http://bioscience.oxfordjournals.org April 2017 / Vol. 67 No. 4 •BioScience 357
JASON BAUMSTEIGER AND PETER B. MOYLE
Most extinction literature focuses on prevention and prediction, not assessment. Determination of extinction can be surprisingly complicated,
with diverse approaches and terminology, leading to a gray area within extinction assessment. A series of five gray-extinction categories
(mitigated, regional, native range, wild, and apparent) are provided to address these ambiguities and highlight how extant lineages may be
effectively extinct. For reference, we use freshwater fishes, a group in serious decline throughout the world. Categories are interwoven into a
decision tree to ensure a practical assessment of extinction and maximize conservation effectiveness. To prevent premature declarations, a
waiting period based on generation time (versus a fixed number of years) is proposed. We also explain how extinction is tied to multilineage
and lineage-specific anthropogenic effects and how dependence on artificial selection equates to a form of extinction. Finally, we touch on the
resurrection of lineages and the impact of artificial hybridization and propagation on the extinction process.
Keywords: freshwater fishes, artificial selection, endangered species, triage, conservation
Throughout evolutionary history, millions of lineages
have gone extinct, by chance or because of an inability
to adapt to new environmental conditions (Darwin 1859).
In particular, sudden global changes have led to five major
extinction events (Raup and Sepkoski 1982). Today, we
are experiencing a sixth major extinction event as a result
of global change caused by humans (Dirzo et al. 2014).
Interestingly, most people see extinction as a simple dichot-
omy: Either a species is gone from the Earth, or it is not. In
an era of rapid change, however, extinction can be surpris-
ingly hard to determine. In this article, we address the basic
question: How do you know when a lineage is extinct?
Before any assessment of extinction can be attempted, a
lineage must be formally recognized as distinct. This distinc-
tiveness can include reproductive barriers, genetic signature,
morphometrics or meristics, location, and range, to name
but a few. For most organisms, lineages are distinguished at
the level of species (IUCN 2012). Often, however, lineages
are also are recognized as subspecies, distinct population
segments (DPS; Endangered Species Act of 1973, section 4),
evolutionarily significant units (ESU; Waples 1991), desig-
natable units (SARA; Canada’s Species at Risk Act of 2002),
or management units (MU; Vogler and Desalle 1994). All
can receive the same legal protection as species, depending
on the country. Therefore, to ensure clarity and comprehen-
siveness, we simply refer to any recognized distinct taxo-
nomic group as a lineage.
From a biologist’s perspective, every evolutionary lin-
eage is unique, with a distinct role in the ecosystem(s) of
which it is part (Ehrlich and Mooney 1983). The science
of conservation biology focuses on preventing extinction
of contemporary lineages. In some countries, prevention is
mandated by law, whereas in others, it is a matter of choice.
Regardless, most countries follow the International Union
for Conservation of Nature (IUCN) system, in which lin-
eages are assessed as to their susceptibility to extinction
in the near future: vulnerable, endangered, and critically
endangered (IUCN 2012). But despite the success of this
system at decreasing extinctions (Rondinini et al. 2014),
many lineages continue to decline. The IUCN (2012) also
uses three extinction categories: regionally extinct, extinct
in the wild (known only to survive in culture, captivity, or
as a naturalized population outside their native range), and
globally extinct (i.e., “there is no reasonable doubt that the
last individual has died”). But it is often difficult to say with
certainty when the last individual is gone because most lin-
eages are cryptic at small population sizes, making it difficult
to determine “no reasonable doubt.”
Determination of extinction is not simple. Is a lineage
extinct if a single individual still exists, as has happened
with some trees and tortoises (Cronk 2016)? What if a lin-
eage is wholly reliant on humans, is absent from its native
range, or has no habitat left? What if it has been genetically
modified or hybridized with another lineage? In all such
situations, the evolutionary trajectory of the lineage has
been changed. But is this change enough to consider it to
be extinct? Defining extinction can be a lot like defining a
species because it requires recognizing one lineage as being
distinct from other lineages (de Quieroz 2007). A species is
generally thought of as being one or more populations that
have separated over time from other related populations by
chance and/or natural selection acting under a particular set
of environmental conditions. As the environment changes,
a species must change as well or go extinct. Unfortunately,
BioScience 67: 357–366. © The Author(s) 2017. Published by Oxford University Press on behalf of the American Institute of Biological Sciences. All rights
reserved. For Permissions, please e-mail: firstname.lastname@example.org.
doi:10.1093/biosci/bix001 Advance Access publication 1 March 2017
Downloaded from https://academic.oup.com/bioscience/article/67/4/357/3015934 by guest on 01 November 2021
358 BioScience •April 2017 / Vol. 67 No. 4 http://bioscience.oxfordjournals.org
recent anthropogenic environmental change is occurring at
such a rapid pace that evolutionary change cannot keep up,
at least for longer-lived species, such as most vertebrates.
Therefore, many species persist mainly at the sufferance of
humans: in zoos, in culture, or in the wild in carefully regu-
lated environments. This means, in turn, that frequent deci-
sions must be made that require understanding the nature
Regrettably, extinction of a lineage is not well defined
legally in most places in the world (with some exceptions,
such as SARA in Canada). In the United States, for example,
there is no formal policy for defining a lineage as extinct.
The Endangered Species Act of 1973 (ESA) was written to
prevent extinction and therefore does not specify how to
determine extinction. Legislation in the European Union
and Australia is similar in this regard. However, under
the implementing regulations for the ESA, the managing
agencies (the US Fish and Wildlife Service, USFWS, and
the National Marine Fisheries Service, NMFS) can delist
an endangered lineage on extinction, suggesting a discre-
tionary act rather than a mandatory procedure (50 C.F.R.
§424.11(d)(1)). This decision, however, also requires a lin-
eage be formally listed as threatened or endangered, so it can
be delisted after the normal 5-year review. This procedure
is only an indirect measure of extinction, with no formal
criteria. The IUCN, which includes over 1200 government
and nongovernment organizations (NGOs), generates most
of the endangered species recommendations but has no
listed formal extinction procedure or enforcement power for
declaring extinction, relying on individual countries to draw
their own conclusions.
