Pacific Salmon, Oncorhynchus spp., and the Definition of "Species" Under the Endangered Species Act

ROBIN S. WAPLES

Introduction (NMFS) and the U.S. Fish and Wildlife provide guidance at three critical stages Service (FWS), three policy position in ESA evaluations: Determination of In conjunction with a review of the papers were drafted for the consideration what constitutes a "species" under the biological status ofPacific salmon, 1 On­ ofColumbia River salmon under the En­ Act (and what, therefore, may merit pro­ corhynchus spp., initiated in 1978 by the dangered Species Act (ESA or "the tection), determination ofthresholds for National Marine Fisheries Service Act' '). The draft papers were intended to listing as threatened or endangered, and

I The term" Pacific salmon" has traditionally re­ the genus Sa/mo to Oncorhynchus calls this usage Robin S. Waples is with the Coastal Zone and ferred to species of the genus Oncorhynchus, five into question. In this document, "Pacific salmon" Estuarine Studies Division, Northwest Fisheries of which (0. gorbuscha, O. keta, O. kisutch, O. is used to include anadromous forms of O. clarki Science Center, National Marine Fisheries Service, nerka, and 0. tshawytscha) occurin North America. and O. mykiss, as well as the five above mentioned NOAA, 2725 Montlake Boulevard East, Seattle, The recent decision to move the western trouts from species. WA981l2.

ABSTRACT-Forpurposes ofthe Endan­ differences betweenpopulations, andevalua­ 2) Does the population occupy unusual or gered Species Act (ESA), a "species" is de­ tions ofthe efficacyofnaturalbarriers. Each distinctive habitat? fined to include "any distinct population ofthese methodshas its limitations. Identifica­ 3) Does the population show evidence of segment ofany species ofvertebrate fish or tion ofphysical barriers to genetic exchange unusual or distinctive adaptation to its envi­ wildlife which interbreeds when mature. " can help define the geographic extent of ronment? Federal agencies charged with carrying out distinctpopulations, butreliance onphysical Several types ofinformation are useful in the provisions ofthe ESA have struggledfor features alone can be misleading in the ab­ addressing these questions. Again, the over a decade to develop a consistent ap­ sence ofsupporting biological information. strengths and limitations ofeach should be proach for interpreting the term "distinct Physical tags provide information about the kept in mindin makingthe evaluation. Pheno­ population segment. " This paper outlines movements of individual fish but not the typic/life-history traits such as size,fecundity, such an approach andexplains in some detail genetic consequences ofmigration. Further­ andageandtime ofspawning mayreflect local how it can be applied to ESA evaluations of more, measurements ofcurrent straying or adaptations ofevolutionary importance, but anadromous Pacific salmonids. recolonization rates provide no direct infor­ interpretation ofthese traits is complicatedby The following definition is proposed: A mationaboutthe magnitude orconsistencyof their sensitivity to environmental conditions. population (or group ofpopulations) will be such rates in thepast. In this respect, datafrom Data from protein electrophoresis or DNA considered "distinct" (and hence a "spe­ protein electrophoresis orDNA analyses can analyses provide valuable insight into thepro­ cies ")forpurposesofthe ESA ifit represents be very useful because they reflect levels of cess ofgenetic differentiation amongpopula­ anevolutionarilysignificantunit (ESU) ofthe geneflow that have occurredoverevolution­ tions but little direct information regarding the biologicalspecies. A population mustsatisfy ary time scales. The beststrategy is to use all extentofadaptivegeneticdifferences. Habitat two criteria to be considered an ESU: available lines of evidence for or against differences suggest the possibility for local 1) It must be substantially reproductively reproductive isolation, recognizing the limita­ adaptations butdo notprove that such adap­ isolated from other conspecific population tions ofeach andtaking advantage ofthe often tations exist. units, and complementary nature ofthe different types Theframework suggested here provides a 2) It mustrepresentan importantcomponent ofinformation. focalpointfor accomplishing the majorgoal in the evolutionary legacy ofthe species. Ifavailable evidence indicates significant ofthe Act-to conserve the genetic diversity Isolation does not have to be absolute, but reproductive isolation, the nextstep is todeter­ ofspeciesandthe ecosystems theyinhabit. At it must be strong enough to permit evolu­ mine whetherthe population in question is of the same time, itallows discretion in the listing tionarily important differences to accrue in substantial ecological/genetic importance to ofpopulations by requiring that they represent different population units. The second cri­ the species as a whole. In other words, ifthe units ofreal evolutionary significance to the terion would be met if the population con­ population became extinct, would this event species. Further, this framework provides a tributes substantiallyto the ecological/genetic represent a significant loss to the ecological/ means ofaddressing several issues ofpar­ diversity ofthe species as a whole. geneticdiversity ofthe species? In making this ticularconcemfor Pacific salmon, including Insights intothe extentofreproductive isola­ determination, the following questions are anadromouslnonanadromous population tion can beprovidedbymovements oftagged relevant: segments, differences in run-timing, groups fish, natural recolonization rates observedin 1) Is thepopulationgeneticallydistinctfrom ofpopulations, introducedpopulations, and otherpopulations, measurements ofgenetic other conspecific populations? the role ofhatcheryfish.

53(3),1991 11 consideration of the possible role of ar­ plants, and any distinct population seg­ lations under the ESA. The approach tificial propagation in recovery plans for ment of any species of vertebrate fish adopted here is an attempt to balance listed "species. " or wildlife which interbreeds when these themes in a framework consistent On receipt of petitions (April-June mature." Unfortunately, although the with both the letter and intent ofthe ESA. 1990) to list several populationsofPacif­ Act thus allows listing ofpopulations that Definition: A vertebrate population ic salmon as threatened or endangered are' 'distinct," it does not explain how will be considered distinct (and hence a "species" under the Act, NMFS sought population distinctness shall be evaluated "species") for purposes ofconservation public comments on draft Policy Position or measured. under the Act ifthe population represents Paper #1, "Definition of Species" (a Nevertheless, there is some guidance an evolutionarily significant unit3 (ESU) summary ofa longerdocument by Utter relevant to this issue. A review oflegis­ of the biological species. An ESU is a (1981». Based on that paper and public lative history indicates that an important population (orgroup ofpopulations) that: comments on it, discussions ofthe issue motivating factor behind the Act was the by the ESA Technical Committee, and desire to preserve genetic variability, 1) Is substantially reproductively iso­ ideas discussed at a Vertebrate Popula­ both within and between species. For lated from other conspecific population tion Workshopconvened inJune 1990by example, the House of Representatives units, and FWS and NMFS, Waples (1991a) pre­ described the rationale for House Resolu­ 2) Represents an important component pared a NOAA Technical Memorandum tion 37, a forerunner to the Act, in the in the evolutionary legacy ofthe species. that formed the basis for NMFS' "In­ following terms (H.R. Rep. 412, 93d terim Policy on Applying the Definition Cong., 1st Sess., 1973): Note that the dual criteria reflect two ofSpecies Underthe Endangered Species common meanings of "distinct": The Act to Pacific Salmon" (56 FR 10542; "Fromthe most narrow possible point of criterion of reproductive isolation em­ March 13, 1991). Afterconsideringpub­ view, it is in the best interests ofmankind phasizes the concept of "separate" or lic comments on the Interim Policy and to minimize the losses of genetic varia­ "apart from," whereas the criterion of the supporting NOAA Technical Memo­ tions. The reason is simple: They are evolutionary importance focuses on randum, NMFS published a Final Policy potential resources. They are keys to characteristics that are "different" or on this issue (56FR 58612; Nov. 20, puzzles which we cannot yet solve, and "unique. " 1991). Those comments were also con­ may provide answers to questions which Isolation does not have to be absolute, sidered in preparation of this paper. we have not yet learned to ask. " but it must be strong enough to allow evolutionarily important differences to Background of the On the other hand, in 1979 the General accrue in different population units. Endangered Species Act Accounting Office (GAO? recom­ Population characteristics that are impor­ The stated purposes ofthe Endangered mended that the authority to list verte­ tant in an evolutionary sense must have Species Act of 1973, as amended (16 brate populations under the ESA be a genetic basis; therefore, the second U.S.C. 1531 et seq) are to "provide a removed. Although this recommenda­ criterion would be satisfied ifthepopula­ means whereby the ecosystems upon tion was not adopted, the Senate Report tion in question contributes substantial­ which endangered species and threatened to the 1979 amendments stated that' 'the ly to the overall genetic diversity of the species depend may be conserved , [and] committee is aware ofthe great potential species. Because ecological diversity to provide a program for the conservation for abuse ofthis authority and expects the may foster local adaptations (and hence of such endangered species and threat­ FWS to use the ability to list populations genetic diversity), a population that oc­ ened species." "Species" is used in a sparingly and only when biological evi­ cupies unusual or distinctive habitat or in more expansive way in the ESA than dence indicates that such action is war­ other ways represents an important biologists or taxonomists generally use ranted" (Sen. Rep. 151, 96thCong., 1st ecological adaptation for the species may the term. In the original (1973) version Sess., 1979). Finally, the ESA (Sec. also be an ESU. ofthe Act, a "species" was defined to in­ 4(b)(1)(A» specifies thatlisting decisions The term "evolutionary legacy" is clude "anysubspecies offish orwildlife shouldbe based"solely on the basis ofthe used in the sense of "inheritance," i.e., or plants and any other group of fish or best scientific and commercial data something received from thepastand car­ wildlife of the same species or smaller available. " ried forward into the future. This reflects taxa in common spatial arrangement that Although not entirely self-contradic­ interbreed when mature." Use of this tory, the charge to conserve irreplaceable language established thatthe scope ofthe genetic resources but to do so sparingly 3 The term "evolutionarily significant unit" can be traced to Ryder (1986), who reported thatthe term Act extends beyond the traditional bio­ (and scientifically) clearly establishes a was used at a 1985 meeting of zoo biologists and logical definition of species to include certain tension in the process of "spe­ systematists in Philadelphia. The usefulness ofthis smaller biological units. Amendments in cies" determination for vertebrate popu­ concept in the context of ESA evaluations was stressed at the Vertebrate Population Biology 1978 (Public Law 95-632 (1978), 92 Stat. Workshop by A. Dizon (NMFS, Southwest Fish­ 3751) provided the current language in eries Science Center, La Jolla, Calif.). See Dizon 2GAO. 1979. Endangered species-a controversial etal. (In press) for additional discussion ofthis and the Act: A "species" is defined to include issue needing resolution. Rep. to Congress, Gov. related ideas with an orientation toward marine "any subspecies of fish or wildlife or Account. Off., Wash., D.C. mammals.

