Shallow Population Histories in Deep Evolutionary Lineages of Marine Fishes: Insights from Sardines and Anchovies and Lessons for Conservation

Shallow Population Histories in Deep Evolutionary Lineages of Marine Fishes: Insights From Sardines and Anchovies and Lessons for Conservation

W. S. Grant and B. W. Bowen

Most surveys of mitochondrial DNA (mtDNA) in marine ®shes reveal low levels of sequence divergence between haplotypes relative to the differentiation observed between sister taxa. It is unclear whether this pattern is due to rapid lineage sorting accelerated by sweepstakes recruitment, historical bottlenecks in population size, founder events, or natural selection, any of which could retard the accumulation of deep mtDNA lineages. Recent advances in paleoclimate research prompt a re- examination of oceanographic processes as a fundamental in¯uence on genetic diversity; evidence from ice cores and anaerobic marine sediments document strong regime shifts in the world's oceans in concert with periodic climatic changes. These changes in sea surface temperatures, current pathways, upwelling intensities, and retention eddies are likely harbingers of severe ¯uctuations in population size or regional extinctions. Sardines (Sardina, Sardinops) and anchovies (Engraulis) are used to assess the consequences of such oceanographic processes on marine ®sh intrageneric gene genealogies. Representatives of these two groups occur in temperate boundary currents on a global scale, and these regional populations are known to ¯uctuate markedly. Biogeographic and genetic data indicate that Sardinops has persisted for at least 20 million years, yet the mtDNA genealogy for this group coalesces in less than half a million years and points to a recent founding of populations around the rim of the Indian±Paci®c Ocean. Phy- logeographic analysis of Old World anchovies reveals a Pleistocene dispersal from the Paci®c to the Atlantic, almost certainly via southern Africa, followed by a very recent recolonization from Europe to southern Africa. These results demonstrate that regional populations of sardines and anchovies are subject to periodic extinctions and recolonizations. Such climate-associated dynamics may explain the low levels of nucleotide diversity and the shallow coalescence of mtDNA genealogies. If these ®ndings apply generally to marine ®shes, management strategies should incorporate the idea that even extremely abundant populations may be relatively fragile on ecological and evolutionary time scales. From the Conservation Biology Division, Northwest Fisheries Science Center, NOAA, Seattle, Washington, and the Department of Fisheries and Aquatic Sciences, A recurring debate in evolutionary biology roki 1997). Although several early hypoth- University of Florida, Gainesville, Florida. We thank A. is over the extent to which microevolution- eses about population regulation are now Bass, A. Clark, and A. Garcia for technical support, and ary processes operating within a species can R. Leslie and A. Payne (Sea Fisheries Research Insti- be extrapolated to explain macroevolution- discounted on the basis of field studies, tute, Cape Town, South Africa), T. Kobayashi (National ary differences among species . . . other hypotheses remain untested be- Research Institute of Fisheries Science, Yokohama, Ja- Avise et al. (1987, p. 489) pan), S. Jablanski (SUDEPPE, Rio de Janeiro, Brazil), cause of the lack of an appropriate tool. and J. Shaklee (CSIRO, Canberra, Australia) for gener- Recent advances in sampling technology ously providing samples for the various studies re- and satellite imagery show considerable viewed in this article. K. Bailey, J. Gold, S. Karl, T. To understand the dynamics of marine Streelman, F. Utter, and R. Waples provided insightful fish populations, researchers must identi- promise, demonstrating, for example, that comments on the manuscript. Genetic studies of sar- fy the conditions that regulate reproduc- in the California Current egg and larval dines and anchovies were supported by the U.S. Na- tional Science Foundation and by the Foundation for tion, population growth, and persistence. production is contingent on small upwell- Research Development, Pretoria, South Africa. Address On short (ecological) time scales, a vari- ing plumes along the coast (Lo et al. correspondence to Dr. Grant, Northwest Fisheries Sci- 1996). ence Center, 2725 Montlake Boulevard East, Seattle, WA ety of factors, including nutrient cycles, 98112-2097, or e-mail: [email protected]. This paper food-chain processes, spawning, preda- One emerging generalization from mo- was delivered at a symposium entitled ‘‘Conservation tion, recruitment, and climate have been lecular analyses is that marine fishes are and Genetics of Marine Organisms’’ sponsored by the American Genetics Association at the University of Vic- proposed as primary regulators of abun- often characterized by shallow population toria, Victoria, BC, Canada, June 7, 1997. dance (Butler 1991; Parrish and Mallicoate genetic architectures, even though genetic ᭧ 1998 The American Genetic Association 89:415–426 1995; Smith et al. 1992; Watanabe and Ku- divergence from sister taxa indicates sep-

415 arations of millions of years. In a review of mitochondrial DNA (mtDNA) diversity in widely distributed marine fishes, Shields and Gust (1995) noted a recurring pattern of a single or a few prevalent haplotypes with numerous rare haplotypes that were one or two mutations removed from the common haplotype. These starlike phylogenies characterize regional populations of haddock (Melanogrammus aeglefinus; Zwa- nenburg et al. 1992), Atlantic cod (Gadus morhua; Carr and Marshall 1991; Smith et al. 1989), cape hake (Merluccius capensis; Becker et al. 1988), deepwater hake (M. paradoxus; Becker et al. 1988), Atlantic herring (Clupea harengus; Kornfield and Bogdanowicz 1987), Pacific herring (C. pallasi; Schweigert and Withler 1990), red drum (Sciaenops ocellatus; Gold et al. 1993), black drum (Pogonias cromis; Gold et al. 1994), greater amberjack (Seriola dumerili; Richardson and Gold 1993), red snapper (Lutjanus compechanus; Camper et al. 1993), Spanish sardine (Sardinella aurita; Tringali and Wilson 1993), orange roughy (Hoplostethus atlanticus; Baker et al. 1995; Ovenden et al. 1989; Smolenski et al. 1993), Atlantic capelin (Mallotus villosus; Dodson et al. 1991), albacore tuna (Thunnus alalunga; Graves and Dizon 1989), and skipjack tuna (Katsuwonus pelamis; Graves et al. 1984). Shallow haplotype divergences atop long lineages are Figure 1. Geographical distributions of sardines (Sardina, Sardinops) and anchovies (Engraulis) with 13ЊC and also clearly illustrated in Figures 4 and 5 25ЊC isotherms (dashed lines). of Bermingham et al. (1997) for species of damselfish isolated about 3 million years ago by the formation of the Panama isth- boundary current systems, and because colonizations), founder events, dispersals, mus. Explanations for this widespread pat- regional populations of both groups show and divergence between isolated popula- tern include a large variance in reproduc- strong fluctuations in abundance that tions. These case histories are used to tive success that leads to the propagation have been attributed to high levels of ex- evaluate the hypotheses that have been of only a few haplotypes (Shields and Gust ploitation (Murphy 1966, 1967). For ex- forwarded to explain shallow gene gene- 1995), overharvesting (Camper et al. ample, the biomass of sardines (Sardinops alogies in other marine fishes. The forces 1993), the physical nature of the pelagic caeruleus) in the California Current peaked that attenuate mtDNA diversity may be a realm (Graves 1995), recent habitat reduc- at an estimated 3,600,000 metric tons key to understanding population regula- tions (Shulman and Bermingham 1995), (MT) in the 1930s (Murphy 1966) then de- tion in marine fishes. Hence these shallow population bottlenecks (Gold et al. 1994), clined during a period of intensive har- intraspecific phylogenies carry implica- or other ‘‘demographic events’’ (Dodson vests to about 5,000–6,000 MT in 1975 tions for microevolution and marine bio- et al. 1991). (Barnes et al. 1992; Wolf 1992). The biogeography as well as resource manage- This phenomenon is also apparent in mass of California anchovies (Engraulis ment in the face of climatic change and sardines (Sardina, Sardinops) and ancho- mordax) has also fluctuated from lows in high exploitation (Hansen et al. 1981; San- vies (Engraulis). Both groups are globally the 1950s to a high in the 1970s (Lo and ter et al. 1996). distributed in temperate zones and have Methot 1989). All of the regional popula- representative species or populations in tions of sardines and anchovies have sim- Long-Term Climatic Variability and most of the world’s temperate boundary ilar histories of declines and partial recov- Population Abundance Cycles current systems (Figure 1). These popu- eries which are attributed to harvests or lations are isolated by vast expanses of to climatic and oceanographic changes It is widely accepted that climatic changes open ocean or by warm tropical waters (Lluch-Belda et al. 1989). are capable of limiting abundances, but that restrict movement across the equa- Here we review genetic evidence from the impact of these changes on marine tor. Sardines and anchovies are a peren- allozyme and mtDNA datasets for sardines biodiversity has only recently been appre- nial concern to marine resource managers and anchovies that may bear imprints of ciated (Hayward 1997; Roemmich and because they represent the majority of the population collapses (bottlenecks), meta- McGowan 1995; Watson et al. 1996). Rapid clupeoid biomass in highly productive population dynamics (extinctions and re- changes in oceanic temperature over the