To help explore the uncertainties or difficulties surround-
ing extinction, it is useful to focus on a particular taxonomic
group. One group with members especially prone to com-
plete extinction is freshwater fishes (Leidy and Moyle 1997,
Moyle etal. 2011, Burkhead 2012). In the twentieth century,
freshwater fishes had the highest extinction rate worldwide
among vertebrates, and conservative estimates project 43–86
extant species will be extinct in North America by 2050
(Burkhead 2012). Furthermore, it is difficult to “prove”
extinction in most freshwater fishes because at low numbers,
they evade capture or detection in places such as deep lakes
or turbid rivers (Jelks etal. 2008). Finally, freshwater fishes
include some of the best examples of conservation-reliant
and captive species, which confound extinction declarations.
To further explore the uncertainties or difficulties sur-
rounding the assessment of extinction, we introduce what
we call gray extinction, a category between traditional IUCN
vulnerable or endangered categorizations and global extinc-
tion. This important classifcation brings to light the many
ways that a lineage may already be “extinct” to some degree
and has strong conservation implications. Multiple catego-
ries are proposed as working criteria to assess the extinction
trajectories of any freshwater fish lineage. In addition, we
have provided a decision tree to document the transition
of a lineage through these categories before reaching global
extinction. And although we focus on freshwater fishes, our
process should be adaptable to any lineage, making it an
acceptable method to assess any extinction.
Human activity affects most lineages in the modern world,
both positively and negatively (Kareiva etal. 2007). These
anthropogenic effects can set a lineage’s trajectory toward
global extinction and range from multilineage to lineage-
specific, with a spectrum of effects in between the extremes.
Multilineage effects are those anthropogenic factors that
affect multiple lineages simultaneously on a habitat, land-
scape, ecosystem or even global scale. They occur as result
of lineages sharing the planet with humans and cannot be
entirely avoided (e.g., climate change). Such effects can also
mimic “natural” effects (e.g., landslides can dam rivers) and
therefore are presumably subject to evolutionary pressures
(i.e., natural selection) consistent with their evolution-
ary history. However, many multilineage anthropogenic
effects are occurring too rapidly, outside the range of nor-
mal environmental variability. Measures taken to alleviate
such effects should, in principle, have a multilineage focus
on conservation up to the planetary scale. This approach
benefits all lineages affected, not just a single lineage. It is
also usually the most cost-effective method over the long
term (Palmer etal. 1997). Approaches for freshwater fishes
might include large-scale habitat restoration, halting the use
of harmful pesticides, improving stream flows, or revers-
ing global warming. The important idea is that a lineage
subject to such conservation measures will still be in its
natural environment, susceptible to natural selection, even
if accelerated, and therefore potentially recoverable if on an
In contrast, lineage-specific effects are those that pri-
marily affect—or at least are perceived to affect—a single
lineage. In most cases, this involves natural selection being
largely replaced by artificial selection (table 1). Artificial
selection can include trait selection, fecundity maximiza-
tion, prezygotic barrier removal, or protection from preda-
tion, among many others. As we discuss later, production
fish hatcheries are classic purveyors of multiple effects on
If exhaustive large-scale, multilineage approaches are
insufficient to alter extinction trajectories, lineage-specific
actions will be required. Individual lineages requiring such
actions become conservation reliant, no longer capable of
surviving on their own and requiring humans to support
their existence (Scott etal. 2005). With this support comes
some unavoidable domestication, because it is unrealistic to
expect managers to fully grasp and replicate every selective
pressure that shapes a lineage—a problem seen time and
time again in fish hatcheries (see below). These actions then
alter a lineage’s evolutionary trajectory from one driven by
natural selection to one driven by artificial selection (deBeer
1958, Akey etal. 2010). Actions to alleviate lineage-specific
problems are purposeful and come with the understanding
Downloaded from https://academic.oup.com/bioscience/article/67/4/357/3015934 by guest on 01 November 2021
http://bioscience.oxfordjournals.org April 2017 / Vol. 67 No. 4 •BioScience 359
that maintaining an altered lineage is better than complete
disappearance. Because lineage-specific effects are generally
additive, ever-increasing artificial selection drives a lineage
further and further from its “natural” roots (Carlson et al.
2007). At some point, such lineages become completely
adapted to artificial habitats (e.g., a fish hatchery or ponds)
and are incapable of maintaining populations in their envi-
ronment of origin. They become, for lack of a better term,
domesticated and extinct as natural lineages.
Recognition of this spectrum of effects on lineages allows
a prioritization system (best to worst case) to be set up for
establishing conservation priorities (Bottrill et al. 2008)
for lineages seemingly headed for global extinction: (a)
Lineages reliant on large-scale multilineage actions still exist
somewhere in their natural environments and will continue
to persist with minimal help for a long time, but ultimately,
they will need the benefits of actions such as the restoration
of natural-flow regimes to rivers. (b) Lineages reliant on lin-
eage-specific actions include two pathways: (1) lineages that
have sufficient genetic diversity so they can eventually be
weaned from reliance on human interference or (2) lineages
that will always require direct management to sustain them,
with evolution taking place, at least in part, through artifi-
cial selection. These lineages are likely to go globally extinct
without continuous action. (c) Lineages for which there is
little or no hope for preventing global extinction, except per-
haps as display populations in zoos or aquaria, constitute the
third category. The application of this prioritization (triage)
scheme requires a general understanding of the varieties of
The varieties of extinction
There have been a number of attempts to deal with impend-
ing extinction on a broad basis, including population viabil-
ity analysis (Boyce 1992, Beissinger and McCullough 2002)
and extensive predictive modeling (Purvis et al. 2000,
Hutchings et al. 2012). One widely used idea is that of
extinction debt, in which diverse lineages persist as declin-
ing populations in habitat so altered or diminished that
eventual extinction is likely, even though the process may
take generations (Tilman etal. 1994, Kuussaari etal. 2009).
