The joint policy of the U.S. Fish and Wildlife Service and the National Marine Fisheries Service provides guidelines for defining distinct population segments below the taxonomic level of species (USFWS-NMFS 1996). The first of two elements to be considered is the discreteness of a population segment with respect to the rest of the populations within the species. Discreteness may result from physical factors that isolate the population segment and may be reflected as life-history differences in physiology, ecology, or behavior between the DPS and other populations. Genetic or morphological differences between the population segment being considered and other populations may also be used to evaluate discreteness. The policy also states that international boundaries within the geographical range of the species may be used to delimit a distinct population segment in the United States. This criterion is applicable if differences in the control of exploitation of the species, the management of the species’ habitat, the conservation status of the species, or regulatory mechanisms differ between countries that would influence the conservation status of the population segment in the United States. In past assessments of evolutionarily significant units (ESUs) in Pacific salmon, however, NMFS has placed the emphasis on biological information in defining DPSs and has considered political boundaries only at the implementation of ESA listings.
A second element in defining distinct population segments is that the segment must be biologically or ecologically significant. Significance is evaluated in terms of the importance of the population segment to the overall welfare of the species and may be considered in the light of, but not limited to, the following factors. The population segment may be considered significant if it persists in an unusual or unique ecological setting for the species. A population segment may also be considered significant, if its loss would result in a significant gap in the geographical range of the species. Such a gap may disrupt the normal connectivity between populations. A segment also meets the significance guideline, if it represents the only surviving natural occurrence of the species that may be more abundant elsewhere as an introduced population outside its historical range. Another guideline is that the population segment differs markedly in its genetic characteristics from other populations of the species. Genetic differences may be detected by molecular genetic methods or may be reflected in unique adaptations to habitats not found in other parts of the species’ geographical range. Other classes of information may also bear on the biological or ecological importance of a discrete population segment.
We considered several kinds of information in this status review to attempt to delineate DPSs of Pacific herring, particularly the geographic limits of the DPS which contains herring from Puget Sound. The first kind of information was to consider geographical variability in life-history characteristics and morphology. Such traits usually have an underlying genetic basis, but are often strongly influenced by environmental factors from one locality to another. The second kind of information consisted of tag and recapture studies, which give insight into the physical movement of individuals between areas. The third kind of information consisted of traits that are inherited in a predictable way and remain unchanged throughout the life of an individual. Differences among populations in the frequencies of markers at these traits may reflect isolation between the populations. The analyses of these kinds of information are discussed briefly in the following sections.
The analysis of habitat characteristics may indicate that a population segment occupies an unusual or distinctive habitat, relative to the biological species as a whole. The persistence of a discrete population segment in an ecological setting unusual or unique for the taxon is one factor identified in the joint DPS policy (USFWS-NMFS 1996) that may provide evidence of the population’s significance. However, Waples (1991a) cautioned against "drawing inferences based on physical characteristics of the habitat without supporting biological information linking the habitat differences to adaptations."
Conversely, the continuous distribution of a population segment within a region possessing similar habitat and ecological characteristics makes it less likely that unique adaptations have arisen in local populations. Without associated compelling phenetic or genetic evidence for a finer population structure, marine fish continuously distributed throughout similar habitat and lacking physical or behavioral barriers to migration are not likely to be composed of multiple DPSs.
Isolation between populations may be reflected in several life-history variables, including differences in behavior (e.g., spawning timing, migration) and demography (e.g., growth rate, fecundity, age structure), among others. Although some of these traits may have a broad genetic basis and may reflect local adaptations of evolutionary importance, they are usually strongly influenced by environmental factors over the lifetime of an individual or over a few generations. Differences can arise among populations in response to environmental variability among areas and they can sometimes be used to infer the degree of independence among populations. However, differences in phenotypic and life-history traits among populations do not provide definitive information on reproductive isolation between populations, because the genetic basis of many phenotypic and life-history traits is weak or unknown. Likewise, elemental profiles present in otoliths, and other structures, reflect local environmental conditions or diets and although they may indicate that different areas or environments are occupied, they also provide little definitive information on the degree of reproductive isolation between populations.
Variation in reproductive behavior within a species of marine fish is an important factor to consider because it may provide the isolating mechanism required for differentiation. The presence of geographically-discrete and temporally-persistent spawning aggregations in a species indicates that reproductive isolation may be occurring. However, it is necessary to evaluate the degree of reproductive isolation by addressing the questions of migration rate, gene flow, and re-colonization rate. These later considerations are dependent on the degree of homing ability and natal-site fidelity of adults.
Studies of parasite incidence can provide important information about the degree of intermingling of marine fish stocks, particularly when a parasite is present in one area and totally absent in an adjoining area. However, parasite studies have some inherent interpretation problems: 1) in most cases parasite incidences exhibit clinal trends with latitude, and the degree to which parasite occurrence is due to environmental differences, acting on the parasite, or to a lack of host stock intermingling, is unknown, 2) the lack of a parasite in an area may be due to a regional absence of an alternate host organism, independent of host distribution, and 3) parasites may not be permanent natural tags in that parasites may be lost during the lifetime of the host.
The analysis of applied or acquired tags can indicate the degree of migration between localities. These tags consist of physical tags that are attached to a fish and later recovered. These tags provide evidence of movement of individuals from one place to another, but not necessarily of population connectivity through gene flow. Since these kinds of population markers largely lack a genetic basis and are not inherited, they must be applied each generation or must arise naturally anew each generation.
The application and recovery of physical tags on adult marine fish on spawning grounds can answer the question of whether fish return to the same locality to spawn in subsequent years, but these studies lack the direct evidence of parent-offspring linkage. In other words, these studies do not provide direct evidence that fish return to their natal area for spawning; however, they may provide evidence of straying and thus, the potential for gene flow between spawning aggregations.
Two problems inherent in the use of morphometric and meristic characteristics to separate marine fish populations are: 1) the characteristics are often under strong environmental influence and are not inherited in a simple Mendelian fashion, and 2) the characteristics are continuously variable and exhibit clinal trends and a high variance about the mean. A further drawback of using morphometric and meristic characteristics to detect population structure in fish is that few of these characteristics have been examined from a genetic standpoint. As shown by studies on several species, environmental parameters such as temperature, salinity, pH, and oxygen concentration can modify the expression of genes responsible for meristic characters (Ihssen et al. 1981).
Molecular genetic evidence can be used to define reproductively-isolated populations or groups of populations of Pacific herring in Puget Sound, as well as throughout the range. Molecular genetic markers appear to be largely unaffected by natural selection, so that geographical differences in gene frequencies can be interpreted in terms of genetic flow and genetic drift. The analysis of the geographical distributions of these markers may reveal historical dispersals, equilibrium levels of migration (gene flow), and past isolation. Evidence for genetic population structure is based on the analysis of protein variants (allozymes), microsatellite loci (variable numbers of short tandem DNA repeats), and mitochondrial DNA (mtDNA).
Evidence of substantial genetic divergence between populations, as shown through analysis of these neutral molecular markers, is an important aspect of distinctiveness because even a small amount of interbreeding between populations will reduce the genetic differentiation between them. Although these molecular genetic methods "provide valuable insight into the process of genetic differentiation among populations" they offer "little direct information regarding the extent of adaptive genetic differences" (Waples 1995).
One widely used method of population analysis is sequence or RFLP (restriction fragment length polymorphism) analysis of mtDNA, which codes for several genes that are not found in the cell nucleus. Mitochondrial DNA differs from nuclear DNA (nDNA) in two important ways. One way is that recombination is lacking in mtDNA, so that gene combinations (haplotypes) are passed unaltered from one generation to the next, except for new mutations. A second way is that mtDNA is inherited from only the maternal parent in most fishes, so that gene phylogenies correspond to female lineages. A greater amount of random genetic drift among populations is expected for mtDNA genes, because the effective population size for mtDNA is about one-fourth of that for nuclear genes. These characteristics permit phylogeographical analyses of mtDNA haplotypes, which can potentially indicate dispersal pathways for females and the extent of gene flow between populations (Avise 1994).
Microsatellite DNA markers can potentially detect stock structure on finer spatial and temporal scales than can other DNA or protein markers, because of higher levels of polymorphism found in microsatellite DNA (reflecting a high mutation rate). When populations are at least partially isolated, genetic markers at loci with high mutation rates may accumulate more rapidly in some areas than in others.
Several standard statistical methods have been used to analyze molecular genetic data to detect reproductive isolation between populations. Comparisons of genotypic frequencies in a sample with frequencies expected under random mating (Hardy-Weinberg proportions) may be used to infer the breeding structure of a population or to detect population mixing (Wahlund's effect). Contingency-table comparisons of allozyme or microsatellite allele frequencies among population samples with chi-square or G (log-likelihood ratio) test statistics, or with randomization tests, can be used to detect significant differences between populations, which may be evidence of reproductive isolation.
A complementary way of assessing genetic isolation between populations is to analyze genetic distances based on allele-frequency estimates. Several genetic distance measures (Cavalli-Sforza and Edwards 1967, Nei 1972, 1978) have been used to study the population genetic structure of anadromous salmonids. It is unclear, however, which measure is most appropriate in a particular case or whether there is one measure that is always most appropriate. Discussions of the features of genetic distances appear in Nei (1978), Rogers (1991), and Hillis et al. (1996). Most of this discussion has focused on the merits of the various measures for phylogenetic reconstruction among species or higher taxa.
Sample sizes and heterozygosity may also influence the power of the genetic-distance approach to resolving genetic population structure. When sample sizes used to estimate allelic frequencies are 50 individuals or more, the difference between Nei's genetic distance, D, (Nei 1972) and Nei's unbiased genetic distance (Nei 1978) is small in absolute terms, but still might be a substantial proportion of D, if D is small. When genetic distances between populations are also small, as they often are between populations of marine fishes, low but significant levels of genetic differentiation may not be detected by an unbiased distance measure because sample-size corrections may reduce estimates of genetic distance to zero. These measures range from 0.0 (identity) to infinity (complete dissimilarity). In most cases, the different genetic-distance measures yield highly-correlated results.
The degree of reproductive isolation between populations can be inferred from an analysis of the pattern of genetic distances between populations. Clustering methods, such as the unweighted pair group method with averages (UPGMA, Sneath and Sokal 1973) and the neighbor-joining method (Saitou and Nei 1987), find hierarchical groupings of genetically similar populations. Multivariate methods, such as multidimensional scaling (MDS, Kruskal 1964) or principal components analysis (PCA), find groupings of genetically-similar populations in several dimensions, which are depicted here in two or three dimensions.