12 Marine Fisheries Review the concern expressed in the Act (Sec. lation4 is seldom a black-and-white stantial differences among populations 2(a)(5» for' 'better safeguarding, for the situation; rather, it is a question ofdegree can be maintained at strongly selected benefit of all citizens, the nation's heri­ (e.g., Ricker, 1972). Although the hom­ loci, while frequencies at neutral loci re­ tage in fish, wildlife, and plants." Spe­ ing instinct is well documented in these main relatively uniform (Slatkin, 1987). cifically, the evolutionary legacy of a species, natural straying does occur Another way to consider the migration species is the genetic variability that is a (Quinn, 1984), and anadromous5 spawn­ problem is to view it as a question of product ofpast evolutionary events and ing populations that are completely replacement. In this context, the relevant which represents the reservoir upon isolated from other conspecific popula­ question is this: Ifall individuals in the which future evolutionary potential tions are probably rare. A relevant ques­ population in question were permanent­ depends. Conservation of these genetic tion thus becomes: How much exchange 1y removed, would the area naturally be resources helps to ensure that the with other populations can a salmon repopulated by individuals of the same dynamic process ofevolution will not be population experience and still be con­ biological species, and ifso, within what unduly constrained in the future. Al­ sidered an ESU? Similarly, it is impor­ time frame? Presumably, an area that though the Act (Sec. 2(a)(3)) also notes tant to consider whether isolation is a would be naturally repopulated at or near that species' 'are ofesthetic, ecological, recent phenomenon or whether it repre­ the previous abundance level in a short recreational, and scientific value to the sents a long-standing condition. time would be unlikely to harbor an ESU. nation and its people, ,, focussing on these One approach to this question empha­ This will bea largely theoreticalexercise attributes without regard to the underly­ sizes the effects ofmigration in inhibiting for populations being considered for pro­ ing genetic basis for diversity is not a the processofgeneticdifferentiation. The tection under the Act. However, infor­ sound strategy for long-term species term" gene flow" is commonly used to mation for other populations and/or survival. Furthermore, societal values describe the movementofgenes from one species may provide some insight into change rapidly on an evolutionary time population to another; that is, gene flow this process. scale, and "species" with no apparent represents genetically effective migra­ The level of migration or gene flow significance today may be found to be tion. An oft-cited maxim based on the occurring among populations can be "valuable" at some point in the future. work of Sewall Wright (1978) is that evaluated in several ways. Approaches It is better, then, to focus on conserving gene flow between populations at the rate that may prove useful for Pacific salmon important genetic resources; ifthis is ac­ ofone individual pergeneration is suffi­ include: complished, then the other benefits of cientto preventthe tendency for different biodiversity (including various societal alleles to be fixed by chance (genetic drift) I) Use oftags to estimate strayingrates; interests) follow naturally. in different populations. Migration (the 2) Intentional genetic marking ofpop­ The framework adopted here provides physical movement ofindividuals) may ulations; a focal point for accomplishing the major occur at a higher rate than gene flow if 3) Use of genetic indices (e.g., goal ofthe Act-toconserve the genetic some migrating individuals have reduced Wright's (1978) FST or Slatkin's (1985) diversity of species and the ecosystems reproductive success or fail to reproduce private allele method) to estimate levels they inhabit. At the same time, it allows entirely. of gene flow; discretion in the listing ofpopulations by Although gene flow at the level ofone 4) Observations of recolonization requiring that they represent units of individual per generation may prevent rates; and evolutionary significance to the biolog­ extreme genetic divergence, it is not suf­ 5) Identification of physical or geo­ ical species. In this framework, repro­ ficient to equalize allele frequencies graphic features likely to act as barriers ductive isolation is a necessary but not a across populations, and Wright (1978) to migration. sufficientcondition for a population to be also pointed out that genetic differentia­ considered "distinct." Given enough tion is by no means negligibleeven ifgene Since the 1950's, extensivedata bases time, an isolate may evolve into an ESU , flow occursatseveral times this rate. Fur­ have been developed that provide impor­ but isolation by itself does not confer thermore, the above comments apply to tant information about the oceanic distri­ distinctness. a balance between migration and genetic bution ofPacific salmon (Hartt, 1962). 6 driftofneutral alleles. Selection for local­ More recently, physical tags have been Application to Pacific Salmon ly adapted alleles can offsetthe homogen­ used to study migrations offish from in­ Application of the ESU concept to izing effects of gene flow. Thus, sub­ dividual populations (Johnson, 1990). Pacific salmon requires consideration of However, tagging efforts typically focus reproductive isolation and ecological/ on hatchery stocks, and much less is 4 This discussion ofreproductive isolation focuses genetic diversity. Forconvenience these on the degree to which a population is isolated from known about natural straying rates. Fur­ concepts are considered separately here, genetic contact with other natural populations. thermore, tagging studies do not provide but it is recognized that they are inherent­ Straying from hatchery stocks and intentional transfers of fish from one area to another are dis­ direct evidence of gene flow. A salmon 1y related. cussed in the section on "Special Considerations: Hatchery Fish." Reproductive Isolation 5 Nonanadromous populations orsegments ofpop­ 6 See also subsequent International North Pacific ulations are considered in the section on "Special Fisheries Commission Bulletins on salmon With Pacific salmon, reproductive iso­ Considerations: Anadromy/Nonanadromy. " distribution.