These “living-dead” lineages may initially survive but will
gradually disappear, even without any further habitat modi-
fication. Extinction debt is especially likely to apply to
lineages with long generation times or to those already on
the threshold of extinction. Although there may be indi-
viduals present in the environment, the lineage is basically
extinct if extreme measures are not undertaken to restore
viability (Kuussaari etal. 2009, Cronk 2016). A similar idea
is functional extinction, in which individuals are present in
their habitat but are so depressed in numbers that they no
longer play a significant role in the ecological community in
which they are embedded (Şekercioğlu etal. 2004). In both
cases, however, lineages persist with an uncertain timeline to
Typically, a lineage is declared globally extinct if an
informal consensus of experts agrees and an extinction dec-
laration is published, usually in a regional faunal work. But
lineages determined to be extinct by this method are some-
times rediscovered (e.g., Miller Lake Lamprey Lampetra
minima, Loiron et al. 2000; Owens Pupfish Cyprinodon
radiosus, Moyle 2002). To combat this problem, the IUCN
recommends waiting 50 years before declaring extinc-
tion. Additional guidelines related to freshwater fishes are
limited. Harrison and Stiassny (1999) proposed a set of
categories contingent on taxonomic status, effective extinc-
tion date, population decline or environmental threat, and
extinction time frame (1500–1948 or 1948–1998), requiring
information often not available. Jelks and colleagues (2008)
recognized three extinction categories for their review of
extinction in North American fishes: extinct (no documen-
tation for 50 years), possibly extinct (no documentation for
20–50 years), and extirpated in nature (lineage only found
in captive populations). But these categories are only guide-
lines for determining global extinction and do not cover all
As we mentioned previously, a lineage is defined by a very
specific set of criteria (e.g., location, range, appearance, and
life history) that define it as a distinct taxonomic group.
Extinction represents the loss of that distinct group. But
Table 1. Differentiation between multilineage and lineage-specific effects with regard to natural and artificial selective
pressures on certain evolutionary tenets.
Multilineage Effects Lineage-specific Effects
Tenet Natural Selection Artificial Selection
Variation in traits Different combinations of traits throughout
Reduced variation, selection for “desirable” traits
Mating success and fecundity is variable Mating success and fecundity is maximized
Heredity Surviving adults pass on their particular
Artificial mating; genetic modification
Prezygotic barriers; predation; biological
interactions (mutualism, commensalism,
parasitism); limited resources; variable
habitat; coevolution; competition
No prezygotic barriers; no predation or other biological interactions;
supplemental feeding; artificial often uniform habitats; maximized
Downloaded from https://academic.oup.com/bioscience/article/67/4/357/3015934 by guest on 01 November 2021
360 BioScience •April 2017 / Vol. 67 No. 4 http://bioscience.oxfordjournals.org
what if that loss was only partial, in which some of the char-
acteristics were lost but not all? Would the lineage still be the
same? We argue that partial losses represent an intermediate
category between traditional IUCN categories (threatened
or endangered) and global extinction, one we call gray
extinction. This term is both a play on the black–white
dichotomy most perceive for extinction as well as a reminder
of the difficulty in assessing extinction for every situation.
Here, we describe five categories within this gray-extinction
classification, representing an amalgamation of categories
generally found in the scientific literature. The initial cat-
egory, mitigated extinction, is defined by the required use
of lineage-specific effects for maintaining the species (Ex1),
whereas categories 2–4 (Ex2–4) can be arrived at with or
without the impact of lineage-specific effects. Combinations
of both lineage-specific and multilineage effects are possible
and would be represented by two subscripts (e.g., Ex13: miti-
gated and native-range extinction).
Ex1: Mitigated extinction. This applies to lineages that are largely
maintained by artificial selection (lineage-specific effects),
as well as to lineages subject to intentional hybridization or
genetic modification. Also included are conservation-reliant
lineages that depend on continuous or intermittent human
action to maintain viable populations, in which “threats can-
not be eliminated, only managed” (Goble etal. 2012, p. 870).
Ex2: Regional extinction. The lineage is extinct in a geographically
or genetically distinct part of its native range, although it may
be abundant elsewhere. This category is often labeled as “extir-
pation” and can vary by spatial and/or evolutionary scale. For
example, the loss of a DPS, ESU, DU, or MU would be global
extinction but would be regional extinction at the species level.
Ex3: Native-range extinction. The lineage is no longer present in
its native range but has been introduced as a “wild” lineage
successfully outside the native range.
Ex4: Wild extinction. The lineage relies on artificial propaga-
tion for its existence. It is maintained as captive popula-
tions in artificial habitats such as fish hatcheries; it may be
reintroduced into the wild, but such populations are not
Ex5. Apparent extinction. No verified observation exists any-
where despite significant efforts. A waiting period based on
generation time is observed.
Ex6: Global extinction. No verified observation exists any-
where, even after waiting period.
The first four categories are a synthesis of widely used
extinction categories, but with a more stepwise approach.
Our goal is to document the progression of extinction, from
the onset of direct human intervention (mitigated extinction)
to recognition that a lineage only exists in a zoo or captive
breeding facility (wild extinction). At each step, it becomes
more and more difficult to reintroduce a lineage back to
its native habitats, even if conditions are ameliorated. The
ever-increasing conservation reliance necessary to maintain
a long-suffering lineage changes the characters that made the
lineage unique, putting it on an evolutionary trajectory that
presumably reduces its adaptability to conditions in the wild.
This categorization of a lineage can help to determine its
conservation priority when resources are limited. Examples
of each category in fishes are given in table 2.
Categories 5 and 6 occur when no documented individu-
als are found anywhere and de facto extinction is assumed.
