Phylogeography of pelagic fishes

December 10, 2012

The Gulf of Mexico (GOM) today consists of a large mediterranean sea with limited water exchanged. In through the Yucatan Channel, and out through the narrow and shallow Straits of Florida.

The loop current and mesoscale eddies it sheds are important oceanographic microcosms that divide Caribbean water masses from the distinct GOM mass. The larvae of marine organisms may be physically entrained in eddies, while slow-swimming adults may be transported into areas of unsuitable environmental conditions before eddies lose vorticity and mix with ambient seawater.

The sill restricting inflow into the basin is deeper and wider than the outflow sill near Florida, so the exiting current is strengthened and curls around the Florida peninsula as the Florida Current to become the Gulf Stream near Cape Hatteras, NC. The strong current and propagation of tropical waters influences the distribution of marine species, and in particular advances the northern range of sub-tropical fish. It is not unusual to find Caribbean species north as far as Massachusetts.

18,000 years ago the oceanographic regime was very different (Cunningham and Collins 1998; Bowen and Avise 1990). Falling Pleistocene sea levels exposed previously submerged land around Florida and the Yucatan Peninsula, choking oceanographic exchange between the Gulf and the western Atlantic.

During this period of isolation, aridity in Florida meant reduced river outflow, and a subsequent lack of estuarine nursery and feeding habitat for fish species. Population ranges receded into the Gulf, where coastal habitats offered sufficient shelter, until sea levels once again rose during the current interglacial.

An alternate, non-exclusive theory is that population separations occur during interglacial periods. Today, for example, the Florida peninsula remains a barrier to temperate species. The conditions along south Florida coast and offshore are simply too tropical for colonization of these areas (Bowen and Avise 1990). During glaciation, lower temperatures expand the southern range of some species, which may become genetically reconnected.

Increasing sea surface temperatures associated with climate change will likely exacerbate this effect in tolerant sub-tropical species with tenuous connectivity.

Scientists now use genetics to establish boundaries (or lack thereof) for marine populations, especially commercially fished species.

The modern genetic structure at the Gulf-Atlantic boundary is a fascination of phylogeographers because of the variety of conclusions emerging from studies considering varying life histories and molecular methods.

Horseshoe crabs, oysters, sea bass, and a menagerie of other species show sharp genetic difference across the boundary. This shared genetic structure is due to vicariance, or the physical (in this case geological) partitioning of once connected populations.

For the most part vicariance, as exemplified by the Florida biogeographic barrier, acts equally amongst species of similar life histories. So, distinguishing characteristics of genetic structure for co-occurring species are indicative of reproductive behavior, while the length of the separate determines the exact climatic isolating events.

Similar concordance in genetic structure is observed in clear biogeographic boundaries like the Strait of Gibraltar and Indonesia. Gibraltar in particular is a well-studied Atlantic boundary, and gateway to the even more isolated Mediterranean Sea.

Several pelagic species have wide ranging individuals shown to spend time in both the Mediterranean and GOM (Bremer et al 2005; Boustany et al 2008). These cosmopolitan species are discussed in more detail below.

For sessile organisms boundaries may be dictated more by currents than geographic features (Wares et al 2001). The convergence of coastal currents can create steep temperature gradients, impermeable to the larvae of intolerant species. At Point Conception, CA this is observed as an excess of southward migration events and a prevalence of species with their northern range limit at this physical oceanographic boundary (Wares et al 2001).

Well before we could sequence the genome, theoreticians were using Mendelian genetics to qualitatively model gene flow in divided populations. Sten Wahlund described expected reduced heterozygote frequency among genetically divided populations in 1928 (Grosberg and Cunningham 2001).

Then in 1951, Sewall Wright described the theoretical result of random genetic drift in divided populations (Neigel 2001). These models were coarse, but made interesting assertions about the dispersal of individuals and the resulting population connectivity. The initial theories have expanded, and collectively help us understand the genetic structure of marine and terrestrial populations.