Various studies have estimated levels of genetic variability within populations, because the level of within-population variability may reflect evolutionary or historical differences in population size and in migration patterns between populations. Within-population gene diversity was measured by the expected proportion of heterozygous genotypes in a population of randomly mating individuals averaged over the number of loci examined (H). Estimates of H based on a small number of individuals are usually accurate, as long as a large number of loci (>30) are surveyed for variability (Nei 1978).
Genetic differentiation between populations at various hierarchical levels has been estimated in many studies with a gene diversity analysis (Nei 1973, Chakraborty 1980), which apportions allele-frequency variability among populations into its geographical or ecological components. For example, the proportion of the total genetic variability in a set of samples that is due to differences among populations may be estimated with FST or the multiallelic equivalent statistics, GST. These variables range from 0.0 (no difference among populations) to 1.0 (fixed allele-frequency differences). The range 0.05-0.15 for FST indicates moderate differentiation, and the range 0.15-0.25 indicates strong genetic differentiation among populations (Wright 1978). These statistics facilitate comparisons among groups of populations that may reveal regional differences in gene flow between populations.
The term "stock" has been used rather loosely in fisheries management and no single definition has been accepted by all fisheries biologists. Stock may be used to refer to groups of fish being harvested in a particular area, whether these fish are genetically related or not. However, in most cases, identification of a group of fish as a stock implies that these fish are in some way different or distinct from those in another stock, and generally implies some genetic relatedness among its members (Ihssen et al. 1981). Evidence of stock structure may be shown through differences in demographic population statistics (age composition, growth rate, fecundity, etc.), morphology (morphometrics and meristics), or genetics (differentiation at allozyme or DNA loci).
Ricker (1972) defined a salmon stock as "the fish spawning in a particular lake or stream (or portion of it) at a particular season, which fish to a substantial degree do not interbreed with any group spawning in a different place, or in the same place at a different season." Larkin (1972) defined a stock as "a population of organisms, which, sharing a common environment and participating in a common gene pool, is sufficiently discrete to warrant consideration as a self-perpetuating system which can be managed." Booke (1981) provided a general definition of a stock as "a species group, or population, of fish that maintains and sustains itself over time in a definable area." Ihssen et al. (1981) defined a stock as "an intraspecific group of randomly mating individuals with temporal or spatial integrity." In none of these definitions is it implied that a fish stock is ecologically or biologically significant in relation to the biological species as a whole.
By contrast, not only must a marine fish DPS be "markedly separated from other populations of the same taxon," it must also exhibit ecological or biological significance in comparison to other population segments of the biological species. Thus, following the guidance supplied by the joint policy statement (USFWS-NMFS 1996), a distinct population segment of marine fish may be viewed as a group of related stocks (or in some cases, if the evidence warrants, a single stock) that form(s) a discrete population and are (is) significant to the biological species as a whole. As stated previously, considerations that can be used to determine a discrete population’s significance to the taxon as a whole include: 1) persistence of the population segment in an ecological setting unusual or unique for the taxon, 2) evidence that loss of the population segment would result in a significant gap in the range of the taxon, 3) evidence that the population segment represents the only surviving natural occurrence of a taxon that may be more abundant elsewhere as an introduced population outside its historic range, and 4) evidence that the population segment differs markedly from other populations of the species in its genetic characteristics.
Phenetic and genetic information examined for evidence for DPS delineations of Pacific herring included presence of geographically-discrete and temporally-persistent spawning aggregations, and variation in seasonal migration patterns, parasite incidence, growth rate and body size, length and age at maturity, fecundity, and meristics and morphometrics.
Genetic studies searching for population structure in Pacific herring have followed a course similar to many other fishes. Early studies concentrated on finding protein electrophoretic variation using starch-gel electrophoresis. This search in Pacific herring was begun by Utter (1972) who identified two polymorphic allozyme loci from four Pacific herring samples from Washington State (three from Puget Sound, one from the Washington coast). No significant differences in allele frequencies were found among the samples. This research was expanded further to include six polymorphic loci collected from samples ranging from Oregon to Kodiak Island, Alaska (Utter et al. 1974). Again, there was no evidence of significant genetic differentiation among these populations either. It should be pointed out that the Washington coast sample in the Utter et al. (1974) and Utter (1972) studies was composed of immature fish. The other samples, as well as all of the samples in the studies described below, were composed of adult fish collected from known herring spawning grounds.
Grant expanded upon Utter’s initial work, and surveyed 40 allozyme loci in Pacific herring (Grant 1979, 1981; Grant and Utter 1984). Grant and Utter (1984) found 26 polymorphic loci in 21 samples collected from five areas throughout the range of Pacific herring -- Asia, the northeastern Bering Sea, the southeastern Bering Sea, the Gulf of Alaska, and the eastern North Pacific Ocean (including samples from the Strait of Georgia and Hale Passage in Puget Sound) (Fig. 15). Their analyses revealed two main genetic stocks: Asian-Bering Sea herring and eastern North Pacific herring, separated by a Nei’s genetic distance (D) (Nei 1972) of 0.039 (Fig. 16). The authors postulated that these two distinct stocks arose because of restricted gene flow between them, due to repeated Pleistocene glaciation on the southern coast of Alaska. Genetic differentiation was also detected among all five areas. However, only the Bering Sea and the Gulf of Alaska showed any significant genetic differentiation within an area. The samples from the eastern North Pacific Ocean, which include the two Puget Sound samples, were not genetically distinct from each other, even though the collection sites ranged from California to southeast Alaska. The average D value between pairs of samples within the eastern North Pacific Ocean was not significantly different from zero. A gene diversity analysis revealed that only 0.5% of the observed variation was due to differences among populations within an area. The authors also reported a north to south cline in allele frequencies of the locus GAPDH-1* for eastern North Pacific Ocean samples. The reason for this is uncertain, but because no other loci showed a similar cline, one possible explanation is that selection is occurring at this locus.
Similar to Grant and Utter’s study, Kobayashi (1993) used allozyme analyses to conduct a genetic study of Pacific herring throughout their range. The majority of the samples were from Asia (N = 18), however the study also included one sample from San Francisco Bay, one sample from Puget Sound, and three samples from Alaska. Analogous to the findings of Grant and Utter (1984), Kobayashi found significant genetic differentiation between Asia-Bering Sea samples and eastern North Pacific samples, separated by a Nei’s distance value of 0.054 (Fig. 17). Very little differentiation was evident among the southeastern samples. Nei’s distance values between the Puget Sound sample and the sample from Vancouver Island (data was obtained from Grant 1981) and San Francisco Bay, were both less then 0.001.
Unfortunately, the Kobayashi study is the last to include any Pacific herring samples from Puget Sound. Currently, microsatellite data is being collected from Puget Sound herring by the Washington Department of Fish and Wildlife (J. Shaklee4) and mtDNA data from the same samples by the University of Washington School of Aquatic and Fishery Sciences (P. Bentzen5), but the results from these studies is not yet available. However, other genetic studies of Pacific herring have been done, some of which include samples from areas geographically close to Puget Sound.
Schweigert and Withler (1990) examined 12 samples of Pacific herring using seven polymorphic allozyme loci, and a restriction endonuclease analysis of mitochondrial DNA (mtDNA). The majority of their samples (N = 10) were collected from southern British Columbia, while two temporally-spaced samples came from one location in northern British Columbia. One of the southern samples was collected from Yellow Point which lies about 50 km from the northern boundary of Puget Sound. Neither the allozyme nor mtDNA data provided any evidence of significant genetic differentiation among the locations. Their cluster analysis of Nei’s (1978) D resulted in D values ranging from 0.000 - 0.004. A gene diversity analysis showed that 99.6% of the variation observed was due to differences within samples, whereas the amount attributed to variation among samples within years was less then 0.3%. Similar to Grant and Utter (1984), a north to south decline in the GAPDH-1*-50 frequency was observed. They also found that samples taken from the same location in two consecutive years showed considerable temporal variation. The gene diversity analyses revealed almost as much variation among years within areas then variation among samples within years.
Sequence variation in ribosomal DNA of Pacific herring, as well as Atlantic herring was investigated by Domanico, Phillips, and Schweigert (1996). Their sampling sites spanned the entire coast of British Columbia, including Tumbo Channel in the Strait of Georgia, which is only 4 km from the northern boundary of Puget Sound. The restriction site they examined was polymorphic, but based upon the percent difference between fragment patterns (Wayne et al. 1991) there was no stock-specific patterns of differentiation for the six British Columbia locations they sampled.
Several genetic studies of Alaska Pacific herring have also been conducted. Burkey (1986) analyzed 16 samples collected from seven locations by commercial herring fisheries within Prince William Sound, Alaska. He analyzed 14 polymorphic allozyme loci but did not find any significant differences among samples, among locations, or between years within a location. A gene diversity analysis showed that over 99% of the total variability was due to variability within samples.
Seven populations of Alaskan Pacific herring were sampled in two different years and examined for microsatellite variation (Wright et al. 1996, Wright and Dillon 1997, O’Connell et al. 1998a, 1998b) and mtDNA variation (Bentzen et al. 1998). The results of these studies were summarized by Seeb et al. (1999). Differences in microsatellite allele frequencies were significant among all samples, whereas the mtDNA haplotype frequency variation was not significant among samples collected in 1995, but were significant among 1996 Prince William Sound samples. Similar to previous studies, the greatest amount of genetic divergence was between samples from the Bering Sea and the Gulf of Alaska. Analogous to what was found by Schweigert and Withler (1990), samples collected from the same location in different years showed a high degree of genetic differentiation. The authors state that "the magnitude of genetic variation among sampling years within locations was equal to or greater than the magnitude of variation among locations within sea basins." They concluded "the DNA data provide no evidence of stable differentiation among populations within sea basins on spatial scales of up to ~700 km. Rather, the DNA data suggest that temporal variation among spawning aggregations dominates genetic variability on these spatial scales."
Two main conclusions about genetic differentiation among Pacific herring populations can be drawn from these studies. First, Pacific herring show considerable temporal variation in allele frequencies. Bentzen et al. (1998), Wright and Dillon (1997) and Schweigert and Withler (1990) all found significant temporal variation in the samples they analyzed. A high degree of temporal variation has the potential to confound genetic population studies. Ideally, all samples for a study should be sampled in the same year. Such was the case for the majority of the studies reviewed here. Additionally, sampling all locations in multiple years as Wright and Dillon (1997) and Bentzen et al. (1998) did, will provide valuable information regarding the nature of any observed genetic variation.
