Hypothesis Testing and the Burden of Proof

1
Artificial propagation of salmonids
2
Hatcheries play a large role in the management,
ecology, and evolution of Pacific salmon. Why
were/are they built? What are the assumptions
behind their operation, and what are the concerns?
3
Hatchery production
coho salmon
0 50 100 150 200
Millions of Fish
Pacific Northwest British Columbia Alaska
1930
1990
1900
1960
Release Year
4
Hatchery production
0 100 200 300 400 500
1900
1930
1960
1990
Release Year
5
Hatchery production Columbia River
Millions of salmonids released
0 50 100 150 200 250
1950 1960 1970 1980 1990
Release Year
Total releases of salmonids (millions) in the
Columbia River basin for release years 1950-1998.

6
Columbia Basin Salmon and Steelhead Hatcheries
7
Estimated of adult wild steelhead
100 80 60 40 20 0
Percent Wild
AK B.C WA Col. R. OR CA
Light 1987
8
Hatcheries

• Why build a hatchery?
• Mitigate for lost habitat (Columbia River)
• Create or enhance a fishery (SE Alaska)
• Supplement a weak or unstable run
• Preserve and restore(?) endangered run

Issaquah Salmon Hatchery, King County, Washington
c.a. 1935 UW library archives
9
Implicit assumptions

• The hatchery by-passes the limiting life-history
  stage other habitats are under-utilized or are
  not limiting
• There are no other biological interaction between
  hatchery-produced and wild fish (e.g. diseases,
  genetic interaction, predation, ect.)
• The survival rate of hatchery-produced fish is
  similar to wild fish.
• The hatchery does not directly of indirectly
  affect habitat needed by wild fish.
• The fisheries can accommodate the more productive
  hatchery population without over fishing the wild
  one.
• The hatcherys productivity will not diminish.

10
Why do hatcheries work so well? They
short-circuit the limiting life history stage
For most fishes, the embryonic and early larval
stages have very high mortality. Many species
are difficult to breed in captivity or their
young are difficult to feed. However, salmon are
easy to breed, the embryos are easy to incubate,
and the juveniles feed readily.
In streams, most salmon die between fertilization
and emergence from the gravel (often 80-90). It
is easy to improve on the survival rates of wild
salmon. Nature is inefficient we can do
better.
Chilko River sockeye salmon Average 8.3
11
Hypothetical coho salmon life tables
Hatchery
Wild
12
Hypothetical coho salmon life tables
Hatchery
Wild
13
The first problem with hatcheries How do we
catch their surplus without overfishing healthy
but less productive wild populations?
14
Hatcheries solution or problem?

• Simulated Effects of stocking and Fishing on Lake
  Trout
• Stocking of non-native lake trout at moderate to
  high exploitation rates caused loss of the
  recipient population, even when the stocked fish
  reproduce, their progeny mask the loss of the
  wild stock. At exploitation and stocking rates
  typical of lake trout populations in Ontario, the
  wild stock was replaced by the hatchery stock in
  a few generations. Native stocks having weak
  recruitment are least resistant to displacement
  by hatchery stocks and are also the most likely
  to be subjected to stocking.

Evans and Willox. 1991. Can. J. Fish. Aquat.
Sci. 48 (Suppl. 1) 134-147
15
The second problem Behavioral and ecological
interactions between wild and hatchery produced
salmon

• Implicit assumptions in most hatchery programs
• The hatchery by-passes the limiting life history
  stage or habitat, and so increased production of
  juveniles leads to increased production of adults
• There are no important biological interactions
  between wild and hatchery fish that would limit
  overall production (i.e., hatchery fish are
  simply added to the wild ones)

16
Behavioral and ecological interactions between
wild and hatchery produced salmon

• Competition
• Hatchery fish are usually larger and may be more
  aggressive, giving them advantages over wild fish
• Wild fish may already have territories when
  hatchery fish are released, giving them an
  advantage
• Large size may give hatchery fish a survival
  advantage but life in the hatchery might make
  them more vulnerable to predation
• If the stream is at carrying capacity, more
  smolts will not be produced and the wild
  population may be reduced.

Atsushi Sakurai
J. Rhodes
17
Behavioral and ecological interactions between
wild and hatchery-produced salmon

• Predation
• Smolts may prey on wild salmon of other species,
  or smaller conspecifics.
• Hatchery produced fish might help swamp predators
  or might concentrate them.
• Responses of predators, and population
  consequences may be complex.

18
Survival rates of hatchery fish are often lower
than those of wild fish, and environmental
conditions play a large role
Smolt to adult survival of Snake River chinook
Raymond (1988) NAJFM
19
The survival of hatchery-released fish is usually
lower than that of wild fish, even though the
hatchery fish are larger. Why?

• Predator avoidance
• Feeding behavior
• Genetics
• Physiology
• Diet
• Release date
• ????