Genetic structure is a function of contemporary and prehistoric ecology, behavior, oceanography, climate and tectonics.

To fully characterize genetic structure it is necessary to examine gene divergence on multiple time scales. These factors are acting directly on the dispersal of individuals. The difficulty of tracking individuals (especially in marine habitats) proved to be a hurdle to those wishing to link theoretical studies to in situ transport and migration. It is nearly impossible to reach estimates of pelagic larvae abundance due to the patchiness.

Furthermore, the presence of eggs and larvae in the water column indicates the range of advection, but does not guarantee recruitment at far field sites (Boustany et al 2008).

Tagging studies of adults, with recapture and electronic tags, has been successful at demonstrating long-term site fidelity in reef fish like red snapper (Garber et al 2004) and cosmopolitan ranges in epi-pelagic predators like tuna (Bremer et al 2005), but are limited in application by small sample sizes and fish of uncertain origin.

The inability to test early genetic theory with conventional field methods necessitated the establishment of genetic methods applicable in all systems.

The development of allozyme electrophoresis in the 1960s first granted the capability to look at the genetic structure of natural populations (Grosberg and Cunningham 2001). Alleles code dissimilar enzymatic proteins called allozymes, which identify individual organisms of shared ancestry.

Allozyme analysis is limited to interpreting coding segments responsible for protein generation, so does not show minor pair mismatches.

Nucleotide sequencing allowed marker database building through the 1980s (Grosberg and Cunningham 2001), and for the most part replaced allozyme analysis as the dominate methodology. The scale of inquiry made possible by the detection of base pair mismatches in DNA finally allowed a connection between the genetic branching of clades and their biogeography, and phylogeography was born.

Detecting smaller genetic differences allows greater precision, so geneticists have since targeted sequences with high mutation rates and rapid replication.

Mitochondrial DNA (mtDNA) is a rapidly evolving segment passed only via maternal transcription. In this case and allele is called a haplotype (Bowen and Avise 1990; Grosberg and Cunningham 2001). MtDNA divergence in vertebrates functions as a clock, with 0.02 (2%) corresponding to about one million years of effective genetic isolation between two populations (Bowen and Avise 1990).

Microsatellites are short, repeating sequences of base pairs with much a higher mutation rate than bulk DNA or even mtDNA [2] and are therefore meaningful on an even finer scale. The results of several studies using mtDNA and microsatellite analysis will be reviewed.

At the most basic level, genetic regimes can be divided into two groups: connected or isolated. Populations showing genetic heterogeneity over the scale of a study are composed of distinct demographic subgroups (demes), while spatially homogenized populations indicate contemporary mixing at sufficient rate to prevent structure (Grosberg and Cunningham 2001; Cunningham and Collins 1998).

Heterogeneity can be observed as a function of distance (reproductive isolation-by-distance) or within geographic areas. A common metric for genetic divergence is F_{st}, or the fixation index, introduced by Wright in 1951 (Grosberg and Cunningham 2001; Neigel 2001).

Originally intended as a measure of inbreeding, representing population division, the variable has since been operationally redefined so that it can be better approximated from allele frequency distribution, or from demographic information (Neigel 2001). Often F_{st} is used with estimates of genetic effective population size (N) or migration rate (m, per generation) to back-calculate the unknown variable (Grosberg and Cunningham 2001),

F_{st} = 1 / (1 + 4Nm)

Values F_{st} near zero indicate population homogeneity, while values near one indicate heterogeneity. The relationship between the three variables is shown below in figure 1. Increasing either m or N independently will increase F_{st}.

The value of m is usually obtained from field experiments following individuals, since N is impossible to estimate from readily available catch data. N is then an abstraction of the population’s reproductive capability useful for theoretical comparison [2]. It is always less than the actual population size (Tringali and Bert 1998), due to variable fecundity and success of individuals.

In studies utilizing mtDNA, effective population size is the number of females (N_f), usually assumed to be half of the whole population (Bowen and Avise 1990).