Second, Pacific herring have comparatively low levels of genetic differentiation among populations. According to Hartl (1980), gene diversity values of 0.05 - 0.15 indicate moderate differentiation among populations. Reported gene diversity values for Pacific herring of 0.005 (Grant and Utter 1984), 0.004 (Burkey 1986), 0.003 (Schweigert and Withler 1990), 0.013 (mtDNA) and 0.030 (microsatellites) (Seeb 1999) for comparisons among samples within a predefined area, are all below this range. While some genetic differentiation was evident in Alaskan samples, neither Utter et al. (1974), Grant and Utter (1984), or Kobayashi (1993) found any evidence of significant genetic differentiation between Puget Sound herring populations and California, Oregon, British Columbia, or southeast Alaska herring populations. Grant and Utter (1984) determined that "very little migration is required to maintain genetic homogeneity at the very large population sizes that are characteristic of herring." Significant migration among Pacific herring populations would result in a high degree of gene flow, and thus little to no genetic differentiation among populations.
Pre-historical and historical persistence in Puget Sound—Tunnicliffe et al. (in press) examined fish remains in a complete Holocene sediment core sequence from Saanich Inlet, Vancouver Island, British Columbia. Pacific herring were one of the first fish species to occur in Saanich Inlet following glacial retreat from the region, after approximately 12,000 years before present (BP) ( Tunnicliffe et al. in press). Fish abundance and species diversity peaked in Saanich Inlet between 7,500 and 6,000 BP, and the last 1,000 years have seen some of the lowest abundances of fishes in Saanich Inlet’s marine history (Tunnicliffe et al. in press). The close proximity of Saanich Inlet to Puget Sound would suggest that Pacific herring were also likely established in Puget Sound by about 12,000 BP. Pacific herring were identified in prehistoric fish skeletal remains from the Duwamish No. 1 archeological site (45-KI-23), located 3.8 km upstream from Elliott Bay on the Duwamish River, utilized by aboriginal humans between A.D. 15 and A.D. 1654 (Butler 1987). However, Pacific herring remains were infrequently found at this site and the family Clupeidae ranked 19th out of the 25 fish groups, in order of abundance (Butler 1987). In historic times, Pacific herring were reported as "exceedingly abundant" in Puget Sound by Jordan and Starks (1895).
The overall distribution and spawn timing of Pacific herring stocks in North America were reviewed by Scattergood et al. (1959), Hay (1985), and Haegele and Schweigert (1985). Pacific herring spawning has been reported in sheltered inlets, sounds, bays and estuaries from San Diego, California in the southern extreme of the range, north along the West Coast of North America to Cape Bathurst in the Beaufort Sea (Hay 1985, Haegele and Schweigert 1985). In general, Pacific herring spawn timing varies with latitude, and over their entire range spawning occurs for nearly 10 months of the year (Hay 1985). The earliest spawning reportedly occurs in southern California in early fall and the latest occurs in August in Bristol Bay, Alaska (Hay 1985). However, it is also apparent that in some regions, the earliest and latest spawnings can be up to six months apart. Figures 13a-d summarize selected information on spawn timing of Pacific herring.
In some locations, Pacific herring are known to spawn in several discrete waves that are separated by several days to weeks, although individual female herring deposit all their eggs in one to two days. It has also been noted that the older and larger fish tend to spawn in the earliest wave, with subsequent waves being made up of smaller fish (Hay 1985, Ware and Tanasichuk 1989, Hay 1990). Pacific herring in some aggregations, such as in Barkley Sound on the west coast of Vancouver Island, can hold in a state of full maturity for several days to weeks, whereas herring in more protected inshore spawning locations usually spawn as soon as full maturity is achieved (Ware and Tanasichuk 1989). Ware and Tanasichuk (1989) examined maturation rates in selected groups of male and female Pacific herring in British Columbia during the month prior to spawning for 1982-87. These latter authors found that the influence of water temperature and body size on the maturation rate (as measured by the gonosomatic index (GSI)) could explain: 1) why Pacific herring spawn in waves, 2) much of the variation in spawn timing between regions, and 3) much of the year-to-year variation in spawn timing at a given location.
Hay (1990) suggested that differences in spawn timing of Pacific herring stocks could possibly be explained by herring having "spawning times that match local zooplankton production schedules, particularly the time of egg production by copepods because copepod eggs are, overwhelmingly, the dominant food organism of larval herring."
California—According to Miller and Schmidtke (1956) and Spratt (1981), Pacific herring have been known to spawn in California in the following estuaries: 1) Crescent City Harbor, 2) Humboldt Bay, 3) Shelter Cove, 4) Noyo River, 5) Russian River, 6) Bodega Bay, 7) Tomales Bay, 8) San Francisco Bay, 9) Elkhorn Slough, 10) Monterey Bay 11) Morro Bay, 12) San Luis River, and 13) San Diego Bay (Fig. 11). Miller and Schmidtke (1956) stated that spawning is also believed to occur in Los Angeles Harbor, Santa Ynez Lagoon, Drakes Bay, and at Fort Bragg; however, the spawning grounds have not been documented in these areas. Within California, the Tomales Bay and San Francisco Bay populations have the largest populations (Barnhart 1988).
Since larval herring have been found in San Francisco Bay as early as October (Eldridge and Kaill 1973), Hay (1985) stated that "earliest spawnings probably occur in the early fall in California." Barnhart (1988) stated that herring spawn from November to June in California, although most spawning occurs from December to February. Figure 13a summarizes known spawn timing for Pacific herring in California.
Oregon—Because herring are of minor economic importance in Oregon, only limited information is available on the species (Scattergood et al. 1959). Scattergood et al. (1959) stated that principal spawning grounds for Pacific herring in Oregon included Yaquina, Tillamook, and Coos bays (Fig. 11). Lassuy (1989) indicated that spawning also occurs in the vicinity of Reedsport, Oregon (Winchester Bay) and in the Columbia River estuary (Fig. 13a). Spawning was reported to occur in Yaquina Bay from January to April and in Tillamook Bay from February to April (Scattergood et al. 1959) (Fig. 13a).
Washington—Currently, WDFW recognizes eighteen spawning stocks of Pacific herring in Puget Sound: 1) Squaxin Pass, 2) Quartermaster Harbor, 3) Port Orchard-Port Madison, 4) South Hood Canal, 5) Quilcene Bay, 6) Port Gamble, 7) Kilisut Harbor, 8) Port Susan, 9) Holmes Harbor, 10) Skagit Bay, 11) Fidalgo Bay, 12) Samish Bay-Portage Bay, 13) Interior San Juan Islands, 14) Northwest San Juan Islands, 15) Semiahmoo Bay, 16) Cherry Point, 17) Discovery Bay, and 18) Dungeness Bay (Lemberg et al. 1997, O’Toole et al. 2000). Geographic distribution of the spawning sites for these stocks are illustrated in Figure 12. Detailed descriptions of these spawning stocks can be found in Lemberg et al. (1997) and O’Toole (2000). Most Puget Sound Pacific herring stocks, as recognized by WDFW(1998), spawn on multiple shoreline locations within a restricted geographic location. For example, the Squaxin Island stock in South Puget Sound spawns in the mouth of Hammersley Inlet, in Totten Inlet at Gallagher Cove, and in Squaxin Passage. Similarly, the Interior San Juan Islands herring stock is a combination of spawners from East Sound and West Sound on Orcas Island; Mud, Hunter, and Swifts bays on Lopez Island; and Blind Bay on Shaw Island.
O’Toole (2000) provided an historical overview of spawning ground locations for Pacific herring in Puget Sound as gathered from the existing literature, which commences with observations by Chapman et al. (1941), and contrasts these with locations currently supporting spawning aggregations. According to O’Toole (2000), most of the Pacific herring spawning grounds that Chapman et al. (1941) reported as existing in their 1936-37 surveys are included in the grounds or stocks that WDFW currently assesses for abundance. These include Cherry Point, Portage Bay, Semiahmoo Bay, Northwest San Juan Islands, Interior San Juan Islands, Fidalgo Bay, Discovery Bay, Sequim Bay (designated Washington Harbor in Chapman et al. 1941), Kilisut Harbor, Holmes Harbor, Port Orchard-Port Madison, Quartermaster Harbor, and certain spawning grounds in central Hood Canal and in southern Puget Sound south of the Tacoma Narrows.
According to O’Toole (2000), Pacific herring spawning locations that were observed in 1936-37 by Chapman et al. (1941) that no longer support spawning aggregations include Port Blakely and Rolling Bay on the east side of Bainbridge Island and Wollochet Bay in southern Puget Sound. Chapman et al. (1941) and Katz (1942) also identified Echo Bay and Shallow Bay on Sucia Island and a small bay on the President Channel side of Waldron Island in the San Juan Islands as herring spawning locations. Updated information in Koenings (unpubl. data) shows that Wollochet Bay was used as a spawning ground by Pacific herring in the year 2000. Koenings (unpubl. data) also noted that "there has been a relative lack of exploratory herring spawn deposition survey efforts" over the years, and that no recent-era WDFW spawn surveys have been conducted in Port Blakely, Eagle Harbor, Rolling Bay, Shallow Bay, Echo Bay, or near Waldron Island. Thus, it is uncertain whether or not every herring spawn site reported in Chapman et al. (1941) is currently utilized by Pacific herring.
Pacific herring do not apparently utilize these areas for spawning at the present time. In addition, O’Toole (2000) identified a number of spawning grounds for Pacific herring that were not included in the list published by Chapman et al. (1941), but have since been documented: 1) Port Gamble, 2) Quilcene Bay, 3) Dungeness Bay, 4) Port Susan, 5) Samish Bay, 6) South Hood Canal, 7) Skagit Bay, and 8) the Point Whitehorn to Lummi Bay and Point Roberts portions of the Cherry Point stock.
Chapman et al. (1941) referred to an unpublished report of observations of Pacific herring spawning grounds in Puget Sound made in 1927 by Arthur S. Einarsen of the Washington Department of Fisheries. According to Chapman et al. (1941), the Einarsen report listed the spawn timing of the Birch Bay (Cherry Point) population as occurring from May 1st to June 10th in 1927. These dates are within the range of spawn dates of late-March to mid-June reported for recent years at Cherry Point by O’Toole (2000). According to Chapman et al. (1941), the Einarsen report listed Eagle Harbor, on the east side of Bainbridge Island, as supporting a spawning aggregation of Pacific herring during the first three weeks of February. Apparently, no Pacific herring spawning has been reported in Eagle Harbor since the Einarsen report (Chapman et al. 1941, O’Toole 2000).
Within Puget Sound, major Pacific herring stocks spawn from late-January through early-April (Trumble 1983b, Lemberg et al. 1997, O’Toole 2000). An exception to this is the Cherry Point stock, which spawns from early April through early June (Lemberg et al. 1997, O’Toole 2000) (Figs. 12 and 13b). According to Lemberg et al. (1997) each stock generally spawns over approximately a two-month period. In summary, O’Toole et al. (2000) stated that:
Although some changes in Puget Sound herring spawning behavior have been observed, the consistency of timing and specific spawning locations, comparing historical and current descriptions, is remarkable. Descriptions of peak spawning timing to date typically vary by two weeks or less and most spawning locations have shifted very little, if at all.