53(3),1991 13 may swim into a nearby stream (and lower frequency may also be indicative eliminated, would become reestablished perhaps be counted as a stray) before of reproductive isolation. naturally. Approaches 1, 2, and 4 can ultimately making its way to its natal Opportunities to observe recoloniza­ provide data on current levels of migra­ stream to spawn. Reproductive success tion are not common with Pacific salmon, tion or gene flow (over periods ofone or ofstrays may also be less than that offish but they do occur. Aspinwall (1974) a few generations). It is unlikely, how­ from the local population. described one experiment thateradicated ever, that migration rates have been con­ A direct measurementofgene flow can an entire run of pink salmon to study stant over long periods oftime. In some be obtained by monitoring changes over straying. Natural events (e.g., theerup­ cases, significant gene flow may occur time in the frequency ofgenetic markers tion of Mount St. Helens) sporadically only at intervals ofdecades orcenturies. characteristic of different populations. occur that cause extinction ofpopulations Genetic methods can be very informative Because Pacific salmon populations are or allow access to previously blocked in this context because they reflect the typically characterized by different fre­ habitat. In evaluating the results ofsuch cumulative effects of gene flow over quencies of the same suite of alleles, "experiments, " it should be recognized evolutionary time scales. rather than by qualitative differences in that ifintraspecific interactions (such as the types of alleles present (Utter et al., competition) are acting to hinder the suc­ Ecological!Genetic Diversity 1980, and many more recent references), cess ofmigrants orstrays, recolonization Ifavailable evidence indicates signifi­ it often will be difficult to measure gene of empty habitat may occur at a higher cant reproductive isolation, the next step flow precisely withoutenhancingthefre­ rate than expected from migration rates is to determine whether the population in quencies ofdifferent alleles in different among fully-seeded populations. question is of substantial ecological! populations (intentional genetic mark­ Because natural straying in Pacific genetic importance to the species as a ing). Although the few genetic marking salmon seems to be largely confined to whole. In other words, ifthe population studies that have been conducted with nearby areas (Quinn and Fresh, 1984), became extinct, would this event repre­ Pacific salmon (Seeb et aI., 1986; Lane geographic proximity of a population sent a significant loss to the ecological/ et al., 1990) have provided important in­ from other conspecific populations may genetic diversity ofthe species? In eval­ formation, opportunities for such studies provide a useful approximation of the uating a population's contribution to involving populations that are potential degree of reproductive isolation. A ecological/genetic diversity, the follow­ candidates for ESA listing are likely to be number ofstudies ofPacific salmon have ing questions are relevant: limited. found genetic clustering of populations An indirect measurement ofgene flow to occur largely along geographic lines 1) Is the population genetically distinct is provided by Wright's and Slatkin's (see references in Waples, 1991b). How­ from other conspecific populations? methods, which measure some of the ever, both types of exceptions to this 2) Does the population occupy unusual genetic consequences ofmigration. Ac­ pattern-pronounced differences be­ or distinctive habitat? curacy ofsuch estimates depends on the tween nearby populations and lack ofdif­ 3) Does the population show evidence degree to which the various assumptions ferences between distant populations­ ofunusual ordistinctive adaptation to its ofthe models used are satisfied. Notably, have also been found, which emphasize environment? both methods assume selective neutral­ the fact that distance is not the only bar­ ity ofthe alleles used, and results may be rier to gene flow. Consideration should Important factors to consider in addres­ sensitive to the geographic configuration also be given to other factors (e.g., geo­ sing these questions include (but are not ofthe study sites included. Both methods logical history and physical and en­ necessarily limited to) the following: are also based on equilibrium models and vironmental gradients) that can affect may overestimate migration rates ifisola­ population structuring. It must also be 1) Genetic traits. Examples include tion has been too recent for a balance to recognized that what appears to be a bar­ presumably neutral characters detected develop between the forces ofmigration rier to a biologist may not be a barrier to by protein electrophoresis or DNA anal­ and genetic drift. fish, and vice versa. Inferring barriers to yses as well as other genetically-based The presence of unique alleles (those migration on the basis ofgeographical or traits that are more difficult to quantify. found in only one population or one geo­ physical features alone can be misleading 2) Phenotypic traits. Examples include graphic region) may also provide insight in the absence of supporting biological morphological and meristic characters, into the degree ofreproductive isolation. information. occurrenceofparasites, and disease and A major concern in evaluating such data As can be seen from this brief discus­ parasite resistance. is sampling error; that is, the failure to sion, each of the above approaches has 3) Life-history traits. Examples in­ find the alleles in other localities may be limitations. Nevertheless, it is important clude time, size, and age at spawning; due to inadequate sampling. Neverthe­ to consider all available information spawning behavior; fecundity; migration less, alleles that have been found in only because the various approaches provide patterns; and timing of emergence and one area and occur there at moderate or different insights into the question of outmigration. high frequency suggest a substantial reproductive isolation. Recolonization 4) Habitat characteristics. This cate­ degree ofreproductive isolation. The oc­ rates provide the most direct indication gory includes such physical character­ currenceofa number ofunique alleles at of the likelihood that a population, if istics ofthe spawning and rearing habitat

14 Marine Fisheries Review as temperature, rainfall, stream flow, and tion of likely adaptive differences else­ the Act. water chemistry, as well as biological at­ where in the genome. 1) Evaluate the degree ofreproductive tributes ofthe local ecosystem. Location Several types of genetic analyses can isolation. Ifthere is gene flow with other within a river drainage (e.g., upstream provide information relevantto ESA con­ populations, it should be at a level low vs. downstream) and elevation can also siderations. Gene diversity analysis (Nei, enough to permit evolutionarily impor­ be important in this regard. In a broader 1973) can be used to partition the total tant divergence. If apparent migration sense, the habitat for a population also genetic variance in a species in a variety rates with adjacent populations are high, includes areas encountered during the of ways, including between-population, the population would not be considered entire life cycle. Thus, a waterfall in the within-population, and between-years isolated unless there is evidence that the migratory route might select for robust (within population) components. Results genetically effective migration rate is fish capable of surmounting it, and a can be compared to data for other species much lower. Approaches outlined in the population with distinctive oceanic (and other salmonids in particular) to pro­ previous sectionon "ReproductiveIsola­ migration patterns probably utilizes vide insight into the degree of genetic tion" should be used to address the ques­ marine habitat differently than do other distinctness ofthe population under con­ tions of migration rate, gene flow, and populations. sideration. Genetic distance indices can recolonization rate. be used in a similar fashion, and com­ If the population is believed to be Two points are important to consider parisons of heterozygosity levels may reproductively isolated, an evaluation in evaluating these types of data. First, also be informative in some cases. under step 2 (below) should be made; if such data can be properlyevaluated only Phenotypic and life history traits may it is not isolated, the population is not an in relation to similar information for the reflect local adaptations, and for this ESU and should not be considered a species as a whole. That is, some refer­ reason they may be relevant to the evalua­ separate "species" under the Act (pre­ ence data are necessary before one can tion ofpopulationdistinctness. However, sumably, however, it would be part ofa determine that a particular population is expression of these traits is known to be larger unit that is an ESU). distinct. affected by environmental as well as 2) Evaluate evidence for ecological/ Second, it is unlikely that complete genetic factors (Barlow, 1961; Clayton, genetic distinctness in the context of biological information will be available 1981), which complicates their inter­ similar data from throughout the species for any given ESA evaluation. This is par­ pretation. Sorting outthe genetic and en­ range, as well as for other species as ap­ ticularly true for status reviews that must vironmental effects on phenotypic and propriate. Often, this process will involve be conducted within a limited time period life history characteristics is a challeng­ difficult judgments concerning the in response to formal petitions for listing. ing task and has been a central focus in relative importance to attach to different ESA determinations must be made on the evolutionary biology (Endler, 1986). types ofevidence. Although a variety of basis of the best scientific information Analysis of habitat characteristics is approaches may prove useful in making available at the time, and all relevant data important in two ways. First, the exis­ this determination, none will provide a should be considered. tence of unusual or distinctive habitat completely objective assessment of Data from protein electrophoresis or features allows for the possibility of evolutionary significance. Nevertheless, DNA analyses permit direct inferences unique adaptations in the local popula­ some general guidelines can be suggested about genetic divergence and thus are tion. Second, identification ofunusual or to aid the evaluation process. particularly applicable to the question of distinctive habitat is one step toward The existence of substantial genetic population distinctness under the Act. achieving the broad purpose ofthe Act­ differences from other conspecific pop­ However, ifthe common presumption is to preserve threatened and endangered ulations based on protein electrophoresis correct that the genetic characters "species" and the ecosystems they in­ or DNA analyses would strongly suggest detected by these methods are largely habit. Again, however, caution should be that evolutionarily important, adaptive neutral with respect to natural selection, used in drawing inferences based on differences also exist? The failure to then it follows that differences among physical characteristics of the habitat find such differences (or the absence of populations in these characters do not by without supporting biological informa­ genetic data) would not rule out the pos­ themselves denote evolutionary signifi­ tion linking the habitat differences to sibility that such adaptive differences cance, except in the sense that neutral adaptations. Just as our perception of exist, but it would place a greater burden genes may provide the raw material for what constitutes a barrier to fish migra­ ofproofon data for other characters. Data future evolution. Rather, these genetic tion can be faulty, so too our understand­ for habitat characteristics should be inter­ characters are primarily useful as indi­ ing of the importance of various habitat preted in a similar fashion: habitat dif­ cators (or proxies) for evolutionary pro­ characteristics to organisms is far from cesses that can lead to local adaptation in complete. other parts of the genome more directly 7 Although protein electrophoresis and DNA tech­ Recommended Approach niques are considered together here for conve­ related to fitness. In a similar way, al­ nience, a variety of parts of the nuclear or mito­ though unique alleles do not necessarily The following two-step approach is chondrial genome can be targeted for genetic study. Sensitivity ofthe different approaches can vary con­ reflect adaptation, they may, ifnumerous suggested for making a determination siderably. and this factor must be considered in or at high frequency, provide an indica­ regarding population distinctness under interpreting the results of genetic analysis.