416 The Journal of Heredity 1998:89(5) last few tens of thousands of years, cor- can be calibrated with divergences be- responding to major climatic shifts, have tween fish populations separated by the been recorded in ice cores from Greenland rise of the Isthmus of Panama about 3 mil- (Dansgaard et al. 1993; GRIP Members lion years ago (Grant 1987; Keigwin 1978, 1993) and Antarctica (Jouzel et al. 1993; 1982; Vawter et al. 1980) and by dispersal Lorius et al. 1985). These regime shifts, through the Bering Strait, also about 3 mil- some of which have occurred on a time lions years ago, prior to the late Pliocene scale of only a few decades, drastically al- cooling of the Arctic Ocean (Grant 1984; tered major ocean circulation and temper- Grant and Sta˚hl 1988; Grant et al. 1984). ature patterns (Lehman and Kiegwin The resulting clock (D ϭ 1.0 about 19 mil- 1992). Recent decade-scale shifts in the lion years) yields an estimate of the diver- Kuroshio Current led, in part, to a rapid gence time between Sardina and Sardinops decline of Japanese sardines in the 1940s of about 19 million years BP (15–24 million (Kawasaki 1993; Kawasaki and Omori years). This time frame coincides closely 1995). Figure 2. Majority-rule bootstrap of neighbor-joining with the collision of the African plate trees representing phylogenetic relationships among On scales of centuries and millennia, taxa of sardines (Sardina, and Sardinops). Percentage against southern Europe (Steininger et al. abundances of sardines and anchovies in bootstrap (over loci for allozymes, over nucleotide 1985), a vicariant event which divided the sites for sequences) support indicated at nodes of the California Current have fluctuated trees. (a) Nei’s unbiased genetic distances based on 34 Tethys Sea into Atlantic and Indian-Pacific greatly. Based on scale deposition rates in protein-coding loci (Grant and Leslie 1996). (b) Se- components. An alternative scenario, pos- anaerobic sediments, Baumgartner et al. quence divergences based on a 220 bp sequence in the tulated by Okazaki et al. (1996), is that the cytochrome b gene of mitochondrial DNA (Grant et al., (1992) identified nine population declines unpublished data). initial split between ancestral sardine pop- and recoveries over the last 1,700 years. ulations occurred via the Isthmus of Pan- During this interval the estimated abun- ama. However, the magnitudes of the allo- dance of the California sardine fluctuated Sea in the Atlantic and from the western zyme genetic distance between Sardina from less than 1 million MT to at least 4 Mediterranean to the western margin of and Sardinops and sequence divergence million MT. These fluctuations predate the Black Sea (Figure 1a). Sardinops inhab- between cytochrome b sequences (Figure fishing activity along the California coast its five upwelling zones of the Indian–Pa- 2) contradict this recent separation. and thus are attributable entirely to natu- cific Ocean characterized by high levels of Although both allozyme and mtDNA ral biotic, climatic, and oceanic changes. primary and secondary productivity, in- data point to a deep evolutionary history The earth’s climate oscillates on several cluding southern Africa, Australia–New for Sardinops, divergence among present- time scales with various amplitudes. For Zealand, Chile–Peru, west-central North day populations reflects only a shallow example, the population fluctuations doc- America, and Japan. In the light of recent history reaching back less than half a mil- umented by Baumgartner et al. (1992) for molecular data, the taxonomy of these re- lion years (Bowen and Grant 1997; Grant sardines occurred during a period of rel- gional populations is uncertain, and we et al., in press; Grant and Leslie 1996). atively minor climatic shifts. Over the pre- will refer to them by their traditional spe- This shallow time frame and the low levels ceding 100,000 years, the amplitude of cies names: S. ocellatus (S. Africa), S. neo- of allozyme and morphological diversity global climatic oscillations, as recorded in pilchardus (Australia), S. sagax (Chile), S. indicate that present sardine populations Greenland ice cores, was much greater caeruleus (California), and S. melanostictus have expanded only recently around the (GRIP Members 1993). These North Atlan- (Japan). The geographic distributions of rim of the Indian– Pacific Ocean. At least tic changes corresponded to temperature the regional populations are generally two legacies of this colonization process and salinity shifts in many of the world’s bounded by the 13ЊC and 25ЊC isotherms, are apparent in the genetic data. One is a oceans (Broecker 1995). Imprints of tem- since temperatures above 27ЊC are lethal significant excess of low-frequency allo- perature fluctuations in ice cores reaching to adults and larvae (Parrish et al. 1989). zyme alleles over that expected with drift- back about 250,000 years also indicate Sardines are notably absent in the western mutation equilibrium in datasets for Indi- strong climatic variability (Dansgaard et Atlantic, even though the temperate zones an–Pacific sardines (Grant and Leslie al. 1993). On yet a deeper temporal scale, of the northwest and southwest Atlantic 1996), for California sardines (Hedgecock climates have greatly fluctuated during appear to be suitable for sardines, and et al. 1989), and for South African sardines the four major Pleistocene glaciations both regions host populations of ancho- (Grant 1985). Such an excess is usually at- reaching back 1.7 million years. These vies. tributed to the retention of new mutations changes have undoubtedly led to popula- The proposal that current Indian–Pacific during population growth or expansion tion crashes and expansions, and possibly sardine populations are ephemeral or (Watterson 1984). Another indication that to extinctions of some regional sardine have become established only recently is Indian–Pacific sardines have recently ex- and anchovy populations in the major based on the observation that present-day panded is a Poisson-like distribution of the boundary currents of the world. populations of Sardinops are shallow twigs number of nucleotide differences ob- at the termini of an ancient lineage ex- served in comparisons of cytochrome b tending back to the Miocene. In an analy- sequences (Grant et al., in press). This dis- Sardines (Sardina, Sardinops) sis of 34 allozyme loci, Grant and Leslie tribution is attributed in other species to Northwest Atlantic and Indian–Pacific sar- (1996) reported a Nei’s genetic distance of mutation-drift disequilibrium caused by dines are divided into two genera. Sardina D ϭ 1.04 (Ϯ0.24 SE) (Figure 2a) between explosive population growth (Rogers and in European waters consists of a single the Atlantic–Mediterranean (Sardina) and Harpending 1992). species, S. pilchardus, that extended in his- Indian–Pacific (Sardinops) forms. An ap- Phylogenetic relationships among re- torical times from West Africa to the North proximate time frame for this separation gional populations are not resolved with

Grant and Bowen • Shallow Genetic Architectures in Marine Fishes 417 Table 1. mtDNA haplotype and nucleotide diversities and allozyme diversities in sardines (Sardina, chovies belong to a single genus, Engrau- Sardinops) lis, but the level of morphological differ- Control regiona Cytochrome bb Allozymesc entiation between Old World and New Region h ␲ n h ␲ n H n World species indicates that a separate genus for Old World species may be war- Sardina ranted (Hubbs 1952; Whitehead 1973). Europe 0.36 0.002 5 0.024 26 The taxonomy of anchovies is further con- Sardinops fused by the inclusion of three tropical South Africa 1.00 0.02 15 0.62 0.004 15 0.036 46 species (genus Cetengraulis) within the Australia 1.00 0.02 15 0.62 0.004 15 0.045 50 Chile 1.00 0.03 18 0.76 0.006 15 0.037 30 morphologically based phylogenetic tree California 1.00 0.03 15 0.76 0.007 15 0.036 30 of Engraulis (Nelson 1984). Molecular data 0.010 149d Japan 0.96 0.01 18 0.67 0.005 14 0.022 50 reinforce the arguments for revision of anchovy taxonomy, but we will refer to re- a Bowen and Grant 1997. gional populations by traditional species b Grant et al., in press. names: E. encrasicolus (Europe), E. capen- c Grant and Leslie 1996. sis (southern Africa), E. australis (Austra- d Hedgecock et al. 1989. lia), E. japonicus (Japan), E. mordax (California–Mexico), E. ringens (Chile–Peru), the allozyme data of Grant and Leslie expected gene genealogical patterns are E. anchoita (Argentina–Brazil), and E. eu- (1996) because of the recency of diver- quite different under the two models, they rystole (Atlantic U.S.–Canada). gence (Figure 2a); estimates of allozyme may have converged in present-day pop- Genetic partitions in anchovies are gene diversities ranged from 0.045 in Aus- ulations due to regional extinctions that markedly different from those in sardines. tralian sardines to 0.022 in Japanese sar- erased evidence of previous population The analyses of 31 allozyme loci (Figure dines (Table 1). Hedgecock et al. (1989) histories (Figure 4). Fortunately, paleocli- 5a) and 521 bp of cytochrome b (Figure reported an estimate of H ϭ 0.010 for Cal- mate and fossil records may eventually of- 5b) indicate that anchovies are divided ifornia sardines and concluded that an an- fer a resolution of these two scenarios. In into four relatively deep lineages corre- cient population bottleneck or founder ef- a study of marine Pleistocene and Plio- sponding to the three New World species fect may have reduced genetic diversity. cene sediments from coastal California, (E. anchoita, E. ringens, E. mordax), and a These diversity values are low relative to Fitch (1969) reported a conspicuous ab- lineage consisting of all Old World species those reported for other marine fishes sence of sardine hard parts but a contin- combined: E. japonicus, E. australis, E. ca- (Ward et al. 1994) and much lower than uous record of other common species pensis, E. encrasicolus (and presumably diversities in other clupeiform fishes (see [hake (Merluccius), mackerel (Trachurus), West Atlantic E. eurystole, which was not Table 8 in Hedgecock et al. 1989). In con- and anchovy (Engraulis)]. Sardines also assayed but which is morphologically trast, polymorphisms in the mtDNA con- were not detected in elevated marine de- very similar to the European anchovy E. trol region and cytochrome b are relatively posits dating from about 100,000 to 3 mil- encrasicolus). Large genetic distances in abundant, presumably due to the elevated lions years BP, but were present in Native the allozyme survey and levels of mtDNA mutation rate in mitochondrial DNA rela- American middens about 7,000 years BP sequence divergence indicate that the tive to nuclear protein-coding loci. The (Casteel 1975). Although temporal resolu- four primary lineages have been isolated mtDNA gene trees, consisting of a network tion in elevated marine deposits is not for 6–10 million years. Shallow genetic dis- of minimal mutational distances between precise, these studies yield an approxi- tances among the Old World species, how- haplotypes for both the control region mate time frame for the arrival of the pres- ever, indicate dispersal events within the (Bowen and Grant 1997) and cytochrome ent Sardinops population in the California last few hundreds of thousands of years b (Grant et al., in press), indicate a prob- Current. It is not yet clear, however, and possibly more recently in some cases able dispersal pathway around the rim of whether this was the initial colonization (Table 2, Figure 6). the Indian–Pacific Ocean connecting South or the most recent event in an extinction/ While the Old World anchovies are Africa and Australia, with Chile, California, recolonization cycle; a fossil record ex- closely related, three of the four (except- and Japan (Figure 3a,b). tending back 5–20 million years is needed ing the southern African population E. ca- Shallow divergences among these re- to resolve this issue. Nonetheless, these pensis) contain high levels of intraregional gional populations may be explained by results are consistent with the genetic genetic diversity relative to the shallow two alternative models of population per- data in indicating that shallow gene gene- separations between species. For exam- sistence and dispersal. First, Sardinops alogies in Sardinops populations are due ple, two deep mtDNA lineages occur in Eu- may have inhabited a limited area for (at least in part) to a late Pleistocene dis- ropean anchovies, the apparent result of most of the last 20 million years before ex- persal around the rim of the Pacific Ocean. isolation between Black and Mediterra- panding to the temperate corners of the nean Sea populations during glacial maxi- Indian and Pacific Oceans in the last few ma in the early Pleistocene (Magoulas et Anchovies (Engraulis) hundred thousand years. Alternatively, re- al. 1996). An average sequence divergence gional Sardinops populations may have Anchovies are active plankton feeders of d ϭ 2.2% between haplotypes in these been extinguished repeatedly and recolon- found in the same temperate boundary two lineages (Grant WS and Bowen BW, ized by transoceanic or transequatorial currents as sardines, but additionally oc- unpublished data) is consistent with this migrants. Genetic analyses of present-day cur in less productive areas off Argentina– time frame. These lineages are in apparent populations alone may not be able to re- Brazil and in the western North Atlantic secondary contact and are codistributed solve these alternative scenarios. While (Figure 1b). Regional populations of an- throughout the Mediterranean Sea and ad-