On the outer coast of Washington, Pacific herring spawning has been reported to occur in Willapa Bay during February (Chapman et al. 1941, Katz 1942, Lemberg et al. 1997). Chapman et al. (1941) reported that a small number of Pacific herring spawn in Grays Harbor on an irregular basis. Although Lemberg et al. (1997) stated that a survey of Grays Harbor in 1988 "failed to yield any evidence of spawning activity," Koenings (unpubl. data) revealed that herring spawn activity has been documented by WDFW in the South Bay/Elk River estuary of Grays Harbor, annually from 1998-2000. Lassuy (1989) indicated that Pacific herring spawn in the Columbia River estuary (Fig. 11). Lemberg et al. (1997) also stated that spawning activity has been reported from the Ilwaco, Washington and Hammond, Oregon areas of the Columbia River estuary; however, this reported activity has not been documented. Lemberg et al. (1997) stated that currently "the only documented Washington coastal herring spawning stock is the Willapa Bay stock."
British Columbia—Records of Pacific herring spawning activities have been collected in British Columbia since 1928, originally by the Fisheries Research Board of Canada, and in recent times by the Department of Fisheries and Oceans Canada (DFO) (Hay and Kronlund 1987, Hay et al. 1989a-f, Hay and McCarter 1999a). These records generally include data on spawn timing, the location of spawning, the shoreline width and length of spawning, and spawning intensity for each location (Hay and Kronlund 1987). Pacific herring spawn data up to 1986 were described in a detailed six-volume publication (Hay et al. 1989a-f) and data through the 1999 season are available at the DFO operated Herring Spawn Home Page at http://www_sci.pac.dfo_mpo.gc.ca/herspawn/default.htm. Approximately 1,300 locations have been identified in British Columbia as having had at least one or more Pacific herring spawnings since 1928 (Hay and Kronlund 1987, Hay and McCarter 1999a, 2000). Generalized locations of major Pacific herring spawning grounds in British Columbia are illustrated in Figures 18 and 19. Detailed spawning site information can be found at the above mentioned website.
Currently, for management purposes, DFO recognizes six Pacific herring management regions in British Columbia: 1) Queen Charlotte Islands, 2) the North Coast British Columbia (Prince Rupert District), 3) the Central Coast, 4) Johnstone Strait, 5) the Strait of Georgia, and 6) the west coast of Vancouver Island. Each of these Regions is further divided into Statistical Areas, which are further divided into Sections (= Subareas), each of which is named and numbered (Hay and McCarter 2000). The boundaries of each of the 108 Pacific herring Sections are illustrated by Region in Figure 20. Schweigert (unpubl. data) stated that a "location" within a Section is an "artificial construct, usually a local geographic name used to identify a section of shoreline" and that within a Section, "locations are often contiguous and often differ markedly in size." Figures 13c and 13d provide mean spawning day of the year (± one standard deviation) and earliest and latest spawning day of the year for most of the Pacific herring Sections in British Columbia.
In general, Pacific herring spawn from January to May in southern British Columbia and from mid-January to June in northern British Columbia (Taylor 1964, Hourston 1980) ( Figs. 13a, c, d). Outram and Haegele (1969) found a difference of six weeks in the mean spawn timing (from March 8-9 to April 20-21) that occurred between spawning areas in extreme southern and northern British Columbia. However, several exceptions to these generalities occur. For instance, Pacific herring in two geographically adjacent spawning sections in the northern Queen Charlotte Islands, Masset Inlet (Section 011) and Naden Harbour (Section 012), possess some of the extreme latest and earliest spawn timings on the British Columbia coast, respectively (Hay 1985) (Figs. 13c, 20 and 21). Masset Inlet spawnings have been documented in late-June to July (Hay 1985, 1990, Hay and McCarter 1999a) with a mean spawn date of June 20 (DFO 2000b), while Naden Harbour spawnings may occur in late-January or in early-February (Hay 1985, 1990, Hay and McCarter 1999a) with a mean spawn date of March 2 (DFO 2000b) (Fig. 13c, and 21). Pacific herring in other nearby spawning sections in the Queen Charlotte Islands Region spawn mainly in April (Hay 1985). Pacific herring from Skidegate Inlet (Section 022, mean spawn date of May 14) and Burke Channel (Section 084, mean spawn date of June 1) also possess consistently later spawn timings than other Sections in their respective regions (Hay 1985, DFO 2000b) (Figs. 13c, 20 and 21).
Barraclough (1967) reported on the occurrence of larval Pacific herring in surface trawls on July 5-8, 1966 in the southern Strait of Georgia, between the Fraser River delta and Vancouver Island. Based on the size of these larval herring, Barraclough (1967) calculated that they were the result of spawning that had occurred between May 22 and June 4, which was considerably later that any previously reported Pacific herring spawn timing in the British Columbia portion of the Strait of Georgia (Barraclough 1967). Based on the counterclockwise flow of currents in the Strait of Georgia, Barraclough (1967) postulated that these larvae had hatched from spawn deposition in the vicinity of Boundary Bay. The estimated timing of this spawn deposition suggests the larvae encountered by Barraclough (1967) were progeny of the WDFW Cherry Point Pacific herring stock.
Hay and McCarter (1999a) examined long-term trends (1928-99) in the minimum, mean, and maximum day of spawning for each of the Pacific herring management regions in British Columbia and found that: 1) the range of spawn timing and the mean spawning day of the year have steadily declined in both the Queen Charlotte Islands and North Coast British Columbia (Prince Rupert District) regions, 2) both the long-term mean and range of spawn timing in the Central Coast and Johnstone Strait regions have remained steady, 3) the Strait of Georgia Region has experienced "a striking contraction of the range of spawning times, mainly from the loss of early spawning fish.", and 4) the west coast of Vancouver Island Region has also experienced a loss of early spawning herring, but the mean spawn day of the year has remained steady. Overall, Hay and McCarter (1999a) noted that the duration of Pacific herring spawn timing in British Columbia is becoming shorter and, in most areas Pacific herring are starting to spawn later and completing spawning earlier than in the past. Hay and McCarter (1999a) postulated that factors such as fisheries or climate change may account for some of the observed temporal changes in spawn timing; however, declining survey efforts, particularly in the non-assessment Sections, could also be a factor in the observed changes (Hay and McCarter 1999a).
Hay (1985) found a statistically significant inverse correlation between the mean temperatures for March and the mean spawning date (as Julian calendar day of the year) for Pacific herring in the Strait of Georgia between 1951 and 1982. Thus, within the limited geographical area of the Strait of Georgia, these analyses indicated that spawning occurs earlier in high temperature years and later in low temperature years (Hay 1985).
Alaska—In Southeast Alaska, six major Pacific herring stocks are recognized on the basis of their particular wintering grounds: 1) Ketchikan, 2) Dear Island-Etolin Island (near Wrangell, Alaska, 3) Craig-Hydaburg, 4) Auke Bay, 5) Sitka (Carlson 1980), and 6) Tenakee Inlet (east side of Chichagoff Island) (Carlile, unpubl. data). Carlile (unpubl. data) indicated that the Tenakee Inlet stock has recently been recognized as a major herring stock in Southeast Alaska. In the year 2000, the Tenakee Inlet stock had the second largest spawn abundance, behind the Sitka stock (Carlile, unpubl. data). Rounsefell (1930), Skud (1960), Blankenbeckler (1978), and Blankenbeckler and Larson (1982, 1985) provided information on the Pacific herring spawning localities and timing in Southeast Alaska (Table 3). Spawning localities identified by Rounsefell (1930) and Skud (1960) are listed in Figure 13a and their general locations are illustrated in Figure 22. Skud (1960) stated that although many Pacific herring spawning localities are utilized year after year, "in others there is a definite change in location of spawning beaches from year to year." Skud (1960) failed to detect spawning at 37 of the 57 localities listed by Rounsefell (1930) as Pacific herring spawning locales (Fig. 13a). Skud (1960) suggested three possible explanations for this discrepancy: 1) spawning locales may have changed between 1930 and the mid-1950s, 2) surveys may have missed spawning events, or 3) information supplied to Rounsefell (1930) may have been incorrect.
Between Southeast Alaska and Prince William Sound, few suitable Pacific herring spawning grounds are available and the continental shelf is of limited extent (Burkey 1986). Pacific herring spawn in south-central Alaska in Yakutat, Kenai Peninsula (Burkey 1986), Prince William Sound, Cook Inlet, and in the Kodiak-Afognak islands vicinity (Rounsefell 1930, Kruse 2000) (Fig. 23).
Pacific herring spawn on the western coast of Alaska in the eastern Bering Sea on the north side of the Alaska Peninsula, in Bristol Bay in the Togiak District, near Nunivak and Nelson islands, at Cape Romanzof, in Norton Sound, in Port Clarence and in Kotzebue Sound (Rounsefell 1930, Barton and Wespestad 1980, Haegele and Schweigert 1985, Kruse 2000) (Fig. 23).
Pacific herring spawning in the Bering Sea and northward is associated with climatological conditions, particularly ice-break-up (Barton and Wespestad 1980). Thus spawning in the eastern Bering Sea commences along the northern coast of the Alaska Peninsula in April-May and occurs progressively later to the north (Barton and Wespestad 1980, Kruse 2000), not occurring until early August in some years at Kotzebue ( Figs. 13a and 23). Pacific herring spawning does not begin until after break-up, in June to July, in the Beaufort Sea (Tanasichuk et al. 1993).
In the context of delineation of Pacific herring population structure, key questions that tagging studies address are: 1) To what degree do the same fish return to spawn on the same grounds year after year? and 2) How much interchange (gene flow) occurs between spawning populations? Unfortunately, adult tagging studies cannot help answer another important question: do adult Pacific herring return to the same spawning grounds where they were hatched?
Puget Sound—Taylor (1973) reviewed the results of tagging of over 70,000 Pacific herring in Puget Sound from 1953-59. Tagging apparently occurred both during summer fisheries in the San Juan Islands, Holmes Harbor, and Hood Canal, and during winter fisheries (presumably on pre-spawning populations) in Holmes Harbor, Hood Canal, Bellingham Bay, and Boundary Bay (Taylor 1973). Taylor (1973) addressed only the 52 tag recoveries that occurred in British Columbia, which were mainly from fisheries off the lower east coast of Vancouver Island. Other recoveries included one Holmes Harbor and four Hood Canal herring, tagged in summer fisheries, recovered at Swiftsure Bank off the southwestern tip of Vancouver Island, and one Bellingham Bay and five Boundary Bay herring, tagged in winter fisheries, recovered off the west coast of Vancouver Island (Taylor 1973). Taylor stated that:
These recoveries suggest that movement of herring from the Puget Sound, the San Juan Islands, and the Boundary Bay-Bellingham Bay regions into British Columbia is mostly from the adjacent areas and is probably no greater than between adjacent stocks of similar size in B.C. The recoveries on Swiftsure Bank and on the west coast of Vancouver Island suggest that the American stocks tagged perhaps move to offshore summer feeding grounds in the same way as the Canadian Strait of Georgia stocks.