53(3), 1991 15 ferences suggest (but do not prove) the sidered as a unit for purposes ofthe Act. tribution of run-times or distinct times possibility of adaptive differences, In determining whether such a population and/or locations ofspawning). whereas the inability to detect habitat unit is an ESU, the anadromous and Assuming that fish with different run­ differences constrains the scope of pos­ nonanadromous traits should be consid­ times are reproductively isolated, they sible local adaptations but does not prove ered in the same manner as other popula­ can be considered distinct populations they do not exist. In evaluating data for tion characteristics discussed in the sec­ under the Act ifthey exhibit evolutionar­ phenotypic and life-history traits, every tion on "Ecological/Genetic Diversity. " ily important ecological/genetic differ­ effort should be made to account for en­ The important questions are whether the ences, as outlined in the section on vironmental effects that are manifested in traits have a genetic basis and whether "Ecological/Genetic Diversity. " In the periods shorter than one generation (and they help to make the population unit absence ofsubstantial isolation between therefore do not reflect adaptations). "distinct" from other populations. For run-times (or if the races are reproduc­ example, an anadromous/nonanadro­ tively isolated but do not individually Special Considerations mous unit might be considered an ESU if satisfy the ecological/genetic diversity other ecologically comparable popula­ criterion), a "population" unitconsisting Anadromy/Nonanadromy tions of the species harbored only the of two or more recognized run-times Some species ofPacific salmon, Onco­ nonanadromous form. In this case, ifthe could be considered an ESU if it were rhynchus nerkn, 0. mykiss, O. clarki, and population unit is considered to be an isolated from and distinct from other perhaps others, have nonanadromous as ESU solely or primarily on the basis of populations. well as anadromous forms that occur the anadromous trait, then the potential together. A similar phenomenon occurs loss ofanadromy should be a legitimate Hatchery Fish in some species ofthe genera Salmo and ESA concern. A key question would be Artificial propagation has been used in Salvelinus. This raises the question whether the nonanadromous form was one form or another with anadromous whether the two forms should be con­ likely to give rise to the anadromous form Pacific salmon for over a century. Hatch­ sidered jointly or separately in deciding after the latter had gone locally extinct. eries have been used both for fisheries if a population is "distinct" under the Therefore, an anadromous/nonanad­ enhancement (largely as mitigation for Act. The following general guidelines are romous population unit could be listed losses ofnative stocks caused by destruc­ suggested. based on a threat to one ofthe life-history tion of habitat or blockage of migratory The two forms should be considered traits, ifthe trait were genetically based routes) and in an attempt to boost produc­ separately ifthey are reproductively iso­ and loss of the trait would compromise tion of naturally-spawning fish (Licha­ lated. As noted above, the question of the"distinctness" ofthe population. towich and McIntyre, 1987). Currently, reproductive isolation is likely to be one hatchery operations in the Pacific North­ of degree. Again, the key question is Differences in Run-time west are carried out on a large scale, and whether isolation is strong enough for In several species of Pacific salmon, the majority of adult fish produced in evolutionarily important differences to biologists recognize different run-times, many systems are of hatchery origin develop in the two forms. Data from or races, offish inhabiting the same gen­ (Washington, 1985; Vreeland, 1986). It protein electrophoresis or DNA analyses eral area. Generally, run-times are deter­ is important, therefore, to consider the can be valuable in making this determina­ mined on the basis ofthe time of year at role that hatchery fish play in ESA con­ tion, as can observations oftimeand place which adults enter fresh water to spawn; siderations for Pacific salmon. ofspawning and behavioral interactions in some cases, fish with different run­ The Act (Sec. 3(3)) identifies' 'prop­ during spawning. Information from other timing also have different juvenile life agation" as one method that may be used populations (and other species) can help history patterns (Healey, 1983; Groot to conserve threatened or endangered provide a context for making the evalua­ and Margolis, 1991). The question wheth­ species, and both NMFS and FWS have tion. However, such studies suggest that er such races represent' 'distinct' , popu­ used captive breeding or other artificial a variety ofscenarios probably exists in lations under the Act can be addressed in propagation techniques with listed "spe­ nature, from substantial reproductive the framework developed above. cies," including several fishes. Artificial isolation of sympatric anadromous and First, it should be determined whether propagation may thus be an appropriate nonanadromous forms to substantiallife­ the different run-times are reproductively tool for use in recovery plans for some history plasticity within presumably a isolated. Often, the formal distinction "species" ofPacific salmon. However, single gene pool (Footeetal., 1989; Kir­ between run-times is rather arbitrary, the Act also mandates conservation of pichnikov et al., 1990). Therefore, such with (for example) fish appearing before native ecosystems' 'upon which endan­ studies are unlikely to provide an un­ a certain date classified as "springs" and gered species and threatened species equivocal answer for an unstudied anad­ those appearing after that date as "sum­ depend. " The key here is the link be­ romous/nonanadromous system. mers. ' , Races that are arbitrarily defined tween threatened and endangered species Ifsubstantial gene flow occurs or has in this fashion may in fact be reproduc­ and their native ecosystems; the link may recent!y occurred between the two forms, tively isolated, but if so this needs to be be jeopardized if either component (the they represent polymorphisms within a demonstrated biologically (for example, species or the ecosystem) is emphasized single population and should be con­ by providing evidence for a discrete dis­ to the exclusion of the other. Because a

