What can population genetics tell us about dispersal and biogeographic history of coral-reef fishes?

Abstract Coral-reef fishes, like many other marine organisms, generally possess a benthic adult stage and pelagic larval stage. What can population genetics studies tell us about the demographic, evolutionary and biogeographic consequences of this life cycle? Ten studies of geographical patterns of intraspecific genetic differentiation in reef fishes have been published. These studies have included 2t > species/species complexes (14 in the family Pomacentridae, the remaining 12 in 9 different families) and have been about equally divided between the tropical Pacific and the tropical western Atlantic. A survey of these studies shows the following: (i) the existence of the pelagic larval stage appears to have led to high levels of gene flow even among populations separated by thousands of kilometres of open ocean; (ii) an apparent pattern of increased gene flow among populations connected by intermediate 'stepping stones’; (iii) very tentative evidence for a relationship between length of pelagic larval life and gene flow; (iv) no clear relationship between egg type (pelagic rs non-pelagic) and gene flow; and (v) suggestive evidence that damselfishes (family Pomacentridae) may have more restricted dispersal (less gene flow) than other reef fishes. The application of current and future molecular tools has the strong potential to clarify some of these relationships, particularly by using relatively neutral genetic markers. Additionally, discoveries of DNA markers having very high rates of mutation may allow tracking of demographically relevant levels of larval dispersal. Molecular tools are becoming especially valuable in uncovering the biogeographic and phylogenetic history of reef fishes. The one molecular study to date has suggested that at least some speciation events may have occurred during the climate changes and sea-level regressions associated with Pleistocene glacial episodes. Molecular tools need to be used to further explore the means by which high species diversity can be generated in the face of the apparently high gene flow observed in most coral-reef fishes.     

Fine scale dispersal in Banggai Cardinalfish, Pterapogon kauderni, a coral reef species lacking a pelagic larval phase

Dispersal in marine species results from complex interactions between biotic and abiotic factors. Importantly, the pelagic larval phase of most marine species adds a significant degree of complexity. Therefore, a growing body of work is focusing on those rare species that lack a pelagic larval phase (usually brooding species). For such species, large-scale gene flow has been shown to be very low, thus following the expectation of a relationship between realized dispersal and pelagic larval duration. Yet, little is known about the dispersal of those species at very small geographic scales. In this study, we focused on the Banggai Cardinalfish, Pterapogon kauderni, a mouthbrooding species that lacks a pelagic larval phase. Based on previously identified microsatellites, we scored 12 populations around the southern island of Bangkulu, in the Banggai Archipelago, Indonesia. While only 60 km in perimeter, we found that this island harbors very distinct populations of P. kauderni. Indeed, assignment tests self-assigned 10 out of those 12 populations. These results mirror the very high level of self-assignment at the level of the entire archipelago, where, out of 13 populations, 70% of the individuals were reassigned to their source population. Therefore, our data show consistency between small and large-scale dispersal. In addition, in light of the recent expansion in the harvesting of this species for the pet trade, our data have important conservation implications.     

Ocean circulation model predicts high genetic structure observed in a long‐lived pelagic developer

Understanding the movement of genes and individuals across marine seascapes is a long‐standing challenge in marine ecology and can inform our understanding of local adaptation, the persistence and movement of populations, and the spatial scale of effective management. Patterns of gene flow in the ocean are often inferred based on population genetic analyses coupled with knowledge of species' dispersive life histories. However, genetic structure is the result of time‐integrated processes and may not capture present‐day connectivity between populations. Here, we use a high‐resolution oceanographic circulation model to predict larval dispersal along the complex coastline of western Canada that includes the transition between two well‐studied zoogeographic provinces. We simulate dispersal in a benthic sea star with a 6–10 week pelagic larval phase and test predictions of this model against previously observed genetic structure including a strong phylogeographic break within the zoogeographical transition zone. We also test predictions with new genetic sampling in a site within the phylogeographic break. We find that the coupled genetic and circulation model predicts the high degree of genetic structure observed in this species, despite its long pelagic duration. High genetic structure on this complex coastline can thus be explained through ocean circulation patterns, which tend to retain passive larvae within 20–50 km of their parents, suggesting a necessity for close‐knit design of Marine Protected Area networks.     