Nm is itself considered a single variable for the number of migrating individuals per generation. A single individual per generation (Nm = 1) is a critical point after which migration will always, eventually, overpower intrapopulation mutation (Grosberg and Cunningham 2001; [2]. This simplification assumes that the populations are immune to mutation and selection (Neigel 2001; Wright 1951).

Though this index has received criticism, it remains appropriate for comparative observations of large equilibrium populations. The time (measure in species-specific generations) necessary to reach stable allele frequency distribution is approximated by,

T = 1 / (2m + 1 / (2N))

This can be equated to a state of reciprocal monophyly, in which the phylogenetic tree for a species is congruent with the geographic distribution of the subpopulations (Grosburg and Cunningham 2001).

Populations on either side of the Florida peninsula are often managed individually because of the geographic separation, and evidence of distinguishable vital rates (Gold et al 1994) [2]. Contemporary genetic studies do not always support management paradigms, and available data should be considered by fishery policymakers.

Take for example king mackerel (Scomberomorus cavalla) and black sea bass (Centropristis striata). The sea bass show a clear divide in the genetic structure, and are appropriately managed from the standpoint of populations (Bowen and Avise 1990). King mackerel are similarly managed based on early evidence that there could be 2-3 recognized demographic groups (Broughton et al 2002). A closer investigation shows them to be genetically well mixed (Broughton et al 2002).

There is no evidence to suggest that the misapplication in this case is harmful. The difference is in, unsurprisingly, the life history of the fishes.

Black sea bass are considered two species based on evidence of little gene flow between Atlantic (C. striata striata) and Gulf subpopulations (C. striata melana) (Bowen and Avise 1990). They are estuarine dependent. Adults spawn along the coast in the spring and pelagic eggs and larvae are distributed for 3-4 days by alongshore physical transport (Bowen and Avise 1990). Eventually they are advected offshore, or washed into coastal wetlands by Ekman currents (McMillen-Jackson et al 2005). Post-larvae and juveniles occupy estuaries for the shelter of vegetation and turbidity, as well as increased food availability. Come fall, competent individuals migrate offshore. Bowen and Avise (1990) used mtDNA extracted from the heart, liver and eggs of black sea bass caught in the GOM, eastern Florida and the Carolinas to measure mean sequence divergence and genotypic diversity.

The five discovered sea bass haplotypes were clearly clustered by geographic source, and had 0.9% divergence equating to about 350,000 years or 100,000 generations since separation (Bowen and Avise 1990). Phylogenetically shallower subpopulations formed from endemic Atlantic and GOM groups some 15,000 years ago during the late pleistocene (Bowen and Avise 1990). Each region had low intra-population mtDNA diversity associated with their relatively small effective population size (Nf = 5,000).

King mackerel on the other hand are much larger (up to 1.7 meters) and highly migratory. They winter around the Florida Keys and the Yucatan, and spawn in the spring and summer on the Atlantic coast and northern Gulf of Mexico. Samples encompassing the Gulf from Veracruz, Mexico to Key West, Florida were not significantly different from those collected in Daytona Beach, Florida and Morehead City, North Carolina (Broughton et al 2002). Various grouping schemes of sampling locations also showed no statistically meaningful genetic structure. A low level of migratory exchange (3%) combined with seasonal breeding range overlap (~42%) is sufficient in this case for genetic homogeneity (Broughton et al 2002).

Animal mobility plays an important role in connectivity, but predictions based on it alone may miss life history factors. Coastal species are expected to be more biogeographically divided (McMillen-Jackson et al 2005), as with sea bass versus king mackerel, but even anadromous fishes like menhaden and sturgeon can show inter-ocean connectivity (Bowen and Avise 1990). Menhaden were examined over the same geographic range as sea bass. Bowen and Avise (1990) observed an exclusive Atlantic clade, but also found closely related genotypes represented in both the Atlantic and GOM. In contrast to the sea bass, menhaden have high intra-population genotype diversity and sequence divergence, suggesting large effective population sizes for both the GOM (Nf = 250,000) and Atlantic (Nf = 800,000).