Current recommendations suggest waiting 50 years, but no
reason is given as to why this number is appropriate. We
argue that because different lineages have very different gen-
eration times (pupfish, approximately 1 year; sturgeon, 25
or more years; Moyle 2002), the application of a fixed time
period can be excessive in some lineages and insufficient in
others. For example, if a lineage only exists in a single spring
and the spring dries up, a 50-year wait to declare extinction
may be unnecessary. We propose a metric for the waiting
period based on generation time. This is similar to D’Elia
and McCarthy (2010), who recommended using generation
time as a metric for determining extinction risk for species
proposed for listing under the federal ESA. We specifically
recommend that for lineages with generation times of 0–5
years, the declaration should wait ten generations, whereas
those with generation times of 5 or more years should wait
five generations. These metrics are in the spirit of those
already proposed, are conservative enough to ensure no
species are “rediscovered,” and are practical enough to work
with any lineage of fishes.
Decision tree for determining extinction
Our decision tree is designed to differentiate between pre-
and postextinction declarations and different categories
of gray extinction (figure 1). To address these questions,
we propose the establishment of an extinction assessment
committee (EAC), perhaps by a professional society such as
the American Fisheries Society or the American Institute of
Biological Sciences. The EAC would be a mixture of agency,
NGO, and academic biologists who would monitor the sta-
tus of potentially extinct fish lineages. The sole purpose of
this committee would be to review information concerning
the loss of fish lineages and make objective decisions at a
series of checkpoints designated as key transitions in assess-
ing vulnerable or endangered (consistent with IUCN cat-
egories), gray-extinction, and global-extinction status. The
committee would also make decisions related to introgressed
lineages or to applying hybridization to “rescue” genetic
diversity in another lineage. This systematic approach by a
dedicated group could greatly alleviate the ambiguity cur-
rently applied to assessing extinction; it could also facilitate
strategic conservation goals that maximize the recovery
potential of any proposed lineage.
http://bioscience.oxfordjournals.org April 2017 / Vol. 67 No. 4 •BioScience 361
Checkpoint 1. The evaluation of previous or novel studies
forms the basis for recommendations concerning the under-
lying causes of decline or extinction, listing as vulnerable or
endangered, and the potential to alleviate any multilineage
effects. Much of this information is readily available from
groups such as the IUCN or the Committee on the Status of
Endangered Wildlife in Canada (COSEWIC). Conservation
priority can be assigned if needed.
Checkpoint 2. This assessment of extinction includes the follow-
ing questions: Have all realistic methods to reduce multilineage
effects been exhausted? Are lineage-specific direct effects the
only way to continue the lineage? Are funds and managers in
place for a program of conservation reliance? The gray-extinc-
tion category is assigned following the decision tree.
Checkpoint 3. An annual reevaluation of gray extinc-
tion for each lineage is applied. If wild extinction has
been reached, monitoring programs should indicate that
no individuals are found anywhere. Intensive targeted
sampling efforts should be employed to look for small
populations. A waiting period is set, based on genera-
tions, if no individuals are found (apparent extinction).
If recovery efforts are sufficient to escape gray extinction
(see below), a lineage may be returned to the top of the
decision tree to be evaluated, starting as a healthy lineage
Checkpoint 4. Final information is collected and global extinc-
tion is confirmed. If rediscovered before waiting period
ends, the lineage returns to checkpoint 2 for reevaluation.
Table 2. Examples of fish lineages exemplifying different extinction categories.
priority Location Comments
Winter-run Chinook salmon
1a California, United States Conser vation hatchery, artificial flows
Fall-run Chinook salmon
2California, United States Production hatcheries maintain populations but some natural
1b Yangtze River, China Maintained by hatchery; no wild reproduction
2California, United States Extinct in California but widespread in the western United
States and Canada
1a Southeast United States Gone from most of native range but reintroduced from extant
2California, United States Abundant in reser voirs outside native range
Mojave tui chub
1a California, United States In ponds near former range (Mojave River)
Charco Palma pupfish
1b Mexico Home spring dry but four populations in captivity
1b Thailand Popular aquarium fish; no recent records from wild
3Lakes Huron and
Michigan, United States
Last seen 1985
3Alabama, United States Last seen 2007
3Amik Lake, Turkey Amik Lake drained; possibly present in nearby lake (polluted)
but no records
(Sander vitreus glaucus)
3Lake Erie, United States Last collected 1983
New Zealand grayling
3New Zealand Last collected 1930s; declared extinct 1986
3California, United States Last collected 1955
Note: Information from IUCN Red List (2012) and Moyle (2002).
362 BioScience •April 2017 / Vol. 67 No. 4 http://bioscience.oxfordjournals.org
Numbers/habitat limited and/or decreasing?
List as vulnerable or endangered?
Are there multi-lineage effects that can be ameliorated?
Lineage is healthy
Is the lineage found
somewhere in its native
Is the lineage found in the
Is the lineage found
Has the appropriate waiting period been completed?
Are lineage-specific anthropogenic effects being
Cs – Conservation Success
Figure 1. A decision tree to assess the extinction progression of a lineage. The checkpoints represent important decision
nodes based on the best available information, and the boxes represent extinction categories. Following checkpoint 2,
two trajectories are indicated, one reflecting decreasing habitat availability (left) and the second reflecting increasing
conservation reliance and decreasing habitat availability (right).
http://bioscience.oxfordjournals.org April 2017 / Vol. 67 No. 4 •BioScience 363
Relevance to conservation
The differences between applying multilineage and lineage-
specific methods to avert extinction has considerable finan-
cial and ecological ramifications. Multilineage methods
benefit all lineages, with the focus on restoring the habitat,
not the lineage. Costs may be substantial but often have
anthropogenic benefits (ecosystem services). Once a system
is restored, it presumably requires only traditional manage-
ment expenditures (Odling-Smee 2005). In contrast, apply-
ing lineage-specific methods leads to artificial selection and
conservation reliance (Lorenzen etal. 2012); every lineage
on an extinction trajectory is treated as unique and must be
supported independently. In addition, lineages undergoing
lineage-specific management can rarely fulfill their former
role in a natural system (Ehrlich and Mooney 1983). It may
initially seem cost effective, but as more and more lineages
approach extinction, costs will be overwhelming. However,
even for lineages dependent on lineage-specific methodolo-
gies, their natural habitats could still be restored if reintro-
duction into the wild in some form is likely to be successful.