Phylogeography and demography of sympatric sister surfperch species, Embiotoca jacksoni and E. lateralis along the California coast: historical versus ecological factors

With 18 closely related endemic species that radiated in a diversity of ecological niches, the California surfperches (Embiotocidae) species flock is a good candidate for the study of sympatric speciation. Resource partitioning has been suggested as an important driving force in the radiation of the surfperch family. Within the family, two congeneric sister species, Embiotoca jacksoni and E. lateralis, are known to compete strongly for a preferred single food resource and may be used as a model of ecological interactions for the family. Along the California coast, the distribution of the two species differs. Embiotoca jacksoni has a continuous range, whereas E. lateralis shows a disjunction with a distribution gap in the Southern California Bight. Two hypotheses may explain this disjunct distribution. Ecological competition may have displaced E. lateralis in favor of E. jacksoni. Alternatively, a common vicariant event may have separated the species into northern and southern populations, followed by secondary contact in E. jacksoni but not in E. lateralis. The two hypotheses predict different phylogeographic and demographic signatures. Using a combined phylogeographic and coalescent approach based on mitochondrial control region data, we show that vicariance can only account for a portion of the observed divergences. Our results are compatible with a significant role played by ecological competition in the southern range of the species.     

Influence of habitat discontinuity, geographical distance, and oceanography on fine-scale population genetic structure of copper rockfish (Sebastes caurinus)

The copper rockfish is a benthic, nonmigratory, temperate rocky reef marine species with pelagic larvae and juveniles. A previous range-wide study of the population-genetic structure of copper rockfish revealed a pattern consistent with isolation-by-distance. This could arise from an intrinsically limited dispersal capability in the species or from regularly-spaced extrinsic barriers that restrict gene flow (offshore jets that advect larvae offshore and/or habitat patchiness). Tissue samples were collected along the West Coast of the contiguous USA between Neah Bay, WA and San Diego, CA, with dense sampling along Oregon. At the whole-coast scale (approximately 2200 km), significant population subdivision (F(ST) = 0.0042), and a significant correlation between genetic and geographical distance were observed based on 11 microsatellite DNA loci. Population divergence was also significant among Oregon collections (approximately 450 km, F(ST) = 0.001). Hierarchical amova identified a weak but significant 130-km habitat break as a possible barrier to gene flow within Oregon, across which we estimated that dispersal (N(e)m) is half that of the coast-wide average. However, individual-based Bayesian analyses failed to identify more than a single population along the Oregon coast. In addition, no correlation between pairwise population genetic and geographical distances was detected at this scale. The offshore jet at Cape Blanco was not a significant barrier to gene flow in this species. These findings are consistent with low larval dispersal distances calculated in previous studies on this species, support a mesoscale dispersal model, and highlight the importance of continuity of habitat and adult population size in maintaining gene flow.     

Does life history predict past and current connectivity for rocky intertidal invertebrates across a marine biogeographic barrier?

The southeast Australian coast potentially includes a complex biogeographic barrier, largely lacking exposed rocky shore that may limit the dispersal of rocky intertidal taxa and contribute to the maintenance of two biogeographic regions. Surprisingly, within the 300-km barrier region, several species considered exposed rocky shore specialists occurred within sheltered sites. We analysed COI sequence variation for 10 rocky intertidal invertebrate species, with a range of life histories, to test the hypotheses that larval type and habitat specificity are strong predictors of gene flow between biogeographic regions. Our data revealed that the southeast corner of Australia includes a strong barrier to gene flow for six of eight species with planktonic larvae, and a coalescence analysis of sequence differentiation (IM model) suggests that a barrier has existed since the Pleistocene. In contrast, two direct developers were not affected by the barrier. Our comparative approach and data from earlier studies (reviewed here) do not support the hypothesis that larval type predicts gene flow across this barrier, instead we found that the ability to utilize sheltered habitat provides a clearer explanation of the phylogeographic break. Indeed, the species that displayed little or no evidence of a phylogeographic break across the barrier each displayed unexpectedly relaxed habitat specificity.     