20
Releasing more smolts does not produce more adults
Oregon Production Index coho salmon
Smolts
smolts (millions)
returns (Millions)
Adults
1960 1965 1970 1975 1980
1985
Year of smolt migration
Trends in smolt entering the coastal environment
and subsequent adult returns for the same year
class (Brodeur, R. 1990 PhD. diss.).
21
The third problem Genetic effects

• Domestication Selection
• Genetic divergence of wild and hatchery
  populations
• Lower fitness of hatchery-produced fish in the
  wild
• Do we make wild and hatchery populations
  isolated, or combined?
• Hybridization with other species

22
Wild female salmon must select, prepare and
defend nest sites and choose mates, males must
choose and compete for access to females.
23
The hatchery staff cannot know which salmon would
have chosen the best nest sites or otherwise been
successful in reproducing, so there is a
relaxation of selection on these traits. There
may also be deliberate or inadvertent selection
on other traits such as date of maturation, size,
etc. In addition to these processes affecting
adults, there is also selection for juveniles
that adapt to confined conditions. Together,
these processes are referred to as domestication
selection.
24
Hatcheries reproductive success

• Reproductive success of native and non-native,
  hatchery produced steelhead (Kalama River)

25
Hatcheries reproductive success

• These effects were cumulative and Leider et al.
  concluded
• the naturally spawning wild summer steelhead we
  examined produced from 7.8 to 9.3 times more
  adult offspring than did their naturally spawning
  hatchery counterparts. However, more than 40
  ...of the naturally produced adult summer
  steelhead were the direct offspring of hatchery
  spawners due to the preponderance in the
  spawning escapement.
• Leider, Hulett, Loch and Chilcote. 1990.
  Aquaculture 88 239-252

26
Hatcheries solution or problem?
Analysis of Oregon steelhead populations
indicated that the higher the proportion of
hatchery steelhead, the lower the productivity of
the population. Below about 50 wild fish, the
populations failed to replace themselves
(estimated mean /- 95 confidence intervals).
2.0
1.5
Recruits produced per spawner
1.0
0.5
0.0
0.4
0.6
0.8
0.2
1.0
Proportion of hatchery fish in the parental
population
Chilcote 2002
27
Hatchery and wild salmon two options

• Integrated hatchery make wild and hatchery fish
  as similar as possible, by breeding some wild
  fish in the hatchery and having some hatchery
  fish spawn in the river in each generation, and
  manage them as a single population.
• Segregated hatchery make the hatchery fish as
  different from the wild fish as possible (e.g.,
  breeding date), limit interbreeding as much as
  possible, and have different fishing rates on the
  two components.

28
Hatcheries Role in Restoration and Enhancement
of Salmon Populations

• Why build a hatchery?
• High survival rates at early life history stages
  improve net productivity over a wild population.
• Nature is inefficient we can do better.
• Objective, and underlying assumptions
• Reasons to question the assumptions
• Concerns
• Short-term fishery management and how to handle
  success
• Intermediate-term behavioral and ecological
  interactions
• Long-term genetic consequences within the
  hatchery and the entire system

29
Hatcheries

• Objective Add hatchery-produced salmon and
  trout to the wild populations without diminishing
  them, resulting in proportional increases in
  adults.
• Key assumptions
• The hatchery by-passes the limiting life-history
  stage other habitats are under-utilized or are
  not limiting
• There are no other biological interaction between
  hatchery-produced and wild fish (e.g. diseases,
  genetic interaction, predation, ect.)
• The survival rate of hatchery-produced fish is
  the same as the of wild fish.
• The hatchery does not directly of indirectly
  affect habitat needed by wild fish.
• The fisheries can accommodate the more productive
  hatchery population without over fishing the wild
  one.
• The hatcherys productivity will not diminish.
• Violation in these assumptions will erode either
  the success of the hatchery program, the
  neighboring wild populations, or both,
  diminishing the overall success.

30
Hypothesis Testing and the Burden of Proof

• H0 Null Hypothesis usually meaning no
  effect
• H0 is either true or false, and your data either
  allow you to reject it, with some chance that it
  is actually true, or you cannot reject it, but in
  fact it could be false

31
Hatcheries solution or problem?
Predicted steelhead recruitment curves for 6
scenarios differing only in the proportion of
hatchery fish in the natural spawning population
(Ph).
Ph 0.0 Ph 0.1 Ph 0.3 Ph 0.5 Ph 0.7 Ph
1.0
Recruits
0 500 1000 1500 2000 2500
0 500 1000 1500 2000 2500
3000
Spawners
Ph proportion of hatchery fish in spawning
population
Chilcote 2002
32
Hatcheries and Columbia River salmon
A. Pre-development 7 - 11 million salmon per
year
B. Recent years 2.2 million salmon per year
Distribution of Columbia River salmon above and
below the site of Bonneville Dam. Area of
circles is proportional to estimated population
sizes. Source NW Power Planning Council