In the event that conservation reliance can be ended and/
or natural habitat restored, can extinction (categories 1–5)
be reversed? The answer, unfortunately, is yes and no. The
restoration of natural habitat to support a self-sustaining
lineage subject to natural selection would constitute the
recovery of a lineage. The recent removal of a number of
dams (e.g., Elwha Dam) has the potential for this kind of
recovery of certain fishes (O’Conner et al. 2015, Thornton
et al. 2016). However, the lineage would likely have been
altered through artificial selection, potentially removing
lineage-specific identifiers such as courtship behavior or
unique genetic alleles. Therefore, the current lineage may
be different from its wild predecessor, although just how
different would depend on the number of individuals in the
lineage, the amount of artificial selection applied, and the
length of time the selection was applied. Therefore, there is
no direct answer to this question. Our approach is to leave
this difficult decision in the hands of the EAC to be handled
on a case by case basis, with the understanding that in order
to be a conservation success (figure 1), the lineage must be
completely free of human reliance and sufficiently stable
so as to be self-sustaining. We would also recommend the
lineage be labeled “Cs” (similar to endangered, Ex2, etc.).
This would serve the dual purpose of promoting success-
ful conservation efforts and identifying those lineages that
experienced gray extinction.
Resurrection? “Jurassic Park” lineages
As technology continues to improve, biologists are faced
with a difficult question: Do we bring back ancestral or
recently extinct lineages (Friese and Marris 2014, Seddon
etal. 2014)? This avenue has interesting possibilities along
with unprecedented complications and fears (Sandler 2014).
The “resurrection” of an extinct lineage requires using
DNA as a blueprint in one of two ways: pure ancient DNA,
in which a lineage is produced using only DNA from the
known lineage, or mixed DNA, in which an incomplete
genome requires using closely related lineages or known
“conserved regions” universal to freshwater fishes for com-
pletion. On the basis of criteria proposed in this article,
these lineages would be considered domesticated and fall
under the mitigated category of extinction. If the lineages
get beyond being treated as novelties and are proposed for
introduction into the wild, we recommend they should be
treated as a potentially invasive species. An exception might
be given to recently globally extinct lineages in which the
ecosystem may still support habitats recently vacated by
the lineage. But timing would be everything, because other
lineages will quickly use resources that once supported the
lost lineage. Ideally, “resurrection” would be the ultimate last
resort, occurring when a lineage reaches apparent or global
Hybridization and/or genetic modification
Conceptual ideas related to artificial and natural hybridiza-
tion are similar to those for extinction. Allendorf and col-
leagues (2004) argued that any introgression from artificial
hybridization equals lineage loss. In contrast, Campton and
Kaeding (2005) contended that if a lineage looks and behaves
like a true lineage, it should be considered one. So how much
introgression is allowable before the lineage is considered a
novel lineage or extinct? What if introgression levels are high
(more than 50%), but a unique lineage-specific gene or allele
is still present? Even more challenging is directed hybrid-
ization as a means of introducing genetic diversity into a
lineage to avert global extinction (Levins 2002, Harbicht
etal. 2014). This is similar to genetic modification, in which
specific genes are targeted artificially to improve traits such
as parasite resistance or growth rates (Hedrick 2001). We
would argue, except in cases of natural hybridization, that
all of the above represent artificial selection and a lineage-
specific impact of humans on fish lineages; they are there-
fore a form of extinction, in this case mitigated extinction
(Ex1). Even though lineages are not conservation reliant per
se, their genomes are permanently altered by humans. But
unlike Allendorf and colleagues (2004), who suggested that
any artificially introgressed population should be eradicated,
we think a lineage that has some resemblance to the original
lineage is superior to none at all, especially if it exists as a
population in the wild that is subject to natural selection.
Devils Hole pupfish: Should we hybridize?
An example of the controversy surrounding the use of
directed hybridization to prevent extinction is the Devils
Hole pupfish (DHP) Cyprinodon diabolis, the most endan-
gered fish in the world (Pister 1990, Martin CH etal. 2016).
Living in Nevada, this lineage has been isolated for tens of
thousands of years (Sağlam etal. 2016) and reduced to 30
individuals at times. Artificial feeding and separate captive
populations are currently ongoing while hybridization (with
364 BioScience •April 2017 / Vol. 67 No. 4 http://bioscience.oxfordjournals.org
nearby Ash Meadows Amargosa pupfish, C. nevadensis)
has been tested as a possible method to combat inbreeding
(Martin AP et al. 2012). A small population still exists in
Devils Hole, although it would probably be globally extinct in
the absence of human protection of its habitat (e.g., through
fencing or lawsuits). Would DHP be regarded as extinct
(Ex6), or perhaps as a mitigated extinction (Ex4), if all indi-
viduals had a modified genome that enabled them to persist
in their tiny habitat, as long as they looked and behaved like
DHP? Would they then lose the protection afforded them
by the federal Endangered Species Act of 1973? As of the
writing of this article, DHP are still a natural population in
their natural habitat and not reliant on humans for survival.
Therefore, they would not qualify as mitigated extinction.
Managers will ultimately have to decide whether hybridiza-
tion in, the introduction of captive-bred fish to, or reliance
on human-provided food in Devils Hole would change their
status to some form of extinction.
Salmon and steelhead: Do hatcheries cause
In many countries, large-scale fish culture, especially for
salmon and trout (Salmonidae), became established to miti-
gate for dams and to increase availability of desirable fish.
Much of this aquaculture was designed to support capture
fisheries by rearing fish through embryo and juvenile stages,
when mortality is highest, before releasing them into the
wild. This strategy has been enormously successful, with
high percentages of salmon and trout caught along the
Pacific Coast now being of hatchery origin. The problem
with hatchery fish is that they quickly become domesticated.