The scale of connectivity in pelagic species can be quite immense. Like king mackerel, tarpon (Megalops atlanticus) are a highly mobile pelagic fish with coastal tendencies and a pelagic larval duration (PLD) of up to 3 months which can carry them up to 250 km offshore (McMillen-Jackson et al 2005). There are no western Atlantic barriers to their dispersal, yet distance alone cleaves longitudinal population exchange (McMillen-Jackson et al 2005). Spanish mackerel (Scomberomorus maculatus) show a similar phylogeography (Buonaccorsi et al 2001), but greater amberjack (Seriola dumerili) which are also strong pelagic swimmers do show division between Gulf and western Atlantic stocks (Bagley et al 1999).

Latitudinal variation in temperature and food availability set limits on north and south range extremes, since continents provide continuous or patchy potential habitat. Fish with thermoregulatory capabilities like tuna and swordfish can bypass even these restrictions, and show high connectivity at great distance. In some studies of tuna, populations appear at the global scale, although interpretation of genetic structure in Scombroid fishes (tuna and mackerel) is confounded by the fact that they have undergone recent speciation and population founding in the Atlantic and have likely not yet reached an equilibrium state (Boustany et al 2008). Expanding populations are easy to identify by a hub shaped phylogenetic tree, with successively more differentiated subpopulations radiating from a central founding population [1].

Migratory fish pose problems for establishing marine management scale. Seasonal fisheries in one region can harvest disproportionately from a common resource, and even a non-target species may become bycatch. Atlantic bluefin tuna (Thunnus thynnus) migrations are targeted by numerous fisheries in the Atlantic and Mediterranean (Boustany et al 2008). One of the great challenges of studying this species has been to determine whether distinct spawning grounds exist. Bluefin are managed as two populations (GOM/Bahamas and Mediterranean), but electronic tagging has shown individuals moving between eastern and western Atlantic (Boustany et al 2008). Various studies have found disparate evidence for both transatlantic homo- and heterogeneity, and authors are careful to undercut their findings with stipulations regarding accuracy and interpretability. Tuna populations may even be subdivided between the eastern and western Mediterranean, which is a much smaller scale than might be expected based on life history (Boustany et al 2008), but matches the scale of hypothetical (though often unfounded) partitioning in the GOM.

A comparative study of bluefin and swordfish (Xiphias gladius) showed that although they share many characteristics, tuna are much less structured (F_{st} = 0.002) compared to swordfish (F_{st} = 0.091) [1]. Both experienced vicariance due to exposure of the Isthmus of Panama during the Pliocene, but the exact timeline is difficult to establish due to the high mutation rates observed in thermoregulatory fishes [1]. The driving factor in their modern structure appears to be that Xiphias are solitary and seem to show stronger philopatric spawning ground association [1].

Demersal reef fishes (snapper, drum, grouper and squirrelfish) are commercially valuable Gulf species, and have been the subject of a great deal of debate over the exact genetic structure of stocks (Garber et al 2004).

As adults, snapper develop a relatively small home range (Bagley et al 1999), quite in opposition to the previously discussed pelagic migrants. The oceanographic dispersal of snapper eggs and larvae lasts 35 to 47 days (Bagley et al 1999), and is the dominant factor controlling genetic structure, as is the case for most benthic species (Wares et al 2001). Red snapper (Lutjanus campechanus) and red grouper (Epinephelus morio) have been shown to have a panmictic GOM population (Bagley et al 1999; Garber et al 2004), suggesting that passive dispersal is sufficient to homogenize on a basin scale.

It follows that species with greater PLD would exhibit greater connectivity since propagules would have more time to be advected to distant demes. The high velocity currents off south Florida combined with long PLD could act to connect GOM and Atlantic coast populations.