O’Toole (2000) also reviewed results of the same Puget Sound tagging experiments conducted in the 1950s on Pacific herring from the Holmes Harbor, Port Orchard-Port Madison, and Quilcene Bay WDFW stocks. O’Toole (2000) emphasized that since none of the tag recoveries occurred during the spawning season, they provide little evidence toward resolving the question of spawning site fidelity for Pacific herring. Two Holmes Harbor tags were recovered in December from southern Johnstone Strait, and another Holmes Harbor herring was recovered in Holmes Harbor five years after tagging (O’Toole 2000). O’Toole (2000) reported on the recovery of four tagged Quilcene Bay herring: one was recovered in December from southern Johnstone Strait, two others were recovered at Swiftsure Bank in July, and one was recovered off Victoria, B.C. in September. Two tagged herring of the Port Orchard-Port Madison stock were recovered in Puget Sound: one in the Tacoma Narrows, three months after tagging, and another four-and-a-half years after tagging in the Waldron Island fishery (O’Toole 2000). O’Toole (2000) reported that both Williams (1959) and Buchanan (1986) felt that these tagging results indicate that Quilcene Bay (Hood Canal) herring migrate out of Puget Sound to summer feeding areas off southwest Vancouver Island.
British Columbia—Hay et al. (1999) stated that over 1.5 million tagged Pacific herring were released and over 42,000 of these were recovered in British Columbia between 1936 and 1991. Pacific herring tagging programs in British Columbia utilized small (19 mm long, 4 mm wide, 1.6 mm thick) metallic "belly tags" inserted into the herring body cavity from 1936-67, and plastic "anchor tags" inserted into the dorsal musculature from 1979-91 (Daniel et al. 1999, Hay et al. 1999). Both Daniel et al. (1999) and Hay et al. (1999) provided extensive bibliographies of technical publications that document these Pacific herring tagging programs. Recovery of belly tags occurred mainly in fish plants processing herring from the reduction fishery. According to Hay et al. (1999), belly tags were detected using magnetic detectors for the most part, and since these tags weren’t retrieved until the end of the fishing season, the date of recovery is known only to the nearest year. According to Daniel et al. (1999) and Hay et al. (1999), locations of many belly tag recoveries were also not exact and were often reported as being recovered from within a large geographic area at the level of one of the DFO Pacific herring Regions. Since anchor tags are readily visible on the external surface of the fish, the recent anchor tag recovery data, on the other hand, was usually reported by the day, year, and precise location of recovery that corresponds to a roe fishery location (Daniel et al. 1999, Hay et al. 1999). Figures 24, 25, and 26 (modified from figures available at the DFO Herring Tag Home Page at (http://www.sci.pac.dfo_mpo.gc.ca/herspawn/hertags/default.htm) (DFO 2000a) illustrate the geographic origin and recovery locations (within the six major Regions) of all Pacific herring tagged in one location and recovered in another location in British Columbia after at least one year at large.
As previously stated, DFO recognizes six Pacific herring management Regions in British Columbia: 1) Queen Charlotte Islands, 2) the North Coast British Columbia (Prince Rupert District), 3) the Central Coast, 4) Johnstone Strait, 5) the Strait of Georgia, and 6) the west coast of Vancouver Island. Each of these Regions is further divided into Statistical Areas, which are further divided into Sections (= Subareas), each of which is named and numbered (Hay and McCarter 2000). The numbered Statistical Areas can be identified by the first two digits of the Section number. The boundaries of each of the 108 Pacific herring Sections are illustrated by Region in Figure 20. In general, each Section contains several spawning beaches or "locations."
Hourston (1982) analyzed Pacific herring belly tag recovery data for British Columbia for the period 1937-67, at the level of the six management Regions and the then current herring-roe fishery "management units," and determined that 77-94% and 54-84% of recovered herring had "homed" to the Region and to the "management unit," respectively, in which they were tagged. Many authors (Schweigert 1991, O’Connell et al. 1998b, O’Toole 2000) have cited Hourston (1982) as evidence that Pacific herring return at high rates to spawning grounds that they spawned on in previous years. However, Hay et al. (1999) have reanalyzed these belly-tag data and included new analyzes of the anchor tag data that show that estimations of spawning site fidelity in Pacific herring in British Columbia at large for more than one year are highly dependant on geographical scale. Although, these analyses corroborate Hourston’s (1982) conclusion that Pacific herring have high homing or fidelity rates (80-100%) at the large geographic scale of the herring management Regions (Hay et al. 1999), at the smaller geographic scales of Statistical Area, Section, and spawning location the mean fidelity rates are much lower; 50-60% for Statistical Areas, 17-24% for Sections, and 1-2% for specific spawning locations (Fig. 27).
Hay et al. (1999) determined that the best fidelity or homing rates would be obtained by restricting the herring tag recovery database to only those herring released during the spawning months of February to April and to those recovered one or more years later during the same three months. Hay et al. (1999) stated that these restrictions essentially eliminate all of the belly tag data from the analyses since exact recovery dates are unknown for the belly tags. When this was done, the total number of informative tag returns dropped to 395 (321 were recovered after one year, 60 after two years, 8 after three years, 5 after four years, and 1 after five years) (Hay et al. 1999). At the level of the six management Regions, the mean fidelity rate after one year at large was 78%, and 82% after two years at large (Hay et al. 1999). Mean estimates of fidelity rate after one year at large at the level of Statistical Area were 56.8% (Hay et al. 1999) and 24.2% at the Section level (Hay et al. 1999). Among the highest fidelity rate estimates identified for Statistical Areas (SA) by Hay et al. (1999) were SA 02 in southeast Queen Charlotte Islands (72% after one year), SA 05 in the North Coast British Columbia Region (72% after one year), SA 07 in the Central Coast Region (82% after one year), and SA 23 in Barkley Sound (84% after one year). Hay et al. (1999) stated that, in general, fidelity rate estimates by Section were lower; however, fidelity rates after one year at large were 60% (n=1) for Section 042 (Chatham Sound), 72% (n=12) for Section 052 (Kitkatla Inlet), 69% (n=13) for Section 072 (Spiller Channel), 67% (n=12) for Section 142 (Lambert Channel), and 75% (n=38) for Section 232 (Macoah Passage) (Fig. 28). Hay et al. (1999) pointed out that because these tagging and recovery data span a three month period, "it is possible that many of these tag return data that show high fidelity to specific Sections, represent fish that were tagged and released before they reached their exact spawning destination or recovered after they had previously spawned elsewhere, perhaps in a different Section." Hay et al. (1999) suggested that fidelity rate of Pacific herring in British Columbia to a "Location" is only slightly higher than 0%. Hay et al. (1999) defined the approximate size of a "Location" as representing about 15 km of coastline per location.
Hay et al. (1999) also suggested that there is potential for episodic changes in rates of fidelity for Pacific herring, and present tagging data that indicates herring may tend to move more in some years than in others. Although Hay et al. (1999) stated that "tags released in each of the 6 Regions have been recovered in each of the other Regions," most British Columbia herring, except those in the Strait of Georgia and in mid-Hecate Strait, may not have a regular migration out of the Region. Conversely, most British Columbia Pacific herring seem to move extensively within their Region (Hay et al. 1999). Hay et al. (1999) postulated that:
‘Migratory’ herring, regardless of whether they move among Regions, may not show the same ‘fidelity’ to spawning areas as is shown by non-migratory herring. In the case of non-migratory herring, the return to the same spawning areas may not reflect active ‘homing’ as much as seeking the best spawning area, within their home range.
It should be noted that the above discussion concerns tagging studies on adult Pacific herring and as such provides little information in regards to homing of Pacific herring to their natal spawning grounds. Hourston (1959) described the results of tagging experiments on juvenile Pacific herring from Barkley Sound on the west coast of Vancouver Island. Hourston (1959) stated that, based on 50 tag returns, 52% of juvenile herring homed to the same sub-district (lower west coast of Vancouver Island) after two years at large and 64% homed after three years at large. This compared to homing rates of adult Pacific herring in the same sub-district after two and three years at large of 82% and 81%, respectively. Hourston (1959) postulated that:
As the young herring mature, they join adult schools on their spawning migration. Presumably, once the newly-maturing fish have joined an adult school, they stay with it. The adult schools, although intermixed, would have had the experience of at least one spawning migration and may thus be better equipped to find their way back to a certain beach.
Schweigert and Schwarz (1993) and Schwarz et al. (1993) modeled Pacific herring migration rates using tag-recovery data and concluded that "although rates of migration between stocks [on the Region level] are quite small and probably insignificant for all practical fishery management purposes," they are sufficient to homogenize genetic differentiation between "geographically distinct stocks of herring." In particular, Schweigert and Schwarz (1993) stated that their results suggested Pacific herring stocks in the North Coast British Columbia (Prince Rupert District) and Central Coast regions "are effectively discrete [based on tagging data] and should be managed as separate production units."
Southeast Alaska—The goal of tagging experiments of Pacific herring on spawning grounds in Southeast Alaska has been, for the most part, to determine the degree of intermingling of stocks in the fishery and thus provides little new information relevant to the question of spawning site fidelity (Rounsefell and Dahlgren 1933, Dahlgren 1936, Skud 1963, Carlson 1977).
Taylor (1964) recognized two types of Pacific herring in British Columbia, based on their migration patterns: 1) large migratory populations that migrate to the open ocean to feed, and 2) minor local populations that are found towards the heads of inlets or as resident populations that remain in inshore regions throughout the year. According to Taylor (1964), separation of these two types of Pacific herring is based on: 1) population abundance (migratory populations greatly exceed the local populations in abundance), 2) seasonal migration (minor local stocks stay inshore throughout the year, while the large migratory stocks feed offshore during the summer and return to inshore waters in the fall and winter), 3) growth and age composition (minor local populations have a slower growth rate), 4) location of spawning grounds (small local stocks tend to spawn at the heads of inlets and large migratory stocks spawn in more exposed regions such as near the mouths of inlets), and 5) homogeneity of individual stocks (minor stocks are complex assemblages of small individual runs, whereas various runs of the large migratory stocks are similar to one another). Similarly, Hay (1985) stated that in British Columbia, particularly in the Strait of Georgia, there appear to be a group of non-migratory herring that do not go to offshore summer feeding grounds, but remain in inside waters throughout the summer.