418 The Journal of Heredity 1998:89(5) Figure 5. Majority rule, neighbor-joining trees representing phylogenetic relationships among taxa of anchovies (Engraulis). Percentage bootstrap support indicated at the nodes of trees. (a) Nei’s (1978) unbiased genetic distances based on 31 protein-coding loci (Grant et al., unpublished data). (b) Sequence divergences based on a 521 bp sequence in the cytochrome b gene of mitochondrial DNA (Grant et al., unpublished data).

chovies are not intermediate between Eu- ropean and Australian anchovies, but are embedded in the network of European haplotypes. This feature of the parsimony network indicates that a previous southern African population has become extinct and has been recolonized from Europe within the last few tens of thousands of years. Notably the reintroduced anchovies in southern Africa contain both of the Eu- ropean mtDNA lineages, implying a colonization event after the reassociation of Figure 3. Parsimony network of mtDNA haplotypes in sardines (Sardina, Sardinops). (a) Cytochrome-b. Cross the Black Sea and Mediterranean forms. bars represent transitions and ovals represent transversions between haplotypes. Asterisks indicate amino acid replacements. Haplotypes in Sardina based on a 220 bp sequence; those in Sardinops based on a 258 bp sequence (Grant et al., in press). (b) Control region. Transversion haplotypes based on a 500 bp fragment (Bowen and Grant Table 2. mtDNA haplotype and nucleotide 1997). diversities and allozyme diversities in anchovies (Engraulis)

jacent Atlantic Ocean. Since the average mtDNAa Allozymesb sequence divergence between Japanese Species h ␲ n H n and European haplotypes is only margin- anchoita 0.44 0.001 19 0.137 60 ally larger, d ϭ 2.9%, the colonization of ringens 0.41 0.001 17 0.087 30 anchovies into Mediterranean waters ap- mordax 0.88 0.007 14 0.063 30 0.075 432 pears to have occurred in the late Plio- japonicus 0.91 0.010 20 0.044 30 cene or early Pleistocene, possibly facili- 0.067 30c tated by a global cooling trend. Because australis 0.90 0.009 16 0.105 51 capensis 0.21 0.004 18 0.091 60 continental conﬁgurations during this in- 0.115 3,019d terval were essentially the same as they encrasicolus 0.94 0.015 16 0.060 25 e f are now, and because a route of coloniza- 0.88 0.016 140 0.055 634 0.75 0.017 749g tion across northern Eurasia was infeasi- ble due to ice accumulation, the only dis- a 521X bp sequence of cytochrome b; from Grant WS and persal pathway between Japan and Eu- Bowen BW, unpublished data, except where noted. b 31 loci; from Grant WS and Leslie RW, unpublished rope was around the tip of southern Africa data, except where noted. and northward to the Mediterranean (see c 22 loci; Fujio and Kato 1979. Figure 6). The intermediate position of d 31 loci; Grant 1985. some Australian haplotypes in the parsi- e RFLP analysis of 2.5 kb PCR fragment of ND5/6 genes mony network is consistent with this of NADH dehydrogenase complex; Bembo et al. 1995. f 24 loci; Bembo et al. 1996. Figure 4. Models of sardine evolution. (a) Ancient route. However, the haplotypes in present- population with a recent geographic expansion. (b) g RFLP analysis of entire mtDNA molecule; Magoulas et Population histories of extinctions and recolonizations. day populations of southern African anal. 1996.

Grant and Bowen • Shallow Genetic Architectures in Marine Fishes 419 vergences within Indian–Pacific populations of sardines and some populations of anchovies (especially the southern Afri- can form) are attributed to recent founder events, because these regional types are closely related to sister taxa. The shallow divergences observed in Argentine–Brazil- ian and Chilean–Peruvian anchovies are attributed to within-region processes because these lineages are distantly related to sister species (on a scale of 6–10 million years). Both within- and between-region comparisons are necessary to demonstrate that at least two processes (founder events and bottlenecks) are responsible for the shallow genetic architecture of anchovy and sardine populations. Figure 6. Parsimony network of 58 cytochrome b haplotypes in anchovies (Engraulis). Haplotype crossbars rep- Second, mechanisms influencing the ge- resent nucleotide substitutions between haplotypes based on a 521 bp sequence (Grant et al., unpublished data). netic architectures of regional sardine and anchovy populations are probably linked A similar scenario, invoking the possi- or founding events, within-region popula- with global trends (or oscillations) in bility of extinction and recolonization, is tion dynamics are clearly implicated, in- oceanography and climate. In recent de- apparent in the relationship between Jap- cluding severe population fluctuations, cades the size of regional sardine and an- anese and Australian anchovies. While strong natural selection on haplotypes, or chovy stocks could be estimated by the some of the Australian haplotypes are in- extinctions and recolonizations on a local magnitudes of commercial catches and re- termediate between Japan and European scale (within-region metapopulation struc- search surveys, and attempts were made lineages, recent recurring contact between ture) (see Lluch- Belda et al. 1989). Para- to correlate abundances with cyclic warm- Australia and Japan is also strongly impli- doxically, E. anchoita and E. ringens have ing events such as El Ninõs (Moser et al. cated. Australian samples include repre- the highest allozyme heterozygosities 1987; Smith and Moser 1988). The south- sentatives from at least two deep branch- among the anchovies, providing another ern oscillations that produce rapid and re- es of the Japanese network, and Japanese reminder that population processes may gionwide changes in sea surface tempera- samples include haplotypes that are more differentially affect mtDNA and allozyme tures and upwelling intensities will direct- closely related to endemic Australian hap- diversity (Grant and Leslie 1992), and that ly influence zooplankton abundance in lar- lotypes than to other Japanese haplo- caution is indicated when inferring popu- val nursery areas, and hence regulate the types. The latter observation strongly im- lation processes from any single class of abundances of spawning biomass. Roem- plies (but does not prove) a back-dispers- genetic loci (Bernatchez and Osinov 1995; mich and McGowan (1995) and Hayward al from Australia to Japan after the most Karl and Avise 1992; Karl et al. 1992; Pal- et al. (1996) note an order of magnitude recent colonization of Australian waters. umbi and Baker 1994). reduction in zooplankton abundance in In contrast to the case for Old World an- the California Current in recent decades. chovies, genetic signatures of extinctions Our terrestrial perspective is apparent Shallow Genetic Architectures in and recolonization are not apparent when a decline in sardines during the Sardines and Anchovies among the three species of New World an- same period is deemed a mystery. A com- chovies: E. mordax off the west coast of The findings outlined above lead to sev- parable decline in faunal biomass of a ter- North America, E. anchoita off Argentina eral conclusions about the genetic archi- restrial ecosystem would be obvious, as and Brazil, and E. ringens off Chile and tectures of clupeoid fishes inhabiting the would the reason for corresponding de- Peru. The deep levels of mtDNA sequence world’s temperate boundary currents. clines of primary consumers. Notably this divergence between these species indicate First, the processes shaping the genetic major decline in sardines occurred during that regional forms have been isolated for architectures of regional populations of a relatively gentle fluctuation in compari- at least 6 million years. Despite these an- globally distributed species can be under- son to the magnitudes of climatic cycles cient origins, these three species also are stood only in light of metapopulation dy- in the last 250,000 years (Dansgaard et al. characterized by low-to-moderate levels of namics on a planetary scale. For example, 1993; GRIP Members 1993). nucleotide diversity (Table 2) and star- we observe low genetic diversity in most Third, we observe considerable vari- shaped phylogenies consisting of a central of the surveyed sardine populations, but ability in the magnitudes of genetic diver- abundant haplotype with a few ‘‘satellite’’ analyses of within-region diversity will not gences between regional forms. On the haplotypes distinguished by one or two reveal whether this is due to recent origin shallowest scale, sardines and anchovies mutations (especially E. anchoita and E. (founder event) or to large fluctuations in of southern Africa share haplotypes with ringens, see Figure 6). As noted for the sar- abundance (bottleneck). The consequences fish in Australia and Europe, respectively. dines, these characteristics may be evi- of these two processes in terms of extant On the deepest scale, we observe se- dence of recent expansion from a small genetic diversity can be nearly identical quence divergences of 17% between an- number of ancestors. Since low mtDNA di- (Figure 4). However, a rangewide compar- chovies from California–Mexico and Peru– versity within these species cannot be ison of sister forms can distinguish be- Chile. Taken as a whole, we see a broad readily attributed to recent colonization tween these explanations. The shallow di- range of genetic separations from the very