Christie and colleagues (2016) found that steelhead rainbow
trout (Oncorhynchus mykiss) showed genetic-based adapta-
tions to the hatchery environment after just one generation.
This problem has been overcome in part by releasing juve-
niles in increasingly large numbers into the wild to over-
come high postrelease mortality. Large numbers of hatchery
fish returning or straying impede and hybridize with wild
spawners, greatly reducing natural production and increas-
ing reliance on hatcheries. Bowlby and Gibson (2011) found
that hatcheries could thus reduce extinction probabilities in
Atlantic salmon (Salmo salar) populations in the short run
but make extinction more likely in the long run.
In California’s Sacramento–San Joaquin River basin, arti-
ficial propagation resulted in a genetically uniform popula-
tion of Central Valley (CV) steelhead, listed as a threatened
DPS under the ESA and treated as if it were a distinct lineage.
However, during the 1950s, CV steelhead were hybridized in
hatcheries with north-coast steelhead, another genetically
distinct lineage. These fish were brought in to “improve” the
fishery with larger fish, irrevocably altering allele frequen-
cies in all CV steelhead. To complicate things further, steel-
head and resident (nonmigratory) rainbow trout typically
have two distinct life history patterns within a single lineage.
Today, the Sacramento River and its tributaries support
populations of resident rainbow trout that are large enough
to support fisheries; these trout are genetically identical to
CV steelhead but rarely produce individuals that go to sea.
Most “true” steelhead (i.e., anadromous rainbow trout) in
the river are of hatchery origin because hatcheries select for
steelhead life history (Zimmerman and Reeves 2000). This
raises the questions: Are CV steelhead effectively extinct
(mitigated extinction, Ex4)? Is a return to an anadromous
population subject to natural selection possible? The answer
to the first question depends on continued defining of CV
steelhead as a separate lineage when the actual lineage con-
sists of both steelhead and resident trout, with different life
histories. The issue is made even more complicated because
a hybrid lineage has replaced the original lineage and is
quite successful as resident trout. Therefore, the answer to
the first question is that steelhead cannot be declared extinct
as a lineage because they are a life-history alternative, not
a lineage. The lineage to which they belong is in no danger
of extinction. The answer to the second question is yes,
because resident hybrid rainbow trout have the capacity to
genetically switch to the steelhead life history if conditions
are right in the river, estuary, and ocean (Zimmerman and
Reeves 2000, Hayes etal. 2012).
A better example of a hatchery-dependent (Ex4) lineage is
the Sacramento winter-run Chinook salmon (Oncorhynchus
tshawytscha). Listed as an ESU under the ESA, large num-
bers of these salmon are reared in a conservation hatchery.
Unlike production hatcheries used to support fisheries, con-
servation hatcheries are designed to produce wild-type fish,
using careful genetic monitoring to reduce domestication
(Winship etal. 2014). Shasta Dam on the Sacramento River
denies access to historical spawning grounds upstream,
so most of the population is maintained by the hatchery
and by spawning below the dam. Flows below the dam are
regulated by cold-water releases, and there is active gravel
augmentation to create spawning habitat. A recent drought
caused “natural” spawning to fail 2 years in a row, making
the population entirely reliant on the conservation hatchery.
Therefore, Sacramento winter-run Chinook salmon is a
good example of mitigated extinction (Ex4) that will require
continuous human intervention to maintain the lineage, if
on a somewhat different trajectory.
Overall, these examples suggest that hatcheries, which pro-
duce millions of salmon and steelhead each year in the Pacific
Northwest, have a major impact on the nature of regional
linages (ESUs, DPSs); this impact can lead to mitigated
extinctions (Ex4) in which lineages will have to be maintained
by artificial means in perpetuity. The release of millions of
hatchery salmonids into the rivers used by wild fish can
replace distinct wild lineages adapted for natural conditions
with lineages maintained through artificial selection. There is
considerable potential for such practices to ultimately lead to
global extinction (Ex6) of distinct salmonid lineages.
Delta smelt: Headed for extinction
The initial impetus for this article is the possible extinc-
tion of delta smelt, Hypomesus transpacificus. This small
http://bioscience.oxfordjournals.org April 2017 / Vol. 67 No. 4 •BioScience 365
planktivorous fish has a 1-year life cycle and is endemic
to the Sacramento–San Joaquin Delta, the central node for
much of California’s water-supply system (Moyle 2002).
Delta smelt is listed as an endangered species and in recent
years has nearly disappeared from both general and focused
fish surveys (Moyle etal. 2016). Extinction in the wild (Ex4)
or global extinction (Ex6) is likely in the near future and has
major implications for water management. However, mak-
ing the final determination with complete certainty will be
difficult because it is a small fish that can move up and down
the estuary and can have local resident populations. There
are two captive populations in which individuals are reared
through their entire life cycle, but they need a continual
influx of wild fish to maintain genetic diversity (Moyle etal.