Penttila (1986) also postulated that some proportion of adult herring remain in Puget Sound throughout the summer while others migrate to offshore feeding grounds. A similar situation obtains in Southeast Alaska. Carlson (1980) reported that the major Sitka and Craig stocks of Pacific herring in Southeast Alaska make extensive summer feeding migrations, whereas the Pacific herring that spawn in the vicinity of Auke Bay and Ketchikan do not migrate to intermingle with the larger stocks.
Trumble (1983b) stated that many juvenile herring in Puget Sound overwinter in southern and central Puget Sound and migrate to Pacific Ocean feeding grounds in March to July, not returning until their first year of maturity. Adult herring were thought to migrate, on an annual, basis between summer feeding grounds (primarily off the Washington and British Columbia coasts) and Puget Sound spawning grounds (Trumble 1983b).
Hay and McCarter (1997b) surveyed Pacific herring larval distributions in several Sections in the Queen Charlotte Islands, North Coast British Columbia (Prince Rupert District), and Strait of Georgia regions in British Columbia and determined that "each major stock [Region] had a discrete larval distribution with continuous larval distributions within stock boundaries." In other words, extensive mixing of larvae occurred within the Regions investigated, but larval mixture was not detected between Regions (Hay and McCarter 1997b). According to Hay and McCarter (1997b), herring stock structure in British Columbia may be established at relatively early life-history stages independent of the exact spawning location, which can vary between years. Iles and Sinclair (1982) also identified larval distributions and larval retention areas as being of prime importance in defining Atlantic herring stock structure.
Mackenzie (1987) provided a recent review of the relationship of Pacific and Atlantic herring to their parasites, including the use of parasites as biological tags in stock structure analyses. According to Mackenzie (1987), this use of "biological tags" has been used for herring (mostly in the case of Atlantic herring) more than for any other marine fish.
Katz (1942) stated that Pacific herring collected in Willapa Bay in 1937 were heavily infested with nematodes, but that similar infestations were not present in Pacific herring collected in Puget Sound in 1936-37. Conversely, Trumbull (1980) stated that "Infestation of the roundworm parasite Anasakis [sic] occurs much more heavily for the sac-roe herring [Cherry Point stock] than for herring elsewhere in Puget Sound." Similarly, O’Toole (2000) stated that "nonquantitative observations indicate that body cavities of Cherry Point herring are normally full of the roundworm Anasakis [sic], which is uncommonly noted in adult herring from other areas in Puget Sound." According to Mackenzie (1987), the definitive host for adult Anisakis simplex are cetaceans such as whales and porpoises, and the first intermediate host(s) are euphausiids. Pacific herring become infected with Anisakis larvae when they ingest infected euphausiids.
Bishop and Margolis (1955) studied the level of infestation of Pacific herring with larvae of the nematode parasite Anisakis sp. in British Columbia. The mean level of infestation with Anisakis was found to increase with the age of the herring host; age-1 Pacific herring were free of the parasite. These differences are likely due to accumulation of the parasite in the diet from year to year (Bishop and Margolis 1955). The incidence of infection with Anisakis sp. was found to be between 90% and 100% in Pacific herring populations from the Queen Charlotte Islands and the west coast of Vancouver Island, and between 80% and 90% in herring from the Strait of Georgia (Bishop and Margolis 1955).
Arthur and Arai (1980) examined parasites of Pacific herring from seven spawning locations as an indicator of geographical origin: 1) Lisianski Strait in Southeast Alaska, 2) Skincuttle Inlet in the Queen Charlotte Islands, 3) Lockhart Bay on the central British Columbia coast, 4) Barkley Sound on the west coast of Vancouver Island, 5) Nanoose Bay, 6) Northwest Bay in the Strait of Georgia, and 7) Port Gamble in Puget Sound. Although Arthur and Arai (1980) stated that "reliable separation of adjacent stocks of spawning herring could not be accomplished," they identified three parasites that served to separate the Port Gamble population from the others: 1) Thynnascaris adunca (=Hysterothylacium aduncum) larvae (Nematoda), 2) bucephalid metacercariae, and 3) Anisakis simplex larvae. Arthur and Arai (1980) stated that "separation of Port Gamble herring from other stocks was due primarily to high counts of Thynnascaris adunca [= Hysterothylacium aduncum] larvae in this collection."
Moser (1991) and Moser and Hsieh (1992) determined that differences in the degree of infestation with Lacistorhynchus dollfusi (Cestoda) and the nematodes Anisakis simplex, Contracaecum sp., and Hysterothylacium indicated that Pacific herring that spawn in Tomales Bay and San Francisco Bay were separate stocks, and that they remain separate when at sea. Moser and Hsieh (1992) suggested that given the distribution of various definitive hosts of these parasites, Tomales Bay herring appear to spend more time offshore where they are infected with A. simplex, whereas San Francisco Bay herring spend more time inshore where they are more likely to be infected with L. dollfusi and Contracaecum sp. In all the above studies, differences between Pacific herring populations were determined on the basis of statistically significant differences in mean parasite prevalence or mean intensity of infection.
Differences in size-at-age and age structure, which should be reflections of differing growth and mortality rates, of herring populations may result from: 1) differences in migration patterns and consequent differences in food production and availability, 2) temperature differences, 3) density-dependant factors, 4) differences in the physical environment, and/or 5) differences in genetically determined growth characteristics. Since the degree to which growth rate is under genetic control in Pacific herring is unknown, it is difficult to determine whether differences in growth between populations are due to genetic or environmental differences.
California—Spratt (1981) found significant differences in the rate of growth of Pacific herring sampled in Tomales and San Francisco Bays, and suggested that "This difference may be evidence that the herring populations in Tomales bays and San Francisco Bay are distinct."
Puget Sound—Trumble (1980), Gonyea and Trumble (1983), and O’Toole (2000) described statistically significant differences in mean growth rate and length-at-age data for three populations of Pacific herring sampled in the mid-1970s in Puget Sound: 1) Case Inlet/Squaxin Pass, 2) Hale Passage-Carr Inlet, and 3) southern Strait of Georgia purse seine catch (Cherry Point stock). Trumbull (1980) found that the Strait of Georgia (Cherry Point) herring were consistently longer at age than the other populations and continued to grow later in life. Herring from Case Inlet/Squaxin Pass grew rapidly to age-3 and then growth slowed, culminating in a smaller size-at-age than was apparent for herring from the Strait of Georgia (Cherry Point) and Hale Passage-Carr Inlet stocks (Trumbull 1980). Herring from Hale Passage-Carr Inlet showed intermediate growth rate and size-at-age, and continued to grow as the fish aged (Trumbull 1980). Lassuy (1989) suggested these differences "may result because the Strait of Georgia stocks are migratory while the Case Inlet stocks are resident."
Buchanan (1985a) reported on the identification of a minor group of Pacific herring spawning in Fidalgo Bay-Padilla Bay that were differentiated from other stocks on the basis of their small size-at-age, scale pattern, and advanced maturity stage.
British Columbia—Tester (1937b) surveyed length-at-age and age composition of Pacific herring in selected commercial fisheries throughout British Columbia in the late-1920s to early-1930s and concluded that:
Due to the tendency for the Pacific herring to form local populations, a considerable diversity in age composition is present among the various runs. In localities separated by but a few miles different year classes may form predominating groups.
Ware (1985) compared growth rates and size-at-age for Pacific herring in five of the six DFO management Regions in British Columbia and determined that herring from the Queen Charlotte Islands and the North Coast British Columbia appear to grow at higher rates after age-4 than do the Central Coast and Johnstone Strait groups. Pacific herring from the west coast of Vancouver Island appeared to grow at intermediate rates (Ware 1985).
Schweigert (1991) applied multivariate statistical analyses to three years of size and age structure data for 26 spawning populations of Pacific herring in British Columbia. Schweigert (1991) stated that these analyzes "indicated stock separation on a smaller spatial scale" than is currently recognized. Schweigert (1991) found evidence that: 1) three distinct stocks occur in the Queen Charlotte Islands, 2) the North Coast British Columbia Region (Prince Rupert District) consists of a single stock, 3) Johnstone Strait and Strait of Georgia stocks are distinct, 4) Jervis Inlet is a distinct stock from the rest of the Strait of Georgia, 5) four spawning areas (Lambert Channel, Powell River, Nanoose Bay, and Deepwater Bay) in the Strait of Georgia represent a single stock, and 6) three separate stocks exist on the west coast of Vancouver Island (Barkley-Clayoquot sounds, Esperanza Inlet-Nootka Sound, and Quatsino Sound). Ware (1985) postulated that some of the resident stocks grow more slowly than migratory stocks due to the poorer production of food in the nearshore environment.
Alaska—Leon (1993) detected statistically significant differences in length-at-age and/or growth rates between three stocks of Pacific herring in Southeast Alaska, separated from one another by a minimum of 160 miles: 1) Sitka Sound, 2) Seymour Canal, and 3) Kah Shakes-Boca de Quadra. Comparison between two spawning sites separated by only 15-20 miles, Annette Island and Boca de Quadra, also indicated area-specific differences in growth characteristics (Leon 1993). Burkey (1986) compared the mean length-at-age of fishery caught samples of Pacific herring from Afognak Island, southern Alaska Peninsula, Cook Inlet, and Prince William Sound and found Prince William Sound herring were consistently smaller at age than any of the other three populations. Similarly, Burkey (1986) found the age composition of Pacific herring from Prince William Sound to be significantly different from the other three areas. However, similar comparisons of length-at-age and age composition between Pacific herring from Prince William Sound and Southeast Alaska did not show significant differences (Burkey 1986). Rowell (1980) stated that three distinct stocks of Pacific herring could be distinguished in the eastern Bering Sea on the basis of differences in scale growth patterns: 1) Togiak, 2) Port Clarence A, and 3) Port Clarence B-Cape Denbigh-Cape Romanzof (McBride and Whitmore 1981).
In general, age at maturity varies with latitude; beginning at age-2 in California and at age-4 or age-5 in the eastern Bering Sea (Hay 1985). Rabin and Barnhart (1986) reported that Pacific herring in Humboldt Bay, California recruit into this spawning population at age-2. Similarly, Spratt (1981) found that Pacific herring recruit into the spawning populations in Tomales and San Francisco bays beginning at age-2 and are fully recruited by age-3. Hay and McCarter (1999b) stated that sexual maturity occurs at age-3 in all major Pacific herring assessment areas in British Columbia, based on ovarian histology and the gonosomatic index. Barton and Wespestad (1980) stated that maturation begins at age-4 or age-5 in Alaskan stocks in the eastern Bering Sea.