420 The Journal of Heredity 1998:89(5) Table 3. Nei’s (1972, 1978) genetic distances (D) between populations (based on allozyme frequencies), next section we explore the general signif- geographic range of samples, and genetic distance from sister species for species of marine ﬁshes icance of shallow genealogies for under- Species D Geographic D with Reference standing the evolution and population bi- between range sister ology of marine ﬁshes. samples species

Anglerfish Inferences About Population Lophius vomerinus 0.0007 SE Atlantic 0.45 Leslie and Grant 1990; Grant and Leslie 1993 History From Genetic Diversity Anchovy Shallow genetic separations within spe- Engraulis spp. 0.047 Old World 0.93 Grant et al., unpublished data cies, relative to large divergence between E. anchoita 0.003 Argentina-Brazil 0.75 Grant et al., unpublished data sister species, are characteristic of many E. mordax 0.002 California 0.85 Hedgecock et al. 1989; Grant et al., unpublished data marine fishes that have been examined Cod with allozymes and mtDNA sequences. Gadus morhua 0.0037 North Atlantic 0.42 Grant and Sta˚hl 1988; Mork et al. 1985 Two results are notable in allozyme sur- G. macrocephalus 0.025 North Pacific 0.42 Grant et al. 1987b; Grant and Sta˚hl 1988 veys (Table 3). One is that sister species Bigeye of many marine fishes have been isolated Heteropriacanthus Ͻ0.01 Central Pacific 0.69 Rosenblatt and Waples 1986 for a few to several million years, based cruentatus on genetic distances calibrated with well- Hake dated geologic events. Second, the level of Merluccius capensis 0.0006 SE Atlantic 0.23 Grant et al. 1987a; Grant, unpublished differentiation between populations within data many species is an order of magnitude M. paradoxus 0.0007 SE Atlantic 0.48 Grant et al. 1987a; Grant, unpublished data less than the level between sister species. Halibut In these cases, the similarity of allele fre- Hippoglossus stenolepis 0.0002 NW Pacific 0.16 Grant et al. 1984 quencies between populations may be at- H. hippoglossus 0.001 North Atlantic 0.16 Fevolden and Haug 1988; Grant et al. tributed to mixing of eggs, larvae, and 1984 adults on extended temporal and spatial Herring scales (Waples 1987). However, surveys of Clupea harengus 0.001 North Atlantic 0.27 Grant 1984, 1986 microsatellite DNA, which has a much 0.0005 Ryman et al. 1984 higher mutation rate, may reveal mixing a C. pallasi 0.039 North Pacific 0.27 Grant and Utter 1984; Grant 1986 on time scales of decades and centuries Milkfish (Bentzen et al. 1996). b Chanos chanos 0.002 Central Pacific Ͼ1.0 Winans 1980 The analysis of mtDNA sequences al- Mullet lows marine fishes (Table 4) to be cate- Mugil cephalus 0.03 Central Pacific 0.62 Rosenblatt and Waples 1986 gorized into four classifications (Table 5) Sardine based on different combinations of small Sardinops sagax 0.005 Indian-Pacific 1.04 Grant and Leslie 1996 and large values for haplotype diversity Pufferfish (h, a measure of the frequencies and num- Spotted green puffer bers of haplotypes among individuals, Arothron hispidus Ͻ0.01 Central Pacific 0.56 Rosenblatt and Waples 1986 varying between 0–1.0) and nucleotide di- Guinaefowl puffer versity ( , average weighted sequence di- A. meleagris 0.03 Central Pacific 0.56 Rosenblatt and Waples 1986 ␲ vergence between haplotypes, varying be- a Average distance between major east-west subdivision in North Pacific. Nei’s distance within each group averages tween 0 for no divergence to over 10% for 0.0004. very deep divergences). Ideally, compari- b Monotypic genus; sister taxon uncertain. sons of gene genealogies between species should be made with homologous seg- young to the very old, probably reflecting versus protein-coding nuclear loci. During ments of DNA, but the scientific literature the diversity of outcomes that can affect population declines, the loss of genetic di- on marine fishes is not yet rich enough to species in fluctuating habitats. versity will be accelerated in mtDNA rela- allow a review based on a single segment. Finally, we observed some discordance tive to nuclear DNA due to the lower ef- Table 4 contains examples from restriction between the levels of diversity in nuclear fective population size of this maternally fragment analyses of the whole mtDNA and mitochondrial assays: Indian–Pacific inherited genome. During population growth molecule as well as direct sequencing of sardines had high haplotype diversity but the mitochondrial genome will accumulate particular mtDNA genes (cytochrome b, low allozyme diversity, while two New mutations more rapidly than protein-cod- ND4/5, and cytochrome oxidase). Muta- World anchovies had low haplotype diver- ing nuclear loci due to a higher rate of se- tion rates are certain to vary somewhat sities but high allozyme diversities. Sex- quence evolution. Hence allozyme diver- among these different sequence assays specific differences in dispersal or strong- sity might be higher shortly after popula- (see Irwin et al. 1991; Saccone et al. 1987; ly skewed sex ratios can explain such dis- tion crashes and mtDNA diversity might Walker et al. 1995), but are probably not parities in other species, but there is no be higher during a recovery phase with radically different for RFLP and mitochon- evidence that these factors operate in clu- high levels of population growth. Given drial coding regions (Birt et al. 1995; Lamb peoid fishes. A more likely explanation in- the climate-associated processes outlined et al. 1994). Direct comparisons among vokes the relative rate of evolution and above, we may expect to see both condi- the different mtDNA assays are justified the inheritance dynamics of mitochondrial tions in sardines and anchovies. In the here because the focus is on the pattern

Grant and Bowen • Shallow Genetic Architectures in Marine Fishes 421 Table 4. Haplotype and nucleotide diversities, geographic range of samples, and percentage sequence divergence from sister taxon for species of marine ﬁshes

Species Haplotype Nucleotide Geographic Sequence Reference diversity diversity range divergence (h) (␲ %) from sister species (%)