Testing comparative phylogeographic models of marine vicariance and dispersal using a hierarchical Bayesian approach

Background  

Marine allopatric speciation is an enigma because pelagic larval dispersal can potentially connect disjunct populations thereby preventing reproductive and morphological divergence. Here we present a new hierarchical approximate Bayesian computation model (HABC) that tests two hypotheses of marine allopatric speciation: 1.) "soft vicariance", where a speciation involves fragmentation of a large widespread ancestral species range that was previously connected by long distance gene flow; and 2.) peripatric colonization, where speciations in peripheral archipelagos emerge from sweepstakes colonizations from central source regions. The HABC approach analyzes all the phylogeographic datasets at once in order to make across taxon-pair inferences about biogeographic processes while explicitly allowing for uncertainty in the demographic differences within each taxon-pair. Our method uses comparative phylogeographic data that consists of single locus mtDNA sequences from multiple co-distributed taxa containing pairs of central and peripheral populations. We use the method on two comparative phylogeographic data sets consisting of cowrie gastropod endemics co-distributed in the Hawaiian (11 taxon-pairs) and Marquesan archipelagos (7 taxon-pairs).     

Vicariance and dispersal across Baja California in disjunct marine fish populations

Abstract.— Population disjunctions, as a first step toward complete allopatry, present an interesting situation to study incipient speciation. The geological formation of the Baja California Peninsula currently divides 19 species of fish into disjunct populations that are found on its Pacific Coast and in the northern part of the Gulf of California (also called the Sea of Cortez), but are absent from the Cape (Cabo San Lucas) region. We studied the genetic makeup of disjunct populations for 12 of these 19 fish species. Phylogeographic patterns for the 12 species can be separated into two major classes: a first group (eight species) showed reciprocal monophyly and high genetic divergence between disjunct populations. A second group (four species) displayed what appeared to be panmictic populations. Population structure between Pacific Coast populations, across the Punta Eugenia biogeographic boundary, was also evaluated. While dispersal potential (inferred by pelagic larval duration) was a poor predictor of population structure between Gulf of California and Pacific populations, we found that population genetic subdivision along the Pacific Coast at Punta Eugenia was always positively correlated with differentiation between Pacific and Gulf of California populations. Vicariant events, ongoing gene flow, and ecological characteristics played essential roles in shaping the population structures observed in this study.     

Nonsignificant isolation by distance implies limited dispersal

Understanding the scale of dispersal is an important consideration in the conservation and management of many species. However, in species in which the high‐dispersal stage is characterized by tiny gametes or offspring, it may be difficult to estimate dispersal directly. This is the case for many marine species, whose pelagic larvae are dispersed by ocean currents by several days or weeks before beginning a benthic, more sedentary, adult stage. As consequence of the high‐dispersal larval stage, many marine species have low genetic structure on large spatial scales (Waples 1998 ; Hellberg 2007 ). Despite the high capacity for dispersal, some tagging studies have found that a surprising number of larvae recruit into the population they were released from (self‐recruitment). However, estimates of self‐recruitment are not informative about mean dispersal between subpopulations. To what extent are limited dispersal estimates from tagging studies compatible with high potential for dispersal and low genetic structure? In this issue, a study on five species of coral reef fish used isolation by distance (IBD) between individuals to estimate mean dispersal distances (Puebla et al. 2012 ). They found that mean dispersal was unexpectedly small (<50 km), given relatively low IBD slopes and long pelagic durations. This study demonstrates how low genetic structure is compatible with limited dispersal in marine species. A comprehensive understanding of dispersal in marine species will involve integrating methods that estimate dispersal over different spatial and temporal scales. Genomic data may increase power to resolve these issues but must be applied carefully to this question.     

Comparative population genetics of seven notothenioid fish species reveals high levels of gene flow along ocean currents in the southern Scotia Arc, Antarctica

The Antarctic fish fauna is characterized by high endemism and low species diversity with one perciform suborder, the Notothenioidei, dominating the whole species assemblage on the shelves and slopes. Notothenioids diversified in situ through adaptive radiation and show a variety of life history strategies as adults ranging from benthic to pelagic modes. Their larval development is unusually long, lasting from a few months to more than a year, and generally includes a pelagic larval stage. Therefore, the advection of eggs and larvae with ocean currents is a key factor modulating population connectivity. Here, we compare the genetic population structures and gene flow of seven ecologically distinct notothenioid species of the southern Scotia Arc based on nuclear microsatellites and mitochondrial DNA sequences (D-loop/cytochrome b). The seven species belong to the families Nototheniidae (Gobionotothen gibberifrons, Lepidonotothen squamifrons, Trematomus eulepidotus, T. newnesi) and Channichthyidae (Chaenocephalus aceratus, Champsocephalus gunnari, Chionodraco rastrospinosus). Our results show low-population differentiation and high gene flow for all investigated species independent of their adult life history strategies. In addition, gene flow is primarily in congruence with the prevailing ocean current system, highlighting the role of larval dispersal in population structuring of notothenioids.     