2016). However, they are currently (as of 2016) not planted
in the wild. If delta smelt disappeared from all fish surveys,
by our criteria, a finding of wild extinction (Ex4) should
occur after 10 years. But consideration would then have to
be given to the two captive populations and their potential
for both domestication and reintroduction into the wild. As
the salmonid example illustrates, domestication can occur
very rapidly in fish (e.g., Lorenzen etal. 2012). Delta smelt
have a 1-year generation time, so the potential for successful
reintroduction into the wild will diminish with each pass-
ing year. If a reintroduction program allows some smelt to
complete their life cycle in the wild but reproductive success
is low and requires continuous input of hatchery-reared
fish, then the lineage will fit into the mitigated extinction
category (Ex1); however, it would be on a different evolu-
tionary trajectory from wild “natural” delta smelt, especially
if the wild population was too small to allow for capture of
individuals to supplement hatchery populations. If such a
program fails, delta smelt would then be considered globally
extinct (Ex6) after 10 years (although Ex4 designation might
be politically more acceptable if a domesticated population
continued to exist, displayed in public aquaria). Delta smelt
therefore illustrate how extinction criteria can help manag-
ers decide on management goals. What level of extinction is
The ongoing extinction crisis requires that extinction be
dealt with systematically, with priorities established for the
use of limited funds available to prevent global extinction
of distinct lineages. Lineages no longer subject to natural
selection in their native habitats can be regarded as suffer-
ing gray extinction, as can lineages dominated by artificial
selection. However, as lineages of hatchery salmonids illus-
trate, lineages totally dependent on aquaculture can persist
and replace natural lineages, at least visually and sometimes
in fisheries. Global (complete) extinction is often hard to
determine, especially in groups such as fishes, so guidelines
for the determination of such extinctions are needed, as
we propose here. Therefore, our decision tree allows for a
straightforward assessment of the extinction of any lineage
and represents a step in the right direction. Although we can
hope extinctions in freshwater fishes and other organisms
will be less than anticipated, this seems unlikely. It is there-
fore time to identify extinction in a more realistic manner.
JB was funded through a postdoctoral research fellowship
from the UC Davis Center for Watershed Sciences. We thank
the members of the Moyle Laboratory and many others who
listened to our presentations as ideas developed and gave us
welcome critiques and questions.
Akey JM, Ruhe A, Akey DT, Wong AK, Connelly CF, Madeoy J, Nicholas
TJ, Neff MW. 2010. Tracking footprints of artificial selection in
the dog genome. Proceedings of the National Academy of Sciences
Allendorf FW, Leary RF, Hitt NP, Knudsen KL, Lundquist LL, Spruell P.
2004. Intercrosses and the US Endangered Species Act: Should hybrid-
ized populations be included as westslope cutthroat trout? Conservation
Biology 18: 1203–1213.
Beissinger SR, McCullough DR. 2002. Population Viability Analysis.
University of Chicago Press.
Bottrill MC, etal. 2008. Is conservation triage just smart decision making?
Trends in Ecology and Evolution 23: 649–654.
Bowlby HD, Gibson AJF. 2011. Reduction in fitness limits the useful dura-
tion of supplementary rearing in an endangered salmon population.
Ecological Applications 21: 3042–3048.
Boyce, MS. 1992. Population viability analysis. Annual Review of Ecology
and Systematics 23: 481–506.
Brook BW, O’Grady JJ, Chapman AP, Burgman MA, Akçakaya HR,
Frankham R. 2000. Predictive accuracy of population viability analysis
in conservation biology. Nature 404: 385–387.
Burkhead NM. 2012. Extinction rates in North American freshwater fishes,
1900–1910. BioScience 62: 798–808.
Campton DE, Kaeding LR. 2005. Westslope cutthroat trout, hybridiza-
tion, and the US Endangered Species Act. Conservation Biology
Carlson SM, Edeline E, Asbjørn Vøllestad L, Haugen T, Winfield IJ, Fletcher
JM, Ben James J, Stenseth NC. 2007. Four decades of opposing natural
and human‐induced artificial selection acting on Windermere pike
(Esox lucius). Ecology Letters 10: 512–521.
Christie MR, Marine ML, French RA, Blouin MS. 2012. Genetic adapta-
tion to captivity can occur in a single generation. Proceedings of the
National Academy of Sciences 109: 238–242.
Cronk Q. 2016. Plant extinctions take time. Science 353: 446–447.
Darwin C. 1859. On the Origin of Species by Means of Natural Selection,
or the Preservation of Favoured Races in the Struggle for Life. John
DeBeer G. 1958. Evolution by Natural Selection. Cambridge University Press.
D’Elia J, McCarthy S. 2010. Time horizons and extinction risk in endan-
gered species categorization systems. BioScience 60: 750–758.
De Queiroz K. 2007. Species concepts and species delimitation. Systematic
Biology 56: 879–886.
Dirzo, R, Young HS, Galetti M, Ceballos G, Case NJB, Collen B. 2014.
Defaunation in the Anthropocene. Science 345: 401–406.
Ehrlich PR and Mooney HA. 1983. Extinction, substitution, and ecosystem
services. BioScience 33: 248–254.
Friese C, Marris C. 2014. Making de-extinction mundane? PLOS Biology 12
(art. e1001825). doi:10.1371/journal.pbio.1001825
Goble DD, Wiens JA, Scott JM, Male TD, Hall JA. 2012. Conservation-
reliant species. BioScience 62: 869–873.
Harbicht A, Wilson CC, Fraser DJ. 2014. Does human‐induced hybridiza-
tion have long‐term genetic effects? Empirical testing with domesti-
cated, wild and hybridized fish populations. Evolutionary Applications
366 BioScience •April 2017 / Vol. 67 No. 4 http://bioscience.oxfordjournals.org
Harrison JJ, Stiassny MLJ. 1999. The quiet crisis: A preliminary listing of
the freshwater fishes of the world that are extinct or “missing in action.”
Pages 271–331 in McPhee RDE, ed. Extinctions in Near Time: Causes,
Contexts, and Consequences. Kluwer Academic.
Hayes, SA, Hanson CV, Pearse DE, Bond MH, Garza JC, MacFarlane RB.
2012. Should I stay or should I go? The influence of genetic origin
on emigration behavior and physiology of resident and anadromous
juvenile Oncorhynchus mykiss. North American Journal of Fisheries
Management 32: 772–780.
Hedrick PW. 2001. Invasion of transgenes from salmon or other genetically
modified organisms into natural populations. Canadian Journal of Fish
and Aquatic Science 58: 841–844.
Hutchings JA, Meyers RA, Garcia VB, Lucifora LA, Kuparinen A. 2012. Life-
history correlated of extinction risk and recovery potential. Ecological
Applications 22: 1061–1067.