Category 1 Cod, Atlantic 0.30 0.13c North Atlantic Carr et al. 1995 0.36 0.18 NW Atlantic Carr and Marshall 1991; Zwanenburg et al. 1992 0.30 NW Atlantic Pepin and Carr 1993 Beaugregory damselfish 0.41a,b 0.30 Caribbean Shulman and Bermingham 1995 Bluefish 0.11 0.07 Australia Graves et al. 1992a Hoki 0.28 0.08 SW Pacific, SE Atlantic Baker et al. 1995 Red snapper 0.13a 0.13 Gulf of Mexico Camper et al. 1993 Red grouper 0.42a 0.08 Gulf of Mexico Richardson and Gold 1993 Weakfish 0.13 0.13 NW Atlantic Graves et al. 1992b Category 2 Blue marlin 0.84a 0.54 Atlantic–Indo-Pacific Graves and McDowell 1995 0.74 0.33 Atlantic 3.5 Finnerty and Block 1992 0.60 0.16 Pacific 3.5 Finnerty and Block 1992 Sailfish 0.80a 0.32 Atlantic–Indo-Pacific 3.5 Graves and McDowell 1995; Finnerty and Block 1992 White/striped marlin 0.82 0.29 Pacific Graves and McDowell 1995 Atlantic–Pacific 3.9 Finnerty and Block 1992 Shortfin mako 0.76a 0.35 Worldwide Heist et al. 1996 Orange roughy 0.37a 0.19 SW Pacific, SE Atlantic Smolenski et al. 1993 0.74 0.59 SW Pacific, SE Atlantic Baker et al. 1995 French grunt 0.78a,b 0.62 Caribbean Shulman and Bermingham 1995 Goldspost goby 0.98a,b 0.68 Caribbean Shulman and Bermingham 1995 Longjaw squirrelfish 0.94a,b 0.62 Caribbean Shulman and Bermingham 1995 Slippery dick 0.78a,b 0.62 Caribbean Shulman and Bermingham 1995 Sergeant major 0.79a,b 0.49 Caribbean 4.50 Shulman and Bermingham 1995 Bermingham et al. 1997 Bluehead 0.55a,b 0.48 Caribbean Shulman and Bermingham 1995 Greater amberjack 0.90 0.34 Gulf of Mexico Richardson and Gold 1993 Haddock 0.87 0.59 NW Atlantic Zwanenburg et al. 1992 Cape hake 0.90a 0.57 SE Atlantic Becker et al. 1988 Deepwater hake 0.68a 0.55 SE Atlantic Becker et al. 1988 Capelin 0.81a 0.42 NW Atlantic 3.42a Dodson et al. 1991 0.98a 0.51 NE Atlantic 3.42a Dodson et al. 1991 Atlantic herring 0.91a 0.55 NW Atlantic Kornfield and Bogdanowicz 1987 Pacific herring 0.90a 0.49 NE Pacific Schweigert and Withler 1990 Spanish sardine 0.83 0.53 W Atlantic Tringali and Wilson 1993 Red Drum 0.95a 0.58 Gulf of Mexico Gold et al. 1993 0.90a 0.56 NW Atlantic Gold et al. 1993 Stickleback 0.93 0.71 N Atlantic–Pacific Orti et al. 1994 Category 4 Bluefish 0.70 1.23 NW Atlantic Graves et al. 1992a Atlantic menhaden 1.00a 3.20 NW Atlantic Bowen and Avise 1990 Gulf menhaden 1.00a 1.00 Gulf of Mexico Bowen and Avise 1990 Redlip blenny 1.00a,b 1.09 Caribbean 12.4 Shulman and Bermingham 1995; Berming- ham et al. 1997 a Based on restriction enzyme analysis of whole mtDNA. b Average within population h. c Average percentage sequence divergence among haplotypes.

of genetic diversity rather than the abso- other example is Atlantic cod (h ϭ 0.32, ␲ points to a regional extinction during lute magnitudes. ϭ 0.15%), which show little genetic diver- Pleistocene glaciation, followed by a post- The ﬁrst category includes species with gence across the North Atlantic (Mork et glacial range expansion (Carr et al. 1995; small values of both (h Ͻ 0.5 and ␲ Ͻ al. 1985). Since ongoing gene ﬂow between Pogson et al. 1995). While such founder 0.5%). One example is the anchovy of the northeast and northwest Atlantic is events are probably an important factor, southern Africa (h ϭ 0.21, ␲ ϭ 0.40%), unlikely based on distribution and life his- these events cannot explain all the species which, as we have shown, may represent tory, the lack of differentiation (in con- in category one. Anchovies off Chile– Peru a recent recolonization from Europe. An- junction with biogeographic evidence) (h ϭ 0.41, ␲ ϭ 0.10%) and off Argentina–

422 The Journal of Heredity 1998:89(5) Table 5. Interpreting haplotype and nucleotide a more recent scale, perhaps the last few with a recent coalescence of mtDNA lin- diversities for marine fishes hundred thousand years. eages and shallow histories. It is clear that h A third category, low h and high ␲, char- shifts in climate or oceanographic condi- ␲ Small Large acterizes populations with a few highly di- tions can be responsible for this condi- vergent haplotypes. This condition may tion. What additional factors contribute to Small 1. Recent population 2. Population bottle- result from secondary contact between this trend? Using recursive simulations, bottleneck or neck followed by rap- founder event by id population growth isolated populations or by a strong bottle- Avise et al. (1984) showed that in a stable single or a few and accumulation of neck in a formerly large, stable population. population there is a high probability that mtDNA lineages. mutations. Secondary reassociation of formerly iso- all haplotypes in the population can be Large 3. Divergence be- 4. Large stable popula- tween geographi- tion with long evolu- lated populations is certain to occur in the traced to a single female after 4N genera- cally subdivided tionary history or marine realm (see Veron 1995), and retic- tions, where N is the female effective pop- populations. secondary contact ulation of isolated lineages may be rela- ulation size. Hence the loss of female lin- between differentiated lineages. tively common, but this must be coupled eages will accelerate in declining popula- with low effective population sizes (to tions or during fluctuations in abundance, maintain low h) in order to fit the criteria and the expected time to coalescence of of category 3. Coastal and oceanic fishes extant lineages will be correspondingly Brazil (h ϭ 0.44, ␲ ϭ 0.10%) also have low are usually not subdivided into small iso- shorter. A second factor known to pro- haplotype and nucleotide diversities, but lated populations, so it may be that few duce shallow coalescence of extant lin- the ancient origin of these forms pre- open-ocean fish fit into this category. Such eages is a large variance in reproductive cludes an explanation based on recent col- conditions may be more applicable to in- success, which can decrease the genetic onization. Other mechanisms such as pe- shore fauna (Burton 1986; Planes and Do- effective size of a population without ac- riodic regionwide bottlenecks or metapop- herty 1997) and freshwater organisms tually reducing population size (Hedge- ulation structure within regions must be (Bermingham and Avise 1986). cock 1994; Hedgecock et al. 1994). Marine invoked to produce the observed low lev- The fourth category consists of species fishes tend to have very large reproduc- els of diversity. Other examples in cate- with large values of both h and ␲. The high tive potentials (although exceptions exist, gory 1 include Beaugregory damselfish (h level of divergence between haplotypes especially among the cartilaginous fishes), ϭ 0.41, ␲ ϭ 0.30%), Australian bluefish (h may be attributed to secondary contact but propitious combinations of biological ϭ 0.11, ␲ ϭ 0.07%), hoki (h ϭ 0.28, ␲ ϭ between previously differentiated allopat- and physical conditions are required for 0.08%), red snapper (h ϭ 0.13, ␲ ϭ 0.13%), ric lineages (as in category 3) or to a long larvae to survive and recruit into the adult and weakfish (h ϭ 0.13, ␲ ϭ 0.13%). Al- evolutionary history in a large stable pop- population. Under conditions of high vari- though little is known about the evolution- ulation. Examples of the first condition ance in reproductive success, an entire ary histories of these fishes, their genetic may include the European anchovy (h ϭ year class may be the product of relatively architectures uniformly indicate periods 0.86, ␲ ϭ 1.6%) and Atlantic menhaden (h few matings. Evidence for such sweep- of low effective population size within re- ϭ 1.0, ␲ ϭ 3.20%), both of which contain stakes recruitment in marine fishes comes cent thousands or tens of thousands of a pair of divergent and twiggy mtDNA lin- from the observation of genetic differ- years. eages which (based on geographic consid- ences among individual schools of Califor- The second category consists of popu- erations) probably arose in isolation. Pos- nia anchovies (Hedgecock 1994), Black lations with high h and low ␲. This con- sible examples of the second condition in- Sea anchovies (Altukhov 1990), south Af- dition is attributed to expansion after a clude the Japanese anchovy (h ϭ 0.91, ␲ rican anchovies (Grant 1985), Norwegian period of low effective population size; ϭ 1.0%), Atlantic bluefish (h ϭ 0.70, ␲ ϭ sprat (Sprattus sprattus; Nævdal 1968), and rapid population growth enhances the re- 1.23%), Caribbean blenny (h ϭ 1.0, ␲ ϭ redfish (Sebastes mentella; Altukhov 1990). tention of new mutations (Avise et al. 1.09%), and Gulf menhaden (h ϭ 1.0, ␲ ϭ In these cases the effective population 1984; Rogers and Harpending 1987). Ex- 1.0%), for which extended geographic iso- size for maternally inherited genes may be amples are typically drawn from large pop- lation is unlikely because of the configu- one or two orders of magnitude smaller ulations or entire species which contain ration of the open coastline where they than the census size (see Bowen and Avise one or two prevalent haplotypes embed- occur (Japanese anchovy and Gulf men- 1990), leading to higher rates of lineage ex- ded in a cluster of ‘‘twigs’’ that are one or haden) or because of strong dispersal ca- tinction than in populations of the same a few mutations removed from the central pabilities (bluefish). It is notable that even size with many successful spawners. haplotypes. This second category in- in category 4 the levels of divergences be- On the other hand, life-history patterns cludes several of the billfishes (h ϭ 0.68– tween lineages are typically an order of producing strong population subdivisions 0.85, ␲ ϭ 0.29–0.54%), shortfin mako (h ϭ magnitude less than the divergence be- may increase the time to coalescence, but 0.76, ␲ ϭ 0.35%), as well as northwest At- tween sister taxa. numerous allozyme studies indicate that lantic capelin, northeast Atlantic capelin, marine fishes do not have strong popula- goldspot goby, French grunt, slippery tion partitions relative to freshwater and Shallow Genetic Architectures in dick, longjaw squirrelfish, greater amber- anadromous fishes (see Ward et al. 1994). Marine Fishes jack, haddock, Cape hake, northwest At- Lower levels of differentiation between lantic and northeast Pacific herring, red These four categories are defined by de- marine fish populations are attributed to drum, and west Atlantic Spanish sardine mographic events that alter the likelihood higher dispersal potential during plank- (h ϭ 0.79–0.98, ␲ ϭ 0.29–0.68%). Many of of mtDNA lineage survival and the time to tonic egg, larval, or adult life-history these species are believed to have origi- ancestral coalescence of lineages. Most of stages, coupled with an absence of physi- nated in the Pliocene or early Pleistocene, the species in Table 4 fit the first or sec- cal barriers to movement between ocean but their mtDNA genealogies coalesce on ond categories, which include populations basins or adjacent continental margins. In