The relationship of fecundity to length of Pacific herring has been studied from Peter the Great Bay off the Asian mainland (Ambroz 1931), Seal Rock (in Hood Canal), Washington (Katz 1942, 1948), British Columbia (Hart and Tester 1934, Nagasaki 1958, Hourston et al. 1981, Hay 1985), Tomales Bay, California (Hardwick 1973), Humboldt Bay, California (Rabin and Barnhart 1977), Prince William Sound, Alaska (Paulson and Smith 1977), and the Yellow Sea (Qisheng 1980) (Table 3 and Fig. 28). Over large geographic distances, fecundity of Pacific herring at a particular length has been found to decrease with an increase in latitude (Paulson and Smith 1977, Hay 1985, Lassuy 1989). Paulson and Smith (1977) suggested that the apparent decline in fecundity with increasing latitude is "offset by an increase in mean length of reproductively active females" with increasing latitude. Hay (1985) expanded on the data set of Paulson and Smith (1977), through inclusion of additional data from California and British Columbia herring populations, and observed a similar decreasing trend in fecundity with an increase in latitude (Table 4 and Fig. 28).
On the other hand, Nagasaki (1958) studied fecundity of Pacific herring in British Columbia and found the opposite relationship to latitude than that observed by Paulson and Smith (1977) and Hay (1985). Nagasaki (1958) stated that "the fecundity of herring of the same body length decreases from north to south and in the northern part of the province from west to east." In addition, Nagasaki (1958) found statistically significant differences in mean fecundity of Pacific herring of the same age and length between northern and southern population groups in British Columbia. Nagasaki (1958) stated that fecundity was significantly higher in northern than in southern British Columbia. Hourston et al. (1981) listed several possible reasons why the fecundity estimates of Nagasaki (1958) may be questionable, including: 1) three of the ten localities sampled by Nagasaki (1958) contain few herring, 2) only three or four of the ten runs sampled by Nagasaki (1958) were large enough to support fisheries, 3) most of the major herring management units in British Columbia were not sampled by Nagasaki (1958), and 4) all of Nagasaki’s (1958) samples came from a single spawning year.
Katz (1948) studied fecundity of Pacific herring from Seal Rock in Hood Canal, Washington and stated that "the Seal Rock herring produce more eggs than the British Columbia herring of the same length, and that the Siberian herring from Peter the Great Bay are less efficient egg producers than the British Columbia herring of the same size." However, Katz (1948) also noted that the Pacific herring in British Columbia studied by Hart and Tester (1934) and in Peter the Great Bay studied by Ambroz (1931) "ultimately grow much larger and produce far more ova" than Pacific herring from Seal Rock.
Ware (1985) found Pacific herring fecundity to be directly proportional to female body weight in the six DFO management Regions in British Columbia. Tanasichuk and Ware (1987) examined Pacific herring fecundity at a grand mean weight of 126 g for seven British Columbian populations in five different years and found significant differences in fecundity between years, but not between locations. Fecundity in 1983, an El Niño year, were 12.7% higher than in the other four years (Tanasichuk and Ware 1987). Tanasichuk and Ware (1987) also determined that sea-temperature 60-90 days prior to spawning "best accounted for variations in weight-specific fecundity." In comparisons between Pacific herring in the Strait of Georgia and the Beaufort Sea, Tanasichuk et al. (1993) found weight-specific fecundities to be 1.5 times greater in the Strait of Georgia fish.
Hay (1985) emphasized the utility of using the number of eggs per gram of adult female body weight as measure of relative fecundity in comparing populations (Table 4). Hay (1985) stated that this relative measure of fecundity "is more similar among size groups than is total fecundity," since the total number of eggs increases exponentially with female length and egg size also increases with adult female body size. Within British Columbia, the mean number of eggs per gram of female body weight for ripe Pacific herring, as presented in (Hay 1985, his Table 2), varied from a low of 184 on the North Coast British Columbia to a high of 224 in the Strait of Georgia. Hay (1985) suggested that a good estimate for British Columbia in general is 200 eggs/g of female body weight. Tanasichuk and Ware (1987) found the number eggs per gram of female body weight to range between 166 and 233 for seven sampling locations in five different years in British Columbia. The mean number of eggs/g of female body weight for Pacific herring in Humboldt Bay, California was estimated by Rabin and Barnhart (1977) at 220 ± 35. In Tomales Bay, California, Hardwick (1973) estimated relative fecundity of Pacific herring to be 227 ± 50 eggs/g female body weight.
Numerous researchers have attempted to separate "races" or stocks of Pacific herring on the basis of differences in body proportions or meristic characters. These differences may be due either to environmental (phenotypic) or hereditary (genotypic) factors, and it is extremely difficult to determine the underlying causes of these differences. In addition, morphometric and meristic differences between groups of fish are not normally apparent in individual fish but only in the mean value of a large number of individuals.
Thompson (1917) compared the head length; distance from the snout to insertion of the dorsal, anal, and ventral fins; and the mean number of vertebrae, gill-rakers, anal fin rays, and dorsal fin rays in Pacific herring from various locations in British Columbia and from San Francisco Bay, California. Thompson (1917) found differences between British Columbia and California populations of Pacific herring in mean vertebral and gill raker counts, but not between populations within British Columbia. Results from analyses of other meristic and morphometric characters were inconclusive (Thompson 1917). Hubbs (1925) extended the work of Thompson (1917) and studied differences in the number of dorsal fin rays, anal fin rays, and vertebrae in Pacific herring from California and British Columbia and found that the mean number of vertebrae increased from south to north. Subsequently, Rounsefell (1929, 1930) examined differences in mean counts of vertebrae, dorsal fin rays, and anal fin rays, and head length measurements between 32 samples of Pacific herring from San Diego, California to the Bering Sea. Evidence from head length and anal and dorsal rays were inconclusive, but Rounsefell (1929, 1930) found marked differences between locations in mean vertebral counts. Rounsefell (1929, 1930) also verified that the mean number of vertebrae in Pacific herring populations increases to the north and westward. Rounsefell (1929, 1930) found that between San Diego and the Shumagin Islands in the Bering Sea the mean vertebral count differed by nearly four vertebrae. When comparing adjacent localities, Rounsefell (1929, 1930) found significant differences in the mean vertebral count in some instances. On the other hand, Rounsefell (1929, 1930) also found significant differences in mean vertebral counts between samples collected in the same location but in different years. These differences most likely reflected variation between year-classes (Ahlstrom 1957). Rounsefell (1929, 1930) identified the following groupings that he considered distinct populations based on variation in vertebral counts: 1) California, 2) southern British Columbia, 3) Stephens Passage in Southeast Alaska, 4) Chatham Strait in Southeast Alaska, 5) Craig in Southeast Alaska, 6) Prince William Sound, 7) Cook Inlet-Shuyak Strait, 8) Shearwater Bay-Old Harbor in south-central Alaska, 9) Chignik in western Alaska, 10) Shumagin Islands, 11) Unalaska, and 12) Golovin Bay.
Rounsefell and Dahlgren (1935) examined herring stock structure in Southeast Alaska, mainly through comparison of the mean number of vertebrae in different year-classes between 32 localities. Individual year-classes were studied, since a high negative correlation was found between temperature during development and the mean vertebral number in different year-classes. Rounsefell and Dahlgren (1935) identified six populations in Southeast Alaska that they considered independent of one another, based on a combination of differences in vertebral counts, growth rates, and year-class strength: 1) Juneau-Icy Strait area, 2) Sitka-Cape Ommaney-Chatham Strait area, 3) Noyes Island-west coast of Prince of Wales Island, 4) inner areas of Southeast Alaska, 5) vicinity of Petersburg, and 6) Todd-Peril Strait.
Similarly, Tester (1937a) examined meristic and morphometric variation (vertebrae count, head length, length to dorsal fin insertion) in Pacific herring from 19 localities in British Columbia, and, as in earlier studies, found differences in mean vertebral counts to be most informative for stock discrimination. Tester (1937a) also confirmed that the mean vertebral count in Pacific herring populations increases with latitude. Tester (1937a) concluded that meristic characters could be used to separate British Columbia herring into the following discrete units or populations: 1) Point Grey, 2) Granite Bay, 3) Saltspring Island-Departure Bay-Nanoose Bay, 4) Barkley Sound-Sydney Inlet, 5) Nootka Sound-Kyuquot Sound, 6) Quatsino Sound, 7) Bella Bella, and 8) Butler Cove-Pearl Harbour and the area currently designated as N. Porcher Island. In a later paper, Tester (1938) stated that "variation in the number of vertebrae and certain other meristic characters in fishes is caused in part at least by variation in environmental conditions, notably water temperature." This statement was based in part on evidence that the mean count of vertebrae in successive year-classes of Pacific herring from Barkley Sound on the west coast of Vancouver Island, varies inversely with water temperature at the time of spawning and early development (Tester 1938). In a later paper, Tester (1949) found this relationship to hold in general for Pacific herring from the entire west coast of Vancouver Island. McHugh (1942) also found significant differences in vertebral counts of juvenile herring from the same year-class sampled at a number of localities within the Strait of Georgia.
Both Tester (1937a) and McHugh (1942) reported that within certain samples, the largest fish tended to have the highest vertebral counts. However, Tester (1949) later stated that analysis of data from the west coast of Vancouver Island between the years 1929-1941 revealed "no significant tendency for older fish of a year-class to have a higher mean [vertebral] count, as had been indicated by results previously published."
Tester (1949) concluded, on the basis of mean number of vertebrae in Pacific herring samples from the west coast of Vancouver Island, that Pacific herring represented "essentially discrete populations, between which mixing was generally limited, but with the reservation that mixing more extensive than "limited" might take place occasionally." However, when Pacific herring tag recovery data were taken into account for west coast of Vancouver Island, Tester (1949) stated that "the latter method demonstrated that mixture did take place and that it was considerably more extensive than would be anticipated by the term "limited"—so much so that for practical purposes the series of intergrading "units" [of Pacific herring determined to exist on west coast of Vancouver Island on the basis of mean vertebral counts] were considered to constitute one major population." In addition, Royce (1957) determined that Tester’s (1949) vertebral count data for west coast of Vancouver Island Pacific herring indicates that mixture between northernmost and southernmost areas "could be as high as 93 percent."
McHugh (1954) reviewed meristic and morphometric studies on Pacific herring and stated that "the number of vertebrae is capable of modification by temperature during early development, so that in any one locality the mean vertebral number may vary from year to year. This finding, however, has not invalidated the general conclusion that the number of vertebrae decreases from north to south."
Schweigert (1981) reanalyzed morphometric and meristic data published by Thompson (1917) for Pacific herring from San Francisco and Point Grey and Departure Bay in British Columbia using pattern recognition, a form of multivariate analysis. Morphometric characters included standard length; length from the snout to the insertion of the dorsal, pelvic and anal fins; head length; and occiput length. Meristic characters included numbers of vertebrae, gillrakers and anal and fin rays. Schweigert (1981) stated that the morphometric characters were more useful in separating the three groups of Pacific herring from one another than were meristic characters. Although these three groups of Pacific herring are known to be dissimilar, Schweigert (1981) stated that "the separation obtained between the two British Columbia stocks in this study based solely on morphometric considerations is sufficient to warrant examination of morphometric differences among British Columbia herring stocks on a larger scale."