16 Marine Fisheries Review fish hatchery is not a substitute for a should generally be presumed that hatch­ be consistent with the twofold criteria natural ecosystem, maintaining a "spe­ ery fish are not partofan ESUunless there that define an ESU. In making this deter­ cies" in a hatchery while allowing degra­ is a compelling reason for including them. mination, it may be useful to consider dation of its native habitat is not consis­ whether the population was likely to have tent with the stated purposes ofthe Act. Effects of Artificial Propagation been an ESU in the past and to ask For this reason, attention in ESA evalua­ and Other Human Activities whether stock mixing (orotheraspects of tions of Pacific salmon should focus on Two possible effects ofartificial prop­ artificial propagation) has compromised fish that spend their entire life cycle in agation are ofparticularconcern in ESA the evolutionarily importantadaptations their native habitat-i.e., those that are evaluations: 1) Genetic changes within a that distinguished the original popula­ progeny of naturally spawning parents. population and 2) mixtures ofgenetical­ tion. Several factors should beconsidered Such fish will be referred to as' 'natural" ly distinct populations. These effects in this context. (It is assumed that the in the ESA contexts. Threshold deter­ shouldbeevaluated from the perspectives population in question was distinct minations also will focus on natural fish, of both the hatchery population(s) in­ enough to be an ESU prior to the effects on the premise that an ESU is not healthy volved and any natural population that of human activities.) unless a viable population exists in the may have been affected. That is, an First, although stock transfers ofanad­ natural habitat. evaluation of the nature and extent of romous Pacific salmon have been wide­ Once the natural component ofa pop­ these effects will help to determine 1) spread in the past (Withler, 1982), evi­ ulation has been identified, the next step whether a natural population is an ESU dence merely ofthe release ofexogenous is to determine whether this population and 2) whether a hatchery population(s) fish is not sufficient to disqualify a pop­ componentis "distinct" for purposes of should be included in an ESU defined on ulation from consideration as an ESU. the Act. In making this determination, the the basis ofa distinct natural population. Stock transfers (or straying) have a direct twofold criteria for defining an ESU Supplementation (the release of genetic effect only ifthe transplanted fish should be used. Ifit is thought that arti­ hatchery-reared fish into habitatoccupied successfully reproduce and contribute to ficial propagation may have affected by fish of the same biological species), subsequent generations of the native the genetic composition of the natural transfer ofeggsorfish among hatcheries, stock. Results ofsupplementationefforts population (either directly through sup­ and unintentional straying are all aspects with Pacific salmon have been quite plementation or indirectly through stray­ of artificial propagation that can lead to variable (Ricker, 1972) and atpresentare ing ofhatchery fish), factors outlined in population mixing. Such mixing is rele­ largely unpredictable. For example, a the following section should be con­ vant to the question of population dis­ recent review ofsalmonid studies (Hin­ sidered. Fishare notexcluded from ESA tinctness because a population that has dar et al., 1991) cited examples in which consideration simply because some of been overplanted with fish of different the native stock had been largely or en­ their directancestors may have spenttime origin may not be an ESU even ifit once tirely displaced, examples ofhybridiza­ in a fish hatchery. However, there are was distinct. Similarly, a hatchery stock tion between native and introduced fish, a number of potential genetic conse­ that has resulted from extensive transfers and examples in which repeated hatchery quences of artificial propagation that ofexogenous fish is unlikely to be part of releases had no detectable genetic effect should be considered in this context. an ESU. on the native population. Therefore, it Thus, fish meeting the definition of Apart from the effects of stock mix­ should not automatically be assumed that "natural" adopted here may subsequent­ tures, artificial propagation can also lead transplantation efforts have permanent­ ly be excluded from ESA consideration to either random or directional genetic ly altered the genetic structure ofnative for other reasons. change within cultured populations. Ran­ populations; similarly, some stock trans­ In developing recovery plans for dom genetic changes, if of sufficient fers among hatcheries may not have "species" listed as threatened orendan­ magnitude, may lead to the erosion of permanent genetic effects. gered, the use of artificial propagation genetic variability and may overwhelm Second, a population that does not may beconsidered. Ifa hatchery is asso­ selection for locally adapted genotypes, represent a completely pure native gene ciated with the listed' 'species," an im­ thus reducing fitness. Directional genetic pool may still qualify as an ESU if it is portantquestion to address in formulating change can occurat the time offounding adapted to its local environment and is a recovery plan is whether the hatchery a hatchery population (or between "distinct" by the criteria outlined in the population is similarenough to the natural generations in a hatchery) through choice section "Application to Pacific Salm­ population that it can be considered part of which individuals will be allowed on. " This point is important for Pacific of the ESU defined on the basis of that to reproduce. In addition, fish hatch­ salmon, because there are relatively few natural population. Factors to consider eries differ from the natural environ­ populations for which the possibility of in this regard are discussed in the next ment in a number ofways that may affect some genetic influence from transplants section. Given various uncertainties, it selective pressures experienced by the or hatchery strays can be completely ex­ population. cluded. How much introgression from Determinations regarding the distinct­ other gene pools must occur before a 8 This corresponds to the usage suggested by Bjornn and Steward ( 1990); some other authors have used ness of populations that may have been population no longer merits considera­ the term in a different way. affected by artificial propagation should tion as an ESU depends to some extenton

53(3),1991 17 the degree ofdistinctness ofthe original number ofstudies ofculturedpopulations denceofpredators)-differso markedly population. oftroutand Atlantic salmonthatprovide from the natural environmentthat selec­ In evaluating the effects ofpopulation evidence for severe inbreeding depres­ tive changes are likely. mixture, the following types ofinforma­ sion and/or substantial loss of genetic Unfortunately, it is easier to identify tion should be gathered whenever variation. Such dramatic effects have not the potential genetic risks posed by ar­ possible: been documented with Pacific salmon, tificial propagation than to evaluate their butthere is indirectevidence (Waples and actual impact on a given population. As 1) Genetic, phenotypic, and lifehistory Teel, 1990) that effectivepopulation size a general principle, it is probably fair to traits and habitat characteristics for all in some hatcheries is small enough that say that genetic changes in a population stocks involved; such problems are a potential concern. that result from adaptation to hatchery 2) Broodstock and rearing protocols Forpopulations under ESAconsidera­ conditions are unlikely to increase the for all relevant stocks; tion, the importance ofrandom changes fitness of the population in the natural 3) Dates ofrelease, number released, attributable to a history ofartificial prop­ environment. How rapidly such effects and developmental stage at release for agation can be evaluated by examining occur, however, and whether they are transplanted stocks (or those involved in the number and sex of spawners each reversibleare open questions at present. straying); year, methods offertilization, and rear­ Some idea of the likely magnitude of 4) Trends in abundance of the local ing protocols. Exports ofeggs orprogeny selective changes due to artificial prop­ population for a time span bracketing the offsite should also be considered; if (as agationcanbegainedby consideringpast periods(s) ofrelease (or strayingevents); has often occurred in the past) the entire hatchery practices for the stock in ques­ and productionofcertainfamilies is shipped tion and the number of generations in 5) Evidence for reproductive success to another hatchery, thosefamilies do not culture. Ifpossible, baseline data from oftransplanted (or stray) fish, including contributeto theeffective sizeofthelocal the original (prehatchery influence) evidence for incorporation of foreign population (Simon etaI., 1986). In addi­ population should be compared to data genes into the local population. tion, the variability among individuals in from the current population. Focus reproductive success is a key factor in should be on 1) possible reductions in the Ideally, information will be available for determining effective population size, ability of the population to survive and the transferred stock as well as for the but this parameter is very difficult to reproduce in the natural environment, local stockboth before and after the mix­ measure for Pacific salmon. For this and 2) possible changes in characteristics ture event(s). This type of data would reason, a monitoring program that uses that help to make the population distinct. allow an evaluation of whether changes indirect genetic methods (Waples, Again, genetic changes within cultured in the local stock are in the direction 1990b) can provide useful insights into populations are important to consider predicted under the assumption that the the magnitude of random genetic from the perspectiveboth ofthe hatchery mixing has had a permanent effect. In changes. population (and its relation to an ESU) practice, such extensive data will not Directional genetic changes can occur andany natural populationsthatmay have always be available, and in this event the from a variety offactors. Thepracticeof been affected by the cultured stock. evaluation can be much more difficult. culling fish according to age, time of Other human activities (e.g., fishing, Nevertheless, thereare someapproaches return, size, or appearance was former­ habitat degradation) can also alter the (Waples and Smouse, 1990) that have ly widespread in Pacific salmon hatch­ genetic structure of native populations, reasonable power to detect population eries (Donaldson and Menasveta, 1961). and the importance of these factors to mixtures under certain conditions. More recent awareness ofthe drawbacks ESA considerations can be evaluated Finally, although genetic changes to this approach can help to minimize in a similar way. For example, fishing within cultured populations are a legiti­ such effects, but they cannot be elim­ pressurecanselectively affectcertain size mate ESA concern, the effects of such inated entirely. In addition, anadromous or agegroups (Ricker, 1981; Nelson and changes on the viability ofnatural popula­ fish hatcheries, ifsuccessful in theirgoal Soule, 1987), and these characteristics tions ofanadromous salmonids arelarge­ of ensuring that a large proportion of may be heritable. The relevant question Iy unknown. Randomchanges occurin all progeny survive to time of release, also is whether the activities have changed the populations, at a rate inversely propor­ dramatically alter the mortality pattern population so much that it no longer tional to the effective population size. for the population. In general, this canbe represents an evolutionarily significant Unless the number of spawners is se­ expectedtolead togeneticchangerelative component of the biological species. verely limited, random changes can to a population that spawns naturally generally be minimized in cultured (Waples, 1991b). Furthermore, a num­ Introduced Populations populations by following appropriate ber of characteristics of the hatchery In general, populations resulting from broodstock practices (Meffe, 1986; environment-both physical (e.g., the introduction of fish into a local area Simon et aI., 1986; Allendorf and substrate type, water temperature and not occupied by the biological species Ryman, 1987). This has not always been flow, and the variability of same) and (particularly if the area is outside the the case with fish hatcheries, however. biological (e.g., density, food type and historic range ofthe species) are probably Allendorfand Ryman (1987) reviewed a source, behavioral interactions, inci­ not ESU's because they do not contrib-