Grant and Bowen • Shallow Genetic Architectures in Marine Fishes 423 contrast, strong population subdivisions be susceptible to regional extinction. The Bermingham E, McCafferty SS, and Martin AP, 1997. Fish biogeography and molecular clocks: perspectives (and corresponding barriers) for freshwa- passenger pigeon analogy may be appro- from the Panamanian Isthmus. In: Molecular system- ter fishes may serve to retain divergent priate for coastal marine fishes, especially atics of fishes (Kocher TD and Stepien CA, eds). San lineages (see Bermingham and Avise 1986; those in upwelling zones and other fluc- Diego: Academic Press; 113–128. Mayden 1993). The physical factors which tuating but productive habitats. Most of Bernatchez L and Osinov A, 1995. Genetic divergence of trout (genus Salmo) from its eastern native range buffer freshwater fish lineages against ex- the demographic indices of healthy ma- based on mitochondrial DNA and nuclear gene varia- tinctions are notably absent from the marine fish populations are ratcheted down- tion. Mol Ecol 4:285–297. rine realm. ward by overharvesting, and in many Birt TP, Friesen VL, Birt RD, Green JM, and Davidson cases the majority of the biomass in heavi- WS, 1995. Mitochondrial DNA variation in Atlantic capelin, Mallotus villosus: a comparison of restriction ly fished populations consists of young ConclusionÐConservation and sequence analyses. Mol Ecol 4:771–776. fish. Under these circumstances, recruit- Lessons Bowen BW and Avise JC, 1990. Genetic structure of At- ment failures over 3 or 4 years could lead lantic and Gulf of Mexico populations of sea bass, men- Marine fishes are generally regarded as re- not only to commercial extinction but to haden, and sturgeon: influence of zoogeographic fac- sistant to extinction because of large dif- the total extinction of a regional popula- tors and life-history patterns. Mar Biol 107:371–381. fuse populations and because marine wa- tion or species. At least for sardines and Bowen BW and Grant WS, 1997. Phylogeography of the sardines (Sardinops spp.): assessing biogeographic ters are often viewed as boundless habi- anchovies, genetic imprints indicate that models and population histories in temperate upwell- tats. As a result, few species of marine fish regional collapses occur without the add- ing zones. Evolution 51:1601–1610. are considered to be strong conservation ed burden of intense harvesting. There- Broecker WS, 1995. Chaotic climate. Sci Am 273:62–68. concerns (Vincent and Hall 1996). How- fore, management strategies for sardine Burton RS, 1986. Evolutionary consequences of re- ever, several factors promoting lineage and anchovy fisheries, which are among stricted gene flow among natural populations of the co- pepod Tigriopus californicus. Bull Mar Sci 39:526–535. turnover and shallow population structure the most productive harvests on the plan- in several time scales may make marine et, must include allowances for the fragil- Butler JL, 1991. Mortality and recruitment of Pacific sardine, Sardinops sagax caerulea, larvae in the Califor- fishes vulnerable to overharvesting and ity of populations in unstable habitats. De- nia Current. Can J Fish Aquat Sci 48:1713–1723. climate change. Lineage sorting from pleted stocks will not invariably recover. Camper JD, Barber RC, Richardson LR, and Gold JR, sweepstakes recruitment takes place on a 1993. Mitochondrial DNA variation among red snapper scale of generations and is proportional to (Lutjanus campechanus) from the Gulf of Mexico. Mol Mar Biol Biotech 2:154–161. the magnitude of the reproductive vari- References Carr SM and Marshall HD, 1991. Detection of intraspe- ance in the population. Major population Altukhov YP, 1990. Population genetics: diversity and cific DNA sequence variation in the mitochondrial cy- fluctuations, at least in sardines and an- stability. London: Harwood Academic. tochrome b gene of Atlantic cod (Gadus morhua) by the chovies, take place on a scale of decades Avise JC, Arnold J, Ball RM, Bermingham E, Lamb T, polymerase chain reaction. Can J Fish Aquat Sci 48:48– Neigel JE, Reeb CA, and Saunders NC, 1987. Intraspe- 52. and centuries. Regional extinctions, dis- cific phylogeography: the mitochondrial DNA bridge Carr SM, Snellen AJ, Howse KA, and Wroblewski JS, persal events, and recolonizations take between population genetics and systematics. Annu 1995. Mitochondrial DNA sequence variation and ge- place on a scale of thousands to millions Rev Ecol Syst 18:489–522. netic stock structure of Atlantic cod (Gadus morhua) of years. The relative importance of these Avise JC, Neigel JE, and Arnold J, 1984. Demographic from bay and offshore locations on the Newfoundland influences on mitochondrial DNA lineage survivorship continental shelf. Mol Ecol 4:79–88. factors may differ between tropical and in animal populations. J Mol Evol 20:99–105. Casteel RW, 1975. An early post-glacial record of the temperate zones, and between species Baker CS, Perry A, Chambers GK, and Smith PJ, 1995. Pacific sardine, Sardinops sagax, from Saanish Inlet, with diverse life histories. Nonetheless, Population variation in the mitochondrial cytochrome Vancouver Island, British Columbia. Copeia 1975:576– they probably all contribute to the ob- b gene of the orange roughy Hoplostethus atlanticus and 579. the hoki Macruronus novaezelandiae. Mar Biol 122:503– Dansgaard W, Johnsen SJ, Clausen HB, Dahl-Jensen D, served trend of shallow genetic architec- 509. Gundestrup NS, Hammer, CU, Hvidberg CS, Steffensen ture in marine fishes. Barnes JT, Jacobson LD, MacCall AD, and Wolf P, 1992. JP, Sveinbjo¨rnsdottir, Jouzel J, and Bond G, 1993. Evi- What are the implications of shallow Recent population trends and abundance of the Pacific dence for general instability of past climate from a 250- mtDNA population structure and low ge- sardine (Sardinops sagax). CalCOFI Rep 33:60–75. kyr ice-core record. Nature 364:218–220. netic diversity for the conservation of ma- Baumgartner TR, Souter A, and Ferreira-Bartrina V, Dodson JJ, Carscadden JE, Bernatchez L, and Colom- rine fishes? First, it is apparent that this is 1992. Reconstruction of the history of Pacific sardine bani F, 1991. Relationship between spawning mode and and northern anchovy populations over the past two phylogenetic structure in mitochondrial DNA of North a widespread phenomenon among marine millennia from sediments of the Santa Barbara Basin, Atlantic capelin Mallotus villosus. Mar Ecol Prog Ser 76: fishes, and therefore only exacerbated in California. CalCOFI Rep 33:24–40. 103–113. recent decades by deteriorating coastal Becker II, Grant WS, Kirby R, and Robb FT, 1988. Evo- Fevolden SE and Haug T, 1988. Genetic population lutionary divergence between sympatric species of and pelagic habitats (Sherman 1994) and structure of Atlantic halibut, Hippoglossus hippoglossus. southern African hakes, Merluccius capensis and M. par- Can J Fish Aquat Sci 45:2–7. fishing activities. If natural conditions adoxus. II. Restriction enzyme analysis of mitochondrial DNA. Heredity 61:21–30. Finnerty JR and Block BA, 1992. Direct sequencing of commonly result in low mtDNA diversity mitochondrial DNA detects highly divergent haplo- in marine fishes, then such findings do not Bembo DG, Carvalho GR, Snow M, Cingolani N, and types in blue marlin (Makaira nigricans). Mol Mar Biol Pitcher TJ, 1995. Stock discrimination among European Biotech 1:206–214. invariably signal inbreeding depression or anchovies, Engraulis encrasicolus, by means of PCR-am- other genetic health problems. However, plified mitochondrial DNA analysis. Fish Bull 94:31–40. Fitch JE, 1969. Fossil records of certain schooling fishes of the California current system. Rep Calif Coop Ocean evolutionarily rapid drops in genetic di- Bembo DG, Carvalho GR, Cingolani N, and Pitcher TJ, Fish Invest 13:70–80. versity due to fishing (Smith et al. 1991), 1996. Electorphoretic analysis of stock structure in northern Mediterranean anchovies, Engraulis encrasi- Fujio Y and Kato Y, 1979. Genetic variation in fish pop- and the loss of low-frequency alleles (not colus. ICES J Mar Sci 53:115–128. ulations. Bull Jap Soc Sci Fish 45:1169– 1178. usually detected by estimates of hetero- Bentzen P, Taggart CT, Ruzzante DE, and Cook D, 1996. Gold JR, Richardson LR, Furman C, and King TL, 1993. zygosity) may be of special concern to the Microsatellite polymorphism and the population struc- Mitochondrial DNA differentiation and population genetic health of marine species and to ture of Atlantic cod (Gadus morhua) in the northwest structure in red drum (Sciaenops ocellatus) from the Atlantic. Can J Fish Aquat Sci 53:2706–2721. Gulf of Mexico and Atlantic Ocean. Mar Biol 116:175– the maintenance of their evolutionary po- 185. Bermingham E and Avise JC, 1986. Molecular zooge- tential (Ryman et al. 1991). ography of freshwater fishes in the southeastern Unit- Gold JR, Richardson LR, Furman C, and Sun F, 1994. Second, even very large populations can ed States. Genetics 113:939–965. Mitochondrial DNA diversity and population structure