Meng and Stocker (1984) applied discriminant function analysis to a set of twenty-seven morphometric and eight meristic characters taken from Pacific herring sampled in commercial fisheries at five localities in British Columbia. Pacific herring from the Strait of Georgia were detectably different from those in northern British Columbia, although exchange between these two geographical regions was indicated (Meng and Stocker 1984). Meng and Stocker (1984) recommended the use of meristic characters to separate Pacific herring populations on a broad geographic scale and the use of 12 identified "best" morphometric characters for finer scale separation of spawning populations.
Schweigert (1990) used univariate and multivariate statistical analyses to compare the ability of traditional morphometric and meristic data sets and the new method of truss measurements to differentiate Pacific herring stock structure in British Columbia. Schweigert (1990) found "no substantial differences in the ability of truss networks and traditional morphometric and meristic data to differentiate among groups of" Pacific herring. Analyses of the morphometric and meristic data did not reveal significant differences between Pacific herring from North Coast British Columbia, Central Coast, and the Strait of Georgia; however, each of these areas was significantly different from the west coast of Vancouver Island (Schweigert 1990). Analyses of the truss network data revealed a significant difference only between Pacific herring from North Coast British Columbia and the west coast of Vancouver Island.
Within Washington, Chapman et al. (1941) and Katz (1942) compared mean vertebral counts between Pacific herring collected at eleven locations in 1936 and 1937. These locations are: 1) Woolochet Bay (in south Puget Sound), 2) Poulsbo (Port Orchard-Port Madison stock), 3) Holmes Harbor, 4) Hales Pass (Samish Bay-Portage Bay stock), 5) Seal Rock (Quilcene Bay stock), 6) Birch Bay (Cherry Point stock), 7) Pt. Migley, 8) Willapa Bay, 9) East Sound (Interior San Juan Islands stock), 10) Gig Harbor, and 11) Steamboat Island (Squaxin Pass stock). Comparison of Pacific herring samples from Willapa Bay, Woolochet Bay, and Birch Bay revealed "good separation of the mean vertebral counts in the two- and three-year-old age classes, but in the four-, five-, and six-year-olds, the differences of the mean vertebral counts were slight" (Katz 1942). Katz (1942) stated that,
... the older age classes of the herring of Washington fail to show a significant difference in their mean vertebral counts and cannot, therefore, be designated as races if vertebral differences are to be used as a racial criterion. Whether this lack of racial distinctiveness in the older age classes is due to intermingling or to other causes cannot be determined with the scanty data on hand.
Katz (1942) also observed that annual variation in mean vertebral counts of Pacific herring was greater for the Woolochet Bay aggregation in southern Puget Sound than in the Birch Bay aggregation, whose spawning location was closer to the open ocean. Katz (1942) postulated that the greater variation in the Wollochet Bay vertebral counts "might be due to a greater temperature fluctuation of the southern waters of the Sound."
Several hypotheses, pertinent to the question of distinct population segments for Pacific herring, have been proposed in Europe and eastern North America to explain the apparent stock structure pattern of Atlantic herring. Iles and Sinclair (1982) and Smith and Jamieson (1986) proposed diverging hypotheses to explain phenetic and genetic data on Atlantic herring stock structure. McQuinn (1997) articulated the differences between these hypotheses and termed the Iles and Sinclair (1982) hypothesis "the discrete population concept" and the Smith and Jamieson (1986) hypothesis "the dynamic balance population concept." McQuinn (1997) then attempted to unify these opposing hypotheses under the metapopulation concept.
The discrete population concept of Iles and Sinclair (1982) states that "the number of herring stocks and the geographic location of their respective spawning sites are determined by the number, location, and extent of geographically stable larval retention areas." This concept depends upon the maintenance of reproductive isolation of spawning populations through both homing to natal spawning sites and natural selection against less fit hybrids that result from straying of individuals from other spawning populations (Iles and Sinclair 1982, McQuinn 1997). Smith and Jamieson (1986) and McQuinn (1997) suggest that the discrete population concept is inconsistent with several lines of evidence, including: 1) observed rates of straying of adult and juvenile tagged Pacific and Atlantic herring, 2) observations of Atlantic herring hatched in autumn (as indicated by their otolith pattern) spawning as adults in the spring, and vice versa, and 3) the lack of consistent genetic differentiation among regional herring populations.
In contrast, the dynamic balance population concept of Smith and Jamieson (1986) contends that herring populations expand and contract their range in response to external pressures and that local population structure is transient on evolutionary time scales (McQuinn 1997, Corten 1999). McQuinn (1997) suggested that the major weakness in this latter concept was "the lack of an explanation for the temporal persistence of populations in geographic space on an ecological time scale."
McQuinn (1997) stated that neither of the above concepts is consistent with all the data on herring stock structure, and that "contradictory evidence supporting both discreteness and mixing has prevented a clear definition of population structure." However, McQuinn (1997) proposed that "Atlantic herring population structure and dynamics are well described within the metapopulation concept" (McQuinn 1997). A metapopulation has been defined by Levins (1968) as:
a population of populations which were established by colonists, survive for a while, send out migrants, and eventually disappear. The persistence of a species in a region depends on the rate of colonization successfully balancing the local extinction rate.
Under this metapopulation concept of herring stock structure, McQuinn (1997) suggested that local herring populations may be perpetuated through a process he termed the "adopted-migrant hypothesis," where juvenile herring that associate with and synchronize their maturation with schools of adult herring will adopt the migration and homing patterns of the adults. Thus, local spawning populations are maintained by "repeat rather than natal homing to spawning areas, while local population persistence is ensured through the social transmission of migration patterns and spawning areas from adults to recruiting individuals" (McQuinn 1997). In McQuinn’s (1997) "adopted-migrant hypothesis," hydrographic forces on larvae and the effects of schooling of juveniles leads to the majority of individuals spawning in their native population. Thus, differences in the mean values of meristic and morphometric measurements that reflect environmental differences during development are maintained, although strays from other populations are adopted by local populations and gene flow is significant (McQuinn 1997). McQuinn (1997) stated that the "adopted-migrant hypothesis" is consistent with genetic studies on herring that have not observed temporally-persistent differences, since no genetic differences would be expected between sympatric herring populations with the hypothesized level of gene flow.
McQuinn (1997) emphasized that local populations should be considered the basic fisheries management unit and that their "conservation is essential for the preservation of spawning potential and for the viability of coastal fisheries," although when fisheries occur on a mixture of local populations, the metapopulation becomes the practical management unit. Although McQuinn’s (1997) metapopulation concept and "adopted-migrant hypothesis" were first formulated for Atlantic herring, the similar ecological characteristics of Pacific herring suggests these concepts are pertinent to stock structure questions for this species as well.
The BRT examined environmental, geologic, biogeographic, life-history, and genetic information in the process of identifying DPSs of Pacific herring. In particular, biogeography, ecological and habitat factors, and genetic population structure were found to be most informative for these species. The DPSs considered in this evaluation were:
A. A separate DPS for each of the five basins of Puget Sound which are: Hood Canal, Main Basin, Whidbey Basin, the Strait of Juan de Fuca/San Juan Islands, and South Sound. The BRT constructed this scenario based on such factors as differences in spawning distributions, behavior and timing, and possible larval transport restrictions brought about by basin physiography in these basins. This might lead to sufficient differentiation of spawning aggregations to meet the "distinct" criteria for definition of a DPS. The WDFW is presently conducting a herring genetics study that may shed some light on whether these basins define DPSs, or simply multiple stocks. DPS of this size would compare with the DFO section or subarea management unit.
B. A DPS for two regions within the Georgia Basin, which are: Puget Sound proper (that portion of Puget Sound south of Admiralty Inlet and east of Deception Pass), and in north Puget Sound including the Strait of Juan de Fuca/San Juan Islands up to the mouth of the Fraser River. In order to construct this DPS scenario, the BRT used evidence for processes that would lead to distinct populations at this scale. These include the limitation of larval transport through the natural barrier at Admiralty Island and Deception Pass, evidence that the Puget Sound proper spawning aggregations are resident fish and that straying rates inferred from Canadian tagging studies do not apply to these Puget Sound resident populations. Fidelity to spawning aggregations could be increased due to the resident nature of the populations. This could result in less mixing and straying of herring populations than is seen in the Canadian populations that were the subject of the tagging studies. The DFO defines this scale as a District or Regional scale management unit.
C. A DPS that encompasses Georgia Basin, extending from the southern end of Puget Sound to the northern end of the Strait of Georgia near Discovery Passage. This DPS extends from the north end of Vancouver Island to the south end of Puget Sound and includes the Strait of Juan de Fuca. This DPS would define the populations in Puget Sound at the DFO District-level management unit. District-level criteria used by DFO are derived by evidence from tagging studies, vertebral counts, assessment of larval distribution and transport, and hydrographical studies of the adjacent area. This type of information, however, is generally not available for the Puget Sound populations.
D. A single DPS that includes the populations in the area from Baja California to Southeast Alaska, with the northern boundary being the border of the zoogeographic zone near Dixon Entrance, or a line between Helm Bay and Lynn Canal, Alaska. At this scale, the DPS is defined by the genetics investigations of Grant and Utter (1984), and by the zoogeographic boundary of Ekman (1953), Hedgpeth (1957), and Briggs (1974). This DPS exceeds any management area defined by DFO for Canadian populations.
A majority of the BRT favored the Georgia Basin, which is option C, (Fig. 29) as the most likely DPS, with options B and D receiving considerably less support. No member of the BRT supported DPS option A. Members of the BRT utilized a variety of evidence to support their identification of a Georgia Basin DPS for Pacific herring. These included tagging studies in the Canadian portion of the Georgia Basin, vertebral counts, information on larval distribution and transport, as well as hydrographic studies conducted by the Department of Fisheries and Oceans Canada (DFO). Genetic studies by Grant and Utter (1984) were also utilized in concert with work by McQuinn (1997) that describes the metapopulation stock structure in herring. Based on this examination, the BRT identified a DPS for the Georgia Basin, which includes Puget Sound, and focused the risk analysis on this DPS.
4 J. Shaklee, Washington Department of Fish and Wildlife, 600
Capital Way N., Olympia, WA 98501. Pers. commun., July 2000.
5 P. Bentzen, Marine Molecular Biotechnology Laboratory,
University of Washington, 3707 Brooklyn Ave. NE, Suite 175, Seattle, WA 98105.
Pers. commun., July 2000.