[IUCN] International Union for Conservation of Nature. 2012. IUCN Red
List Categories and Criteria, Version 3.1, 2nd ed. IUCN Species Survival
Jelks HL, et al. 2008. Conservation status of imperiled North American
freshwater and diadromous fishes. Fisheries 33: 372–407.
Johnson RC, Weber PK, Wikert JD, Workman ML, MacFarlane RB, Grove
MJ, Schmitt AK. 2012. Managed metapopulations: Do salmon hatchery
“sources” lead to in-river “sinks” in conservation? PLOS ONE 7 (art.
Kareiva P, Watts S, McDonald R, Boucher T. 2007. Domesticated nature:
Shaping landscapes and ecosystems for human welfare. Science
Kuussaari M, etal. 2009. Extinction debt: A challenge for biodiversity con-
servation. Trends in Ecology and Evolution 24: 564–571.
Leidy RA, Moyle PB. 1997. Conservation status of the world’s fish fauna: An
overview. Pages 187–227 in Fiedler PA, Karieva PM, ed. Conservation
Biology for the Coming Decade. Chapman and Hall.
Levins D. 2002. Hybridization and extinction: In protecting rare species,
conservationists should consider the dangers of interbreeding, which
compound the more well-known threats to wildlife. American Scientist
Lorenzen K, Beveridge M, Mangel M. 2012. Cultured fish: Integrative biol-
ogy and management of domestication and interactions with wild fish.
Biological Reviews 87: 639–660.
Lorion CM, Markle DF, Reid SB, Docker MF. 2000. Redescription of the
presumed-extinct Miller Lake lamprey, Lampetra minima. Copeia 2000:
Martin AP, Echelle AA, Zegers G, Baker S, Keeler-Foster CL. 2012.
Dramatic shifts in the gene pool of a managed population of an endan-
gered species may be exacerbated by high genetic load. Conservation
Genetics 13: 349–358.
Martin CH, Crawford JE, Turner BJ, Simons LH. 2016. Diabolical survival
in Death Valley: Recent pupfish colonization, gene flow and genetic
assimilation in the smallest species range on earth. Proceedings of the
Royal Society B 283 (art. 20152334).
Moyle PB. 2002. Inland Fishes of California. University of California Press.
Moyle PB, Katz JVE, Quiñones RM. 2011. Rapid decline of California’s
native inland fishes: A status assessment. Biological Conservation
Moyle PB, Brown LR, Durand JR, Hobbs JA. 2016. Delta smelt: Life history
and decline of a once abundant species in the San Francisco Estuary. San
Francisco Estuary and Watershed Science 14 (art. 6). (9 January 2016;
O’Connor JE, Duda JJ, Grant GE. 2015. 1000 dams down and counting.
Science 348: 496–497.
Odling-Smee L. 2005. Conservation: Dollars and sense. Nature
Palmer MA, Ambrose RF, Poff NL. 1997. Ecological theory and community
restoration ecology. Restoration Ecology 5: 291–300.
Pister EP. 1990. Desert fishes: An interdisciplinary approach to endan-
gered species conservation in North America. Journal of Fish Biology
Purvis A, Gittleman JL, Cowlishaw G, Mace GM. 2000. Predicting extinc-
tion risk in declining species. Proceedings of the Royal Society of
London B 267: 1947–1952.
Raup DM, Sepkoski JJ. 1982. Mass extinctions in the marine fossil record.
Science 215: 1501–1503.
Rondinini C, Marco M, Visconti P, Butchart SH, Boitani L. 2014. Update
or outdate: Long‐term viability of the IUCN Red List. Conservation
Letters 7: 126–130.
Sağlam IK, Baumsteiger J, Smith MJ, Linares-Casenave J, Nichols AL,
O’Rourke SM, Miller MR. 2016. Phylogenetics supports an ancient
common origin of two scientific icons: Devils Hole and Devils Hole
pupfish. Molecular Ecology 25: 3962–3973.
Sandler R. 2014. The ethics of reviving long extinct species. Conservation
Biology 28: 354–360.
Scott JM, Goble DD, Wiens JA, Wilcove DS, Bean M, Male T. 2005.
Recovery of imperiled species under the Endangered Species Act: The
need for a new approach. Frontiers in Ecology and the Environment
Seddon PJ, Griffiths CJ, Soorae PS, Armstrong DP. 2014. Reversing defau-
nation: Restoring species in a changing world. Science 345: 406–411.
Şekercioğlu ÇH, Daily GC, Ehrlich PR. 2004. Ecosystem consequences
of bird declines. Proceedings of the National Academy of Sciences
Thornton EJ, Duda JJ, Quinn TP. 2016. Influence of species, size and rela-
tive abundance on the outcomes of competitive interactions between
brook trout and juvenile coho salmon. Ethology Ecology and Evolution
Tilman D, May RM, Lehman CL, Nowak MA. 1994. Habitat destruction
and the extinction debt. Nature 371: 65–66.
Vogler AP, Desalle R. 1994. Diagnosing units of conservation management.
Conservation Biology 8: 354–363.
Waples RS. 1991. Pacific salmon, Oncorhynchus spp., and the definition of
“species” under the Endangered Species Act. Marine Fisheries Review
Winship AJ, O’Farrell MR, Mohr MS. 2014. Fishery and hatchery effects on
an endangered salmon population with low productivity. Transactions
American Fisheries Society 143: 957–971.
Zimmerman CE, Reeves GH. 2000. Population structure of sympatric anad-
romous and nonanadromous Oncorhynchus mykiss: Evidences from
spawning surveys and otolith microchemistry. Canadian Journal of Fish
and Aquatic Sciences 57: 2152–2162.
Jason Baumsteiger (email@example.com) is a postdoctoral research fel-
low at the Center for Watershed Services and Peter Moyle (pbmoyle@ucdavis.
edu) is a distinguished emeritus professor of the Department of Fish Wildlife
and Conservation Biology at the University of California, Davis. JB studies
the evolution of freshwater fishes using genetic or genomic approaches, and
PBM studies the ecology and conservation of freshwater and estuarine fishes.