424 The Journal of Heredity 1998:89(5) in marine fish species from the Gulf of Mexico. Can J Hansen J, Johnson D, Lacis A., Lededeff, S, Lee P, Rind Lluch-Belda D, Crawford RJM, Kawasaki T, MacCall AD, Fish Aquat Sci 51(suppl 1):205–214. D, and Russel G, 1981. Climate impact of increasing at- Parrish RH, Schwartzlose RA, and Smith PE, 1989. mospheric carbon dioxide. Science 213:957–966. World-wide fluctuations of sardine and anchovy stocks: Grant WS, 1984. Biochemical population genetics of At- the regime problem. S Afr J Mar Sci 8:195–205. lantic herring, Clupea harengus. Copeia 1984:357–364. Hayward TL, 1997. Pacific Ocean climate change: at- mospheric forcing, ocean circulation and ecosystem re- Lo NCH, Green-Ruiz YA, Cervantes MJ, Moser HG, and Grant WS, 1985. Biochemical genetic stock structure of sponse. Trends Ecol Evol 12:150–154. Lynn RJ, 1996. Egg production and spawning biomass the southern African anchovy, Engraulis capensis of Pacific sardine (Sardinops sagax) in 1994, deter- Gilchrist. J Fish Biol 27:23–29. Hayward TL, Cummings SL, Cayan DR, Chavez FP, Lynn mined by the daily egg production method. Calif Coop RJ, Mantyla AW, Niller PP, Schwing FB, Veit RR, and Ven- Grant WS, 1986. Biochemical genetic divergence be- Oceanic Fish Invest Rep 37:160–174. rick E, 1996. The state of the California Current in 1995: tween Atlantic, Clupea harengus, and Pacific, C. pallasi, continuing declines in macrozooplankton biomass dur- Lo NCH and Methot RD, 1989. Spawning biomass of the herring. Copeia 1986:714–719. ing a period of normal circulation. CalCOFI Rep 37:22– northern anchovy in 1988. Calif Coop Oceanic Fish In- Grant WS, 1987. Genetic divergence between congener- 37. vest Rep 30:18–31. ic Atlantic and Pacific Ocean fishes. In: Population ge- Hedgecock D, 1994. Does variance in reproductive suc- Lorius C, Jouzel J, Ritz C, Merlivat L, Barkov NI, Korot- netics and fishery management (Ryman R and Utter cess limit effective population sizes of marine organ- kevich YS, and Kotlyakov VM, 1985. A 150,000-year cli- FM, eds). Seattle: University of Washington Press; 225– isms? In: Genetics and evolution of aquatic organisms matic record from Antarctic ice. Nature 316:591–596. 246. (Beaumont A, ed). London: Chapman & Hall; 122–134. Magoulas A, Tsimenides N, and Zouros E, 1996. Mito- Grant WS and Leslie RW, 1992. Effect of metapopulation Hedgecock D, Hutchinson ES, Li G, Sly FL, and Nelson chondrial DNA phylogeny and the reconstruction of structure on nuclear and organellar DNA variability in K, 1989. Genetic and morphometric variation in the Pa- the population history of a species: the case of the Eu- semi-arid environments of southern Africa. S Afr J Sci cific sardine, Sardinops sagax caerulea: comparisons ropean anchovy (Engraulis encrasicolus). Mol Biol Evol 89:287–293. and contrasts with historical data and with variability 13:178–190. Grant WS and Leslie RW, 1993. Biochemical divergence in the northern anchovy, Engraulis mordax. Fish Bull 87: Mayden RL (ed), 1993. Systematics, historical ecology, and biogeography of anglerfish of the genus Lopius 653–671. and North American freshwater fishes. Stanford: Stan- (Lophiiformes). J Zool Lond 231:465–485. Hedgecock D, Hutchinson ES, Li G, Sly FL, and Nelson ford University Press. Grant WS and Leslie RW, 1996. Late Pleistocene dis- K, 1994. The central stock of northern anchovy (En- Mork J, Ryman N, Sta˚hl G, Utter F, and Sundnes G, 1985. persal of Indian-Pacific sardine populations in an an- graulis mordax) is not a randomly mating population. Genetic variation in Atlantic cod (Gadus morhua) cient lineage of the genus Sardinops. Mar Biol 126:133– CalCOFI Rep 35:121– 136. throughout its range. Can J Fish Aquat Sci 42:1580– 142. Heist EJ, Musick JA, and Graves JE, 1996. Genetic pop- 1587. Grant WS and Sta˚hl G, 1988. Evolution of Atlantic and ulation structure of the shortfin mako (Isurus oxyrin- Moser, HG, Smith PE, and Eber LE. 1987. Larval fish Pacific cod: loss of genetic variation and gene expres- chus) inferred from restriction fragment length poly- assemblages in the California Current region during sion in Pacific cod. Evolution 42:138–146. morphism analysis of mitochondrial DNA. Can J Fish Aquat Sci 53:583–588. 1954–1960, a period of dynamic environmental change. Grant WS and Utter FM, 1984. Biochemical population CalCOFI Rep 28:97–127. genetics of Pacific herring (Clupea pallasi). Can J Fish Hubbs CL. 1952. Antitropical distribution of fishes and Murphy GI, 1966. Population biology of the Pacific sar- Aquat Sci 41:856–864. other organisms. Proc 7th Pac Sci Congr 3:324–329. dine (Sardinops caerulea). Proc Calif Acad Sci 34:1–84. Grant WS, Clark AM, and Bowen BW, in press. Why Irwin DM, Kocher TD, and Wilson AC, 1991. Evolution Murphy GI, 1967. Vital statistics of the Pacific sardine RFLP analysis of mitochondrial DNA failed to resolve of the cytochrome b gene of mammals. J Mol Evol 32: (Sardinops caerulea) and the population consequences. sardine (Sardinops) biogeography: insights from mito- 128–144. Ecology 48:731–736. chondrial DNA cytochrome b sequences. Can J Fish Jouzel J, Barkov NI, Barnola JM, Bender M, Chappallez, Aquat Sci. Genthon C, Kotlyakov VM, Lipenkov V, Lorius C, Petit Nei M, 1972. Genetic distance between populations. Am Nat 106:283–292. Grant WS, Leslie RW, and Becker II, 1987a. Genetic JR, Raynaud D, Raisbeck G, Ritz C, Sowers T, Stievenard stock structure of the southern African hakes Merluc- M, Yiou F, and Yiou P, 1993. Extending the Vostok ice- Nei M, 1978. Estimation of average heterozygosity and cius capensis and M. paradoxus. Mar Ecol Prog Ser 41: core record of palaeoclimate to the penultimate glacial genetic distance from a small number of individuals. 9–20. period. Nature 364:407–412. Genetics 89:583–590. Grant WS, Teel DJ, Kobayashi T, and Schmitt C, 1984. Karl SA and Avise JC, 1992. Balancing selection at allo- Nelson G, 1984. Identity of the anchovy Hildebrandi- Biochemical population genetics of Pacific halibut (Hip- zyme loci in oysters: implications from nuclear RFLPs. chthys setiger with notes on relationships and bioge- poglossus stenolepis) and comparison with Atlantic hal- Science 256:100–102. ography of the genera Engraulis and Cetengraulis. Co- ibut (H. hippoglosus). Can J Fish Aquat Sci 41:1083– Karl SA, Bowen BW, and Avise, JC, 1992. Global popu- peia 1984:422–427. 1088. lation structure and male-mediated gene flow in the Nævdal G, 1968. Studies on haemoglobins and serum Grant WS, Zhang CI, Kobayashi T, and Sta˚hl G, 1987b. green turtle (Chelonia mydas): RFLP analyses of anon- proteins in sprat from Norwegian waters. Fishdir Skr Lack of genetic stock discretion in Pacific cod (Gadus ymous nuclear DNA regions. Genetics 131:163–173. Ser Havunders 14:160–182. macrocephalus). Can J Fish Aquat Sci 44:490–498. Kawasaki T, 1993. Recovery and collapse of the Far Okazaki T, Kobayashi T, and Uozumi Y, 1996. Genetic Graves JE, 1995. Conservation genetics of fishes in the Eastern sardine. Fish Oceanogr 2:244–253. relationships of pilchards (genus: Sardinops) with anti- pelagic marine realm. In: Conservation genetics, case Kawasaki T and Omori M, 1995. Possible mechanisms tropical distributions. Mar Biol 126:585–590. histories from nature (Avise JC and Hamrick JL, eds). underlying fluctuations in the Far Eastern sardine pop- Orti G, Bell MA, Reimchen TE, and Meyer A, 1994. Glob- New York: Chapman & Hall; 335–366. ulation inferred from time series of two biological al survey of mitochondrial DNA sequences in the three- Graves JE and Dizon AE, 1989. Mitochondrial DNA se- traits. Fish Oceanogr 4:238–242. spine stickleback: evidence for recent migrations. Evo- quence similarity of Atlantic and Pacific albacore tuna Keigwin LD, 1978. Pliocene closing of the Isthmus of lution 48:608–622. (Thunnus alalunga). Can J Fish Aquat Sci 46:870–873. Panama based on biostratigraphic evidence from near- Ovenden JR, Smolenski AJ, and White RWG, 1989. Mi- Graves JE and McDowell JR, 1995. Inter-ocean genetic by Pacific Ocean and Caribbean Sea cores. Geology 6: tochondrial DNA restriction site variation in Tasmanian divergence of istiophorid billfishes. Mar Biol 122:193– 630–634. populations of orange roughy (Hoplostethus atlanticus), 203. Keigwin LD, 1982. Isotopic paleoceanography of the Ca- a deep-water marine teleost. Aust J Mar Freshwater Res 40:1–9. Graves JE, Ferris SD, and Dizon AE, 1984. Close genetic ribbean and east Pacific: role of Panama uplift in late similarity of Atlantic and Pacific skipjack tuna (Katsu- Neogene time. Science 217:350–353. Palumbi SR and Baker CS, 1994. Contrasting views of wonus pelamis) demonstrated with restriction endo- Kornfield I and Bogdanowicz SM, 1987. Differentiation humpback whale population structure using mitochon- nuclease analysis of mitochondrial DNA. Mar Biol 79: of mitochondrial DNA in Atlantic herring, Clupea har- drial and nuclear DNA sequences. Mol Biol Evol 11:426– 315–319. engus. Fish Bull US 85:561–568. 435. Graves JE, McDowell JR, Beardsley AM, and Scoles DR, Lamb T, Lydeard C, Walker RB, and Gibbons JW, 1994. Parrish RH and Mallicoate DL, 1995. Variation in the 1992a. Stock structure of the bluefish Pomatomus sal- Molecular systematics of map turtles (Graptemys): a condition factors of pelagic fishes and associated en- tatrix along the mid-Atlantic coast. Fish Bull US 90:703– comparison of mitochondrial DNA restriction site ver- vironmental factors. Fish Oceanogr 4:171–190. 710. sus sequence data. Syst Biol 43:543–559. Parrish RH, Serra R, and Grant WS, 1989. The monotyp- Graves JE, McDowell JR, and Jones ML, 1992b. A ge- Lehman SJ and Kiegwin LD, 1992. Sudden changes in ic sardines, Sardina and Sardinops: their taxonomy, dis- netic analysis of weakfish Cynoscion regalis stock struc- North Atlantic circulation during the last deglaciation. tribution, stock structure, and zoogeography. Can J ture along the mid-Atlantic coast. Fish Bull US 90:469– Nature 356:757–762. Fish Aquat Sci 46:2019–2036. 475. Leslie RW and Grant WS, 1990. Lack of congruence be- Pepin P and Carr SM, 1993. Morphological, meristic, GRIP Members, 1993. Climate instability during the last tween genetic and morphological stock structure of the and genetic analysis of stock structure in juvenile At- interglacial period recorded in GRIP ice core. Nature southern African anglerfish Lophius vomerinus. S Afr J lantic cod (Gadus morhua) from the Newfoundland 364:203–207. Mar Sci 9:379–398. shelf. Can J Fish Aquat Sci 50:1924–1933.

