of breeding to solve problems such as resistance to
disease before plunging into biotechnology.

Dunham has now reached the stage when he wants to test his fish outside
the laboratory. He has submitted an environmental impact assessment to the
US Department of Agriculture, asking for permission to test fish with introduced
genes for human and fish growth hormones, in small outdoor ponds. The fish
will be isolated by screens on the inlet and outlets for water, and by fences
and bird netting. American scientists think the chances of gaining permission
are promising.

‘We know little about the ecology,’ says Dunham, stressing the need
to move carefully. He also wonders whether starting with growth hormone
will be disappointing. ‘There is so much variation in natural growth rates,’
he says, ‘that if faster growth was advantageous, wild fish would have it
²¹±ô°ù±ð²¹»å²â.’

Next week, two further articles consider the ecological problems that
have accompanied the rise of the fish farming industry.