18 Marine Fisheries Review ute to maintaining diversity ofthe species should be considered in making this pression and allowing a smallereffective in its native habitats. Again, the key is the evaluation. Although there is no consen­ size than would ordinarily be the case. It link between a "species" and its native sus among geneticists regarding the is possible that, in some populations, a habitat, and this link is broken when fish minimum effective population size per similar process-evolution of demo­ are moved from one ecosystem to generation (Ne ) necessary to avoid long­ graphic parameters, for example-may another. Some introduced populations term problems ofinbreeding and loss of have occurred that modulates the effects should not be excluded from ESA con­ genetic variability, most estimates are in of environmental variability. Because sideration, and these include populations the range ofseveral hundred (see discus­ such populations would contain adapta­ occupying habitat that is ecologically sions by Lande and Barrowclough (1987) tions that might truly be considered to be similar and geographically proximate to and Simberloff (1988». For a species of evolutionary significance to the spe­ the source population, and those that with overlapping age classes and an cies, and because small populations in represent the only remaining component average age at spawning of 3-5 years general are alikely source ofevolutionary ofa native gene pool. In the former case, (typical ofmany populations ofsteelhead innovation, it is prudent to exercise cau­ the introduced population may be deter­ and chinook, chum, and sockeye tion in eliminating a population from ESA mined to be part of the same ESU as the salmon), this would correspond to an consideration simply on the basis of parent population; in the latter case, the effective number of breeders per year historic size. In particular, theoretical population could be determined to be an (Nb ) of perhaps 50-100. (Waples considerations about the likely persis­ ESU if it met the criteria outlined in the (1990a) provides a discussion ofthe rate tence time ofsmall populations should not section" Application to Pacific Salmon. " of loss of genetic variability in Pacific override strong evidence for long-term salmon and the relationship between N reproductive isolation. Nevertheless, this Historic Population Size e and Nb .) Because not all individuals suc­ concept should prove useful in focusing For a population that once was abun­ cessfully spawn, and because the vari­ attention on population units with the dant but since has declined in numbers, ance among individuals in reproductive greatest probability of representing there should be no minimum size for ESA success may be high, the total number of ESU's. consideration. However, populations adults must generally be somewhat more may also be small because of limiting than this (perhaps several times as many). Groups of Populations physical or biological factors. For Pacific Thelong-termpersistence ofan isolated As anadromous species, Pacific salm­ salmon, suitable habitat may severely population also depends on its ability to on spawn in a freshwater environment limit the potential number of spawners withstand inevitable (and often large) that is often naturally organized in a and hence the carrying capacity ofsmall fluctuations in abundance caused by the hierarchical fashion: Major river systems streams. Given the large temporal fluc­ interplay ofpopulation dynamics, chang­ may contain several large tributaries, tuations in abundance documented for ing environmental conditions, and each with numerous streams fed by every species (and many populations) of chance events. The importance of these smaller creeks, etc. Other areas may be Pacific salmon, and given the likelihood factors varies among species and among characterized by numerous smaller that even greater fluctuations have occur­ populations within species and must be streams, each entering directly into a red overevolutionary time, theremustbe evaluated on a case-by-case basis. In tidewater area. In both cases, geograph­ some size below which a spawning general, however, such fluctuations may ic, environmental, or other factors may population is unlikely to persist in isola­ place greater constraints on the long­ naturally lead to genetic structuring of tion for along period oftime. The fact that term survival of small populations than the various spawning aggregations into small spawning aggregations are regular­ do genetic factors associated with more or less discrete units. The first step ly observed may reflect a dynamic pro­ inbreeding. in determining the appropriate hierar­ cess ofextinction, straying, and recolo­ A Pacific salmon population should chicallevel for consideration as an ESU nization. Such small populations are not be considered an ESU ifthe historic is to identify units within which levels of unlikely to be ESU's, although a collec­ size (or historic carrying capacity) is too gene flow are high relative to the rate of tion of them might be. Therefore, the small for it to be plausible to assume the exchange between units. Often, how­ historic size9 of a population may be population has remained isolated over an ever, there will be more than one hier­ useful in evaluating whether it is an ESU. evolutionarily important time period. In archical level for which this is true. Both genetic and demographic factors making this evaluation, the possibility Therefore, it is important to identify should be considered that small popula­ reproductively isolated units that also tions observed at present are still in contribute substantially to ecological! "This concept is related to, but differs from, the con­ cept of minimum viable population size (MVP) existence precisely because they have genetic diversity ofthe species as a whole. (Shaffer, 1981; Soule, 1987). The MVP concept evolved mechanisms for persisting at low Determining the appropriate level for considers the future and asks how large a popula­ tion must be to have an acceptably high probability abundance. Population genetics theory consideration as an ESU is a challenging ofsurviving a specified period oftime. The historic indicates that gradual inbreeding over a task with Pacific salmon. Although the population size concept considers the past and asks period of time may purge deleterious, strong homing instinct ofthese species in­ how small a population must be before it becomes unreasonable to assume it has persisted in isolation recessive alleles from a population, dicates that even small spawning ag­ long enough for important adaptations to evolve. lessening the effects of inbreeding de­ gregations may potentially represent