Grant and Bowen • Shallow Genetic Architectures in Marine Fishes 425 Planes S and Doherty PJ, 1997. Genetic and color inter- resis and mitochondrial DNA analysis. Am Fish Soc Caribbean: support for the molecular clock. Evolution actions at a contact zone of Acanthochromis polyacan- Symp 7:459–469. 34:705–711. thus: a marine fish lacking pelagic larvae. Evolution 51: Sherman K, 1994. Sustainability, biomass yields, and Veron, JEN, 1995. Corals in space and time; the bioge- 1232– 1243. health of coastal ecosystems: an ecological perspec- ography and evolution of the Scleractinia. Ithaca, New Pogson GH, Mesa KA, and Boutilier RG, 1995. Genetic tive. Mar Ecol Prog Ser 112:277–301. York: Comstock/Cornell. population structure and gene flow in the Atlantic cod Shields GF and Gust JR, 1995. Lack of geographic struc- Gadus morhua: a comparison of allozyme and nuclear Vincent CJ and Hall HJ, 1996. The threatened status of ture in mitochondrial DNA sequences of Bering Sea RFLP loci. Genetics 139:375–385. marine fishes. Trends Ecol Evol 11:360–361. walleye pollock, Theragra chalcogramma. Mol Mar Biol Richardson LR and Gold JR, 1993. Mitochondrial DNA Biotech 4:69–82. Walker D, Burke VJ, Barak I, and Avise JC, 1995. A comparison of mtDNA restriction site vs. control region se- variation in red grouper (Epinephelus morio) and great- Shulman MJ and Bermingham E, 1995. Early life histo- er amberjack (Seriola dumerili) from the Gulf of Mexico. quences in phylogeographic assessment of the musk ries, ocean currents, and the population genetics of Ca- turtle (Sternotherus minor). Mol Ecol 4:365–373. ICES J Mar Sci 50:53–62. ribbean reef fishes. Evolution 49:897–910. Roemmich D and McGowan J, 1995. Climatic warming Waples RS, 1987. A multispecies approach to the anal- Smith PE and Moser HG, 1988. CalCOFI time series: an ysis of gene flow in marine shore fishes. Evolution 41: and the decline of zooplankton in the California Cur- overview of fishes. CalCOFI Rep 29:66–77. rent. Science 267:1324–1326. 385–400. Smith PE, Lo HCH, and Butler JL, 1992. Life-state du- Ward RD, Woodward M, and Skibinski DOF, 1994. A Rogers AR and Harpending H, 1992. Population growth ration and survival parameters as related to interde- makes waves in the distribution of pairwise genetic dif- comparison of genetic diversity levels in marine, fresh- cadal population variability in Pacific sardine. Calif water, and anadromous fishes. J Fish Biol 44:213–227. ferences. Mol Biol Evol 9:552–569. Coop Ocean Fish Invest Rep 33:41–49. Watanabe Y and Kuroki T, 1997. Asymptotic growth tra- Rosenblatt RH and Waples RS, 1986. A genetic compar- Smith PE, Santander H, and Alheit J, 1989. Comparison jectories of larval sardine (Sardinops melanostictus) in ison of allopatric population of shore fish species from of the mortality and dispersal of sardine (Sardinops sa- the coastal waters off western Japan. Mar Biol 127:369– the eastern and central Pacific Ocean: dispersal or vi- gax sagax) and anchovy (Engraulis ringens) eggs off 378. cariance? Copeia 1986:275–284. Peru. Fish Bull US 87:497–508. Watson RT, Zinyowera MC, and Moss RH (eds), 1996. Ryman N, Utter F, and Laikre L, 1991. Protection of Smith PJ, Francis RICC, and McVeagh M, 1991. Loss of Climate change 1995: impacts, adaptations and mitiga- aquatic biodiversity. In: The state of the world’s fish- genetic diversity due to fishing pressure. Fish Res 10: tion of climate change. Scientific-technical analyses. eries resources: Proceedings of the World Fisheries 309–316. Congress plenary sessions (Voigtlander CW, ed). New IPCC Working Group II. Cambridge: Cambridge Univer- Delhi: Oxford; 92–115. Smolenski AJ, Ovenden JR, and White RWG, 1993. Evi- sity Press. dence of stock separation in southern hemisphere or- Ryman N, Lagercrantz U, Andersson L, Chakraborty R, ange roughy (Hoplostethus atlanticus, Trachichthyidae) Watterson GA, 1984. Allele frequencies after a bottle- and Rosenberg R, 1984. Lack of correspondence be- from restriction-enzyme analysis of mitochondrial neck. Theor Popul Biol 26:387–407. tween genetic and morphologic variability patterns in DNA. Mar Biol 116:219–230. Whitehead PJP, 1973. The clupeoid fishes of the Guian- Atlantic herring (Clupea harengus). Heredity 53:687– as. Bull Br Mus (Nat Hist), Zool, suppl. 5:1–227. 704. Steininger FF, Rabeder G, and Ro¨gl F, 1985. Land mam- mal distribution in the Mediterranean Neogene: a con- Winans GA, 1980. Geographic variation in the milkfish, Saccone C, Attimonelli M, and Sbisa E, 1987. Structural sequence of geokinematic and climatic events. In: Geo- Chanos chanos: I. Biochemical evidence. Evolution 34: elements highly preserved during the evolution of the logical evolution of the Mediterranean basin (Stanley 558–574. D-loop-containing region in vertebrate mitochondrial DJ and Wezel FC, eds). New York: Springer Verlag; 559– DNA. J Mol Evol 26:205–211. 571. Wolf P, 1992. Management of the recovery of the Pacific sardine and the recent California fishery. CalCOFI Rep Santer BD, Taylor KE, Wigley TML, Johns TC, Jones PD, Tringali MD and Wilson RR Jr, 1993. Differences in hap- 33:76–86. Karoly DJ, MItchell JFB, Oort AH, Penner JE, Ramas- lotype frequencies of mtDNA of the Spanish sardine wamy V, Schwarzkopf MD, Stoffer RJ, and Tett S, 1996. Sardinella aurita between specimens from the eastern Zwanenburg KCT, Bentzen P, and Wright JM, 1992. Mi- A search for human influences on the thermal structure Gulf of Mexico and southern Brazil. Fish Bull 91:362– tochondrial DNA differentiation in western North Atlan- of the atmosphere. Nature 382:39–46. 370. tic populations of Haddock (Melanogrammus aeglefinus). Can J Fish Aquat Sci 49:2527–2537. Schweigert JF and Withler RE, 1990. Genetic differenti- Vawter AT, Rosenblatt R, and Gorman GC, 1980. Genetic ation of Pacific herring based on enzyme electropho- divergence among fishes of the eastern Pacific and the Corresponding Editor: Fred W. Allendorf

426 The Journal of Heredity 1998:89(5)