Bagley et al (1999) estimated F_{st} = 0.001-0.002 for vermillion snapper (Rhomboplites aurorubens) in the southeast USA, supporting the conclusion that the species did not have a clear genetic structure (Nm > 100), and showed no influence by the Florida peninsula. Although clear geographic delineations do not exists in this case, the authors caution that structuring may exist. Dynamic physical conditions influencing dispersal and the time of spawning may set up complex hereditary patterns which could be unraveled with more careful sampling (Bagley et al 1999).

Just as adult dispersal in pelagic species is not always a good predictor of genetic structure, other factors are at work in the propagation of demersal fishes. Bowen et al (2006) studied the genetic structure of two species of squirrelfish found at GOM reef sites. Both have early pelagic life stages lasting nearly two months, but Holocentrus ascensionis can manifest an additional pre-juvenile stage sustaining PLD by nearly two weeks. The authors hypothesized a homogeneous population due to longer PLD. Instead they found that Myripritis jacobus (F_{st} = 0.008) in fact was less structured than H. ascensionis (F_{st} = 0.091) (Bowen et al 2006).

Pelagic stages may not disperse very far despite having the time to do so, meaning that either there is hydrodynamic retention due to slow or oscillatory currents or that larvae are using chemosensory abilities to home to natal reef sites. This has serious implications for marine stock management. The idea of seed populations whose young will disperse to damaged areas is severely undermined if the replenishing recruits are swimming back home.

Population subdivision is of non-trivial ecological importance. The placement of marine protected areas (MPA) and the sourcing of individuals for stock-enhancement for these pelagic and reef fish requires knowledge of their connectivity. Genetic isolation increases the risk of extinction for specific populations, but minimizes risk for the species as a whole. Though this does not hold true for species whose subpopulations experience equally distributed fishing pressure, such as tuna (Boustany et al 2008).

By placing MPAs to sustain a source of migrants in homogenous populations, fishermen can continue to harvest at far field sites with less risk to the species as a whole. In extremely divided populations with limited migration events unsustainable harvesting will drive the stock to zero. Protecting dispersal corridors is then a major priority, as well as maintaining numbers through informed stock-enhancement.

Effective population size is an important consideration in stock management, especially in species where stock-enhancment is being utilized or considered (Tringali and Bert 1998). The proliferation of gene anomalies to wild fish by hatchery-raised salmon is a major concern of managers in the Pacific northwest (Tringali and Bert 1998), and the introduction of genetically homogeneous seed populations can lead to poor fitness or high disease rate in the long term. Homogeneity is particularly insidious in populations which repeatedly cycle through stock-and-crash scenarios due to continued harvesting (Tingali and Bert 1998). Even when actual population size is constant, genetic effective population size may vary over time. The value of N for gendered fishes with environmental sex determination could decrease with rising sea surface temperatures due to wandering sex ratios (Bagley et al 1999). The same effect has been seen in sex-selecting fisheries (Bagley et al 1999).

When considering the role of biogeographic barriers in management decision making it is vital to utilize genetic methods to at least be able to estimate contemporary rates of exchange, if not robust models of metapopulation connectivity. Studies of pelagic fish have shown that long distance migration and seasonal overlap of adults is sufficient to maintain homogeneity between the GOM and western Atlantic, but not necessarily between eastern and western Atlantic. Active migrators like tuna can span these distances, but exhibit spawning site fidelity that results in greater partitioning than may be expected. Coastal fishes will have greater population structure due to linear dispersal and hydrodynamic retention of eggs and larvae at estuaries. However, even anadromous fish which may live in freshwater for years can show connectivity between GOM and Atlantic. Sedentary demersal fishes like snapper are subject to pelagic egg and larval advection, though the scale of connectivity does not appear proportional to PLD.

Clearly much has been accomplished in this field, especially in well-studied regions like the GOM. The metrics used are necessarily vague, due to inconsistencies in results from varying sample size and methods.