53(3),1991 19 biological populations, such populations habitat can pose a threatto larger popula­ ticular, the combination ofoverlapping may not meet the criteria to be considered tion units. The underlying concern age classes with one-time reproduction) "distinct" under the Act. A group of should be whether important genetic provide some additional opportunities for populations, however, might be distinct resources ofthe biological species are at sampling bias. These factors should be from, and isolated from, other groups of risk because ofthe fragmentation. Ifso, kept in mind in designing sampling plans populations. Such a group ofpopulations then the appropriate action may be to pro­ and in evaluating results. can constitute an ESU and, ifdetermined tect the larger population as a whole, to be threatened or endangered, can be rather than the individual fragments. In Temporal changes withinpopulations. afforded protection under the Act. this context, NMFS recognizes that Although most Pacific salmon spend the In evaluating the appropriate grouping thresholds for threatened and endangered majority oftheir life at sea, they exist in level, a balance must be struck between status must be flexible enough to deal with recognizably discrete populations only two opposing concerns. On the one hand, threats to groups of populations (meta­ during rearing as juveniles and spawning it is important to identify the smallest populations) and clinal populations as as adults. By necessity, samples are units that meet the criteria set out in the well as more discrete population units. generally taken from local spawning section on "Application to Pacific Salm­ Just as there is no simple formula for populations, and often only a single brood on," because this allows the greatest determining evolutionary significance, year is sampled. It must be realized that flexibility in ensuring the appropriate there is no universally applicable numer­ the population as a whole includes several level of protection for different ESU's ical threshold for a listing determination; brood years, and values for a given within a more comprehensive group. On in both types of evaluation, a variety of character will show year-to-year varia­ the other hand, we have seen in the factors must be considered. Recovery tion around the mean for the population previous section, "Historic Population plans for listed" species" could take this as a whole. Theeffects oftemporal varia­ Size," that the smallest units supporting into account by ensuring protection for tion within a population must be con­ local populations of salmon may not be smaller units within a more comprehen­ sidered in comparing single-brood-year evolutionarily independent from other sive ESU. This might be appropriate, for samples from different populations; in nearby populations. A key question is: example, if the smaller units differ in general, this factor will inflate the observ­ How can evolutionarily important units various characteristics but it is uncertain ed level ofdifference above that predicted be protected without running the risk of how these differences relate to evolu­ by the null hypothesis, even ifthe overall attempting to artificially maintain units tionary significance. population means do not differ (see that might naturally undergo episodes discussion of a similar point in Waples of extinction/recolonization on some­ Interpreting Results and Teel (1990)). A study plan that in­ thing short ofevolutionary time scales? of Statistical Tests volves temporally spaced samples within The following approach is suggested. sites as well as samples from geograph­ Identifiable ESU's should not be com­ Sampling Considerations ically distinct localities is the best way to bined for the sake of convenience. In Rigorous analyses ofdata used in ESA evaluate the significance and stability of general, however, ESU's should corre­ considerations will include testing between-population differences. spond to more comprehensive units hypotheses whenever possible, and unless there is clear evidence that evolu­ sampling protocols are important to Life history stage sampled. Waples tionarily important differences exist consider in this context. In general, and Teel (1990) showed that, in compar­ between smaller population segments. regardless of the characters being con­ ing two (or more) samples, the probabil­ This approach is consistent with the sidered, the appropriate null hypothesis ity ofa statistically significant result may recommendation that NMFS and FWS to test is that no differences exist between depend on the life history stage sampled. should use sparingly their authority to list the populations being compared. Sam­ In general, sampling juveniles will tend vertebrate populations, and only ifbio­ pling from the populations introduces a to produce larger differences (and a logical evidence clearly warrants it. In source of random error with magnitude higher probability of a significant test addition, this approach reflects 1) the inversely proportional to sample size. In result) than sampling adults. This will be view that population "distinctness" many statistical tests, the implicit a minor effect if the sample size is small should be supported by positive scientific assumption behind the null hypothesis is relative to the population size, but this evidence and 2) the concern that frag­ that the samples being compared were will not always be true for populations menting groups ofpopulations into mul­ randomly drawn from the same popula­ under ESA consideration. tiple ESU's on the basis of insufficient tion. There are several ways in which this data may create artificial units without a basic assumption might be violated by the Nonrepresentative sampling. Most biological basis. method of sampling, and the effects of statistical tests assume random sampling, Nevertheless, it is recognized that the violating the assumption are often mag­ which means that every individual in the long-term viability of a larger unit may nified in small populations (as may fre­ population(s) has an equal opportunity to also depend on the continued existence of quently be encountered in ESA evalua­ appear in the sample. There are several multiple, semi-independent units it com­ tions). Furthermore, the unusual life ways in which this condition might not be prises. Fragmentation or gradual loss of history features ofPacific salmon (in par­ met with Pacific salmon. For example,

20 Marine Fisheries Review adults may be sampled during only part populations, suggesting that this could upon, and many of the ideas expressed of the spawning run, or only in limited lead to absurdities such as the listing of in that earlier paper appear here in sim­ areas ofa stream. Undercertaincircum­ squirrels in a specific city park2 . Such ilar form. Second, NMFS solicited pub­ stances, samples ofjuveniles may include a result is unlikely under the present lic comments on a summary ofthe 1981 large numbers of individuals from the framework. Although squirrels in a park paper, and a number of those sugges­ same family. Methods for sampling might at present be effectively isolated tions, including ideas discussed at ESA either life history stage may select for from squirrels in other parks or natural Technical Committee meetings, have certain types of individuals. habitat, such a population would be been incorporated into the present doc­ unlikely to meet the second criterion ument. Third, many ofthe concepts out­ Significance ofResults for an ESU (evolutionary significance). lined here were discussed in one form Itis important to realize that' 'statistical Similarly, a population unit recently or another at the FWS-NMFS Verte­ significance" is a different concept than isolated as a result ofhuman activity (by brate Population Workshop held in "evolutionary significance" as it relates the construction ofa dam, for example) Washington, D.C., in June 1990. The to the Act. In the present context, a sta­ probably also does not meet the second goal of the workshop was to develop a tistically significant result indicates that criterion, because most of the diversity consistent approach for determining the means for a pair or group ofsamples presumably would be retained in the whether vertebrate populations are differ by more than would reasonably be larger source population. The same "distinct" underthe Act. In particular, expected if a single population were might be true for some natural isolates, I want to acknowledge Andrew Dizon sampled repeatedly. The conclusion, particularly those of recent origin. (NMFS, La Jolla, Calif.) for introducing then, would be thatthe population means Many will have noted that although this the ESU concept and Mark Shaffer are different for the character under con­ paper establishes a framework for con­ (FWS, Washington, D.C.) for stress­ sideration. Being' 'different," however, sidering populations of Pacific salmon ing the importance of considering is not the same as being' 'distinct' , under under the Act and provides guidance for ecological and habitat characteristics. the Act. Fora population to beconsidered its application, it does not provide a sim­ Fourth, public comments on the NMFS an ESU, it must differ from other popula­ ple formula for determining whether the Interim Policy and accompanying tech­ tions in an evolutionarily important way. unit under consideration is a "species. " nical paper were useful in preparing this Statistical tests can be useful in making To the extent that the process would be revised document. Finally, a number of this determination but do not in them­ simpler and more objective with such a people within NMFS and FWS and the selves provide direct evidence regard­ formula, the approach adopted here is a NMFS Peer Review Group critically ing evolutionary significance. Similarly, disadvantage. However, use ofa simple reviewed drafts of the manuscript, and failure to find a statistically significant (or even a complex) formula does not Pat Montanio (NMFS, Silver Spring, difference does not disprove the exis­ seem consistent with the stipulationofthe Md.) and Karl Gleaves (Office ofNOAA tence of population differences. Power Actto make decisions" solely on thebasis General Counsel, Silver Spring, Md.) to detect true differences in population ofthe best scientific and commercial data provided valuable guidance relating to means is a function ofsample size, so this available. " The process ofevolution and the legislative and legal history ofthe En­ factor should also be considered in differentiation within and between spe­ dangered Species Act. evaluating results of statistical tests. cies is manifest in so many different ways that no simple yardstick will be univer­ Literature Cited General Comments sally applicable. Ryder (1986: 10) came Allendorf, F. W., and N. Ryman. 1987. Genetic management of hatchery stocks. In N. Ryman This paper presents a simple, flexible to essentially the same conclusion re­ and F. Utter (Editors), Population genetics and framework for interpreting language in garding the difficulties in identifying fishery management, p. 141-159. Univ. Wash. the Endangered Species Actpertaining to important conservation units within Press, Seattle. Aspinwall, N. 1974. Genetic analysis of North vertebrate populations: Simple because mammalian species: American populations of the pink salmon (On­ a pair ofcriteria can be applied to deter­ corhynchus gorbuscha): Possible evidence for the mine whether a population segment is "IdentificationofESU's within a species neutral mutation-random drift hypothesis. Evolu­ tion 28:295-305. distinct and hence a "species" for pur­ was recognized as adifficult task, requir­ Barlow, G. W. 1961. Causes and significance of poses ofthe Act, and flexible becausethe ing the use ofnatural history information, morphological variation in fishes. Syst. Zool. morphometrics, range and distribution 10: 105-117. two criteria can be used to address a vari­ Bjomn, T. C., andC. R. Steward. 1990. Concepts ety of issues of particular concern for data, as well as protein electrophoresis, for a model to evaluate supplementation ofnatural Pacific salmon. cytogenetic analysis, and restriction salmon and steelhead stocks with hatchery fish. mapping of nuclear and mitochondrial In W. H. Miller(Editor), Analysis ofsalmon and By focusing on evolutionary signifi­ steelhead supplementation, pI. 3,30p. Bonneville cance, the ESU concept also provides a DNA." Power Admin., Portland, Oreg. means ofdealing with several recurring Clayton, J. W. 1981. The stock concept and the un­ Acknowledgments coupling oforganismal and molecularevolution. problems posed by the term "distinct Can. J. Fish. Aquat. Sci. 38:1515-1522. population segment. " For example, in This paper has drawn heavily on sev­ Dizon, A. E., C. Lockyer, W. F. Perrin, D. P. DeMaster, and J. Sisson. 1992. Rethinldng the 1979 the GAO pointed out potential eral sources. First, the paper by Utter stock concept: A phylogeographic approach. abuses in the authority to list vertebrate (1981) laid a solid foundation to build Conserv. BioI. 6(1):24-36.

53(3),1991 21 Donaldson, L. R., and D. Menasveta. 1961. Selec­ Nei, M. 1973. Analysis of gene diversity in sub­ Utter, F. 1981. Biological criteria for definitionof tive breeding of chinook salmon. Trans. Am. divided populations. Proc. Nat!. Acad. Sci. 70: species and distinct intraspecific populations of Fish. Soc. 90: 160-164. 3321-3323. anadromous salmonids under the U.S. Endan­ Endler, J. A. 1986. Natural selection in the wild. Nelson, K., and M. Soule. 1987. Genetical con­ geredSpecies Act of 1973. Can. J. Fish. Aquat. Princeton Univ. Press, Princeton, N.J. servation of exploited fishes. In N. Ryman and Sci. 38: 1626-1635. Foote, C.J.,C. C. Wood, andR. E. Withler. 1989. F. Utter (Editors), Population genetics and fishery -----=:-c::-----:--, D. Campton, S. Grant, G. B. Milner, Biochemical genetic comparison of sockeye management, p. 345-368. Univ. Wash. Press, J. Seeb, and L. Wishard. 1980. Population struc­ salmon and kokanee, the anadromous and non­ Seattle. tures of indigenous salmonid species of the Pa­ anadromous forms ofOncorhynchus nerka. Can. Quinn, T. P. 1984. Homing and straying in Pacific cific Northwest. In W. J. McNeil and D. C. J. Fish. Aquat. Sci. 46: 149-158. salmon.lnJ. D. McCleave, G. P. Arnold, J. J. Himsworth (Editors), Salmonidecosystemsofthe Groot, C., andL. Margolis (Editors). 1991. Pacific Dodson, and W. H. McNeil (Editors), Mech­ North Pacific, p. 285-304. Oreg. State Univ. salmon life histories. UBC Press, Univ. Brit. anisms ofmigration in fishes, p. 357-362. Plenum Press, Corvallis. Columbia, Vancouver, 564 p. Press, N.Y. Vreeland, R. R. 1986. Evaluationofthecontribu­ Hartt, A. C. 1962. Movementofsalmon in the North and K. Fresh. 1984. Homing and tion ofchinook salmon reared at Columbia River Pacific Ocean and Bering Sea as determined by straying in chinook salmon (Oncorhynchus hatcheries to the Pacific salmon fisheries. Final tagging, 1956-1958.Int. North Pac. Fish. Comm. tshawytscha) from Cowlitz River Hatchery, Rep. 1986, Bonneville Power Admin., Portland, Bull. 6, 157 p. Washington. Can. J. Fish. Aquat. Sci. 41: Oreg. 103 p. Healey, M. C. 1983. Coastwide distribution and 1078-1082. Waples, R. S. 1990a. Conservation genetics of ocean migration patterns of stream- and ocean­ Ricker, W. E. 1972. Hereditary andenvironmen­ Pacific salmon. II. Effective population size and type chinook salmon, Oncorhynchus tshawy­ tal factors affecting certain salmonid popula­ the rateofloss ofgenetic variability. J. Heredity tscha. Can. Field-Nat. 97:427-433. tions.lnR. C. Simon and P. A. Larkin (Editors) , 81 :267-276. Hindar, K., N. Ryman,andF. Utter. 1991. Genetic ThestockconceptofPacific Salmon, p. 19-160. ____ . 1990b. Conservation genetics of effects ofcultured fish on natural fish populations. H.R. MacMillan Lectures in Fisheries. Univ. Pacific salmon. II. Estimating effective popula­ Can. J. Fish. Aquat. Sci. 48:945-957. B.C., Vanc. tion size. J. Heredity 81:277-289. Johnson, J. K. 1990. Regional overview ofcoded ____ . 1981. Changes in the average size and ____ . 1991a. Defmitionof" species" under wire tagging of anadromous salmon and steel­ average age of Pacific salmon. Can. J. Fish. the Endangered Species Act: Application to head in Northwest America. Am. Fish. Soc. Aquat. Sci. 38: 1636-1656. Pacific salmon. U.S. Dep. Commer., NOAA Symp.7:782-816. Ryder, O. A. 1986. Species conservation and Tech. Memo. NMFS F/NWC-194, 29 p. Kirpichnikov, V. S., G. A. Muske, A. D. Scholl­ systematics: Thedilemmaofsubspecies. Trends ----:_--,--_ . 1991b. Genetic interactions between Engberts, V. M. Chernov, and S. N. Borch­ Ecol. Evol. 1(1):9-10. hatchery and wild salmonids: Lessons from the senius. 1990. Genetic structure and allele fre­ Seeb,J. E., L. W. Seeb,andF. M. Utter. 1986. Use Pacific Northwest. Can. J. Fish. Aquat. Sci. quency dynamics in the sockeye salmon popula­ of genetic marks to assess stock dynamics and 48(SuppI.1): 124-133. tion ofLake Dalneye, Kamchatka. Aquaculture management programs for chum salmon. Trans. _--,-_--,- , and P. E. Smouse. 1990. Gametic 84: 13-25. Am. Fish. Soc. 115:448-454. disequilibrium analysis as a means ofidentifying Lande, R., andG. F. BarrowcJough. 1987. Effec­ Shaffer, M. L. 1981. Minimum population sizes for mixtures ofsalmon populations. Am. Fish. Soc. tive population size, genetic variability, and their species conservation. Bioscience 31: 131-134. Symp. 7:439-458. use in population management. In M. E. Soule Simberloff, D. 1988. The contribution ofpopula­ ____ , and D. J. Teel. 1990. Conserva­ (Editor), Viable populations for conservation, p. tion and community biology to conservation tion genetics of Pacific salmon. I. Temporal 87-123. Camb. Univ. Press, Camb., U.K. science. Ann. Rev. Ecol. Syst. 19:473-511. changes in allele frequency. Conserv. BioI. 4: Lane, S., A. J. McGregor, S. G. Taylor, andA. J. Simon, R. C., J. D. McIntyre, and A. R. Hem­ 144-156. Gharrett. 1990. Genetic marking ofan Alaskan ingsen. 1986. Family size and effective popula­ Washington, P. J. 1985. Survey ofartificial produc­ pink salmon population, with an evaluation of tion size in a hatchery stock of coho salmon tion of anadromous salmonids in the Columbia the markand the marking process. Am. Fish. Soc. (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. River basin. Final Rep 1985, Bonneville Power Symp.7:395-406. 43:2434-2442. Admin., Portland, Oreg., 221 p. Lichatowich,J. A., andJ. D. McIntyre. 1987. Use Slatkin, M. 1985. Rare alleles as indicators ofgene Withler, F. C. 1982. Transplanting Pacific of hatcheries in the management of Pacific flow. Evolution 39:53-65. salmon. Can. Tech. Rep. Fish. Aquat. Sci. 1079, anadromous salmonids. Am. Fish. Soc. Symp. ____ . 1987. Gene flow and the geographic 27 p. 1:131-136. structure of natural populations. Science 236: Wright, S. 1978. Evolution and the genetics of Meffe, G. K. 1986. Conservation genetics and the 787-792. populations, volA. Variability within and among management of endangered fishes. Fisheries Soule, M. E. 1987. Viable populations for conser­ natural populations. Univ. Chicago Press, 11(1): 14-23. vation. Camb. Univ. Press, Camb., U.K. Chicago, 580 p.

22 Marine Fisheries Review