The trend is to reach finer temporal and spatial scales, and to use multiple techniques to improve statistical confidence. As the discipline evolves further we should see quantitative estimates of subpopulation age (and be able to date to specific geological events), as well as the ability to meaningfully compare species.


Tringali MD, TM Bert. 1998. Risk to genetic effective population size should be an important consideration in fish stock-enhancement programs. Bulletin of Marine Science 62(2):641–659.
Broughton RE, LB Stewart, JR Gold. 2002. Microsatellite variation suggests substantial gene flow between king mackerel (Scomberomorus cavalla) in the western Atlantic Ocean and Gulf of Mexico. Fisheries Research 54:305–316.
Boustany AM, CA Reeb, BA Block. 2008. Mitochondrial DNA and electronic tracking reveal population structure of Atlantic Bluefin tuna (Thunnus thynnus). Marine Biology 156:13–24.
Buonaccorsi VP, E Starkey, JE Graves. 2001. Mitochondrial and nuclear DNA analysis of population subdivision among young-of-the-year Spanish mackerel (Scomberomorus maculatus) from the western Atlantic and Gulf of Mexico. Marine Biology 138:37–45.
Bowen BW, JC Avise. 1990. Genetic structure of Atlantic and Gulf of Mexico populations of sea bass, menhaden, and sturgeon: influence of zoogeographic factors and life-history patterns. Marine Biology 107:371–381.
Garber AF, MD Tringali, KC Stuck. 2004. Population structure and variation in red snapper (Lutjanus campechanus) from the Gulf of Mexico and Atlantic Coast of Florida as determined from mitochondrial DNA control regional sequence. Marine Biotechnology 6:175–185.
Alvarado-Bremer JR, J Vinas, J Mejuto, B Ely, C Pla. 2005. Comparative phylogeography of Atlantic Bluefin tuna and swordfish: the combined effects of vicariance, secondary contact, introgression, and population expansion on the regional phylogenies of two highly migratory pelagic fishes. Molecular Phylogenetics and Evolution 36:169–187.
Bagley MJ, DG Linquist, JB Geller. 1999. Microsatellite variation, effective population size, and population genetic structure of vermilion snapper, Rhomboplites auroruben, off the southeastern USA. Marine Biology 134:609–620.
Bowen BW, AL Bass, A Muss, J Carlin, DR Robertson. 2006. Phylogeography of two Atlantic squirrelfishes (Family Holocentridae): exploring links between pelagic larval duration and population connectivity. Marine Biology 149:899–913.
Cunningham CW, TM Collins. 1998. Beyond area relationship: extinction and recolonization in molecular marine biogeography. Molecular Approaches to Ecology and Evolution
Gold JR, LR Richardson, C Furman, F Sun. 1994. Mitochondrial DNA diversity and population structure in marine fish species from the Gulf of Mexico. Canadian Journal of Fisheries and Aquatic Sciences 51(S.1):205–214.
Grosberg R, CW Cunningham. 2001. Genetic structure in the sea: from populations to communities. Marine Community Ecology :61–84.
McMillen-Jackson AL, TM Bert, H Cruz-Lopez, S Seyoum, T Orsoy, RE Crabtree. 2005. Molecular genetic variation in tarpon (Megalops atlanticus Valenciennes) in the northern Atlantic Ocean. Marine Biology 146:253–261.
Neigel JE. 2002. Is Fst obsolete?. Conservation Genetics 3:167–173.
Wares JP, SD Gaines, CW Cunningham. 2001. A comparative study of asymmetric migration events across a marine biogeographic boundary. Evolution 55(2):295–306.

Accountable prosperity in autonomous seas ⚡ 2018-2021

We value you, regardless of technical training, domain knowledge, or identity. We invite questions, listen more than we speak, and embody a “yes/and” over a “no/but” mentality.

When you arrive we track your user agent, actions, and location. These data are not transmitted. On our end, we manage risk with backups, monitoring, and encryption.