Coalescent and biophysical models of stepping‐stone gene flow in neritid snails

Marine species in the Indo‐Pacific have ranges that can span thousands of kilometres, yet studies increasingly suggest that mean larval dispersal distances are less than historically assumed. Gene flow across these ranges must therefore rely to some extent on larval dispersal among intermediate ‘stepping‐stone’ populations in combination with long‐distance dispersal far beyond the mean of the dispersal kernel. We evaluate the strength of stepping‐stone dynamics by employing a spatially explicit biophysical model of larval dispersal in the tropical Pacific to construct hypotheses for dispersal pathways. We evaluate these hypotheses with coalescent models of gene flow among high‐island archipelagos in four neritid gastropod species. Two of the species live in the marine intertidal, while the other two are amphidromous, living in fresh water but retaining pelagic dispersal. Dispersal pathways predicted by the biophysical model were strongly favoured in 16 of 18 tests against alternate hypotheses. In regions where connectivity among high‐island archipelagos was predicted as direct, there was no difference in gene flow between marine and amphidromous species. In regions where connectivity was predicted through stepping‐stone atolls only accessible to marine species, gene flow estimates between high‐island archipelagos were significantly higher in marine species. Moreover, one of the marine species showed a significant pattern of isolation by distance consistent with stepping‐stone dynamics. While our results support stepping‐stone dynamics in Indo‐Pacific species, we also see evidence for nonequilibrium processes such as range expansions or rare long‐distance dispersal events. This study couples population genetic and biophysical models to help to shed light on larval dispersal pathways.     

Relative impacts of adult movement, larval dispersal and harvester movement on the effectiveness of reserve networks

Movement of individuals is a critical factor determining the effectiveness of reserve networks. Marine reserves have historically been used for the management of species that are sedentary as adults, and, therefore, larval dispersal has been a major focus of marine-reserve research. The push to use marine reserves for managing pelagic and demersal species poses significant questions regarding their utility for highly-mobile species. Here, a simple conceptual metapopulation model is developed to provide a rigorous comparison of the functioning of reserve networks for populations with different admixtures of larval dispersal and adult movement in a home range. We find that adult movement produces significantly lower persistence than larval dispersal, all other factors being equal. Furthermore, redistribution of harvest effort previously in reserves to remaining fished areas ('fishery squeeze') and fishing along reserve borders ('fishing-the-line') considerably reduce persistence and harvests for populations mobile as adults, while they only marginally changes results for populations with dispersing larvae. Our results also indicate that adult home-range movement and larval dispersal are not simply additive processes, but rather that populations possessing both modes of movement have lower persistence than equivalent populations having the same amount of 'total movement' (sum of larval and adult movement spatial scales) in either larval dispersal or adult movement alone.     

DISCORDANT DISTRIBUTION OF POPULATIONS AND GENETIC VARIATION IN A SEA STAR WITH HIGH DISPERSAL POTENTIAL

Patiria miniata, a broadcast‐spawning sea star species with high dispersal potential, has a geographic range in the intertidal zone of the northeast Pacific Ocean from Alaska to California that is characterized by a large range gap in Washington and Oregon. We analyzed spatial genetic variation across the P. miniata range using multilocus sequence data (mtDNA, nuclear introns) and multilocus genotype data (microsatellites). We found a strong phylogeographic break at Queen Charlotte Sound in British Columbia that was not in the location predicted by the geographical distribution of the populations. However, this population genetic discontinuity does correspond to previously described phylogeographic breaks in other species. Northern populations from Alaska and Haida Gwaii were strongly differentiated from all southern populations from Vancouver Island and California. Populations from Vancouver Island and California were undifferentiated with evidence of high gene flow or very recent separation across the range disjunction between them. The surprising and discordant spatial distribution of populations and alleles suggests that historical vicariance (possibly caused by glaciations) and contemporary dispersal barriers (possibly caused by oceanographic conditions) both shape population genetic structure in this species.     

High connectivity among habitats precludes the relationship between dispersal and range size in tropical reef fishes

The hypothesis that pelagic larval duration (PLD) influences range size in marine species with a benthic adult stage and a pelagic larval period is intuitively attractive; yet, studies conducted to date have failed to support it. A possibility for the lack of a relationship between PLD and range size may stem from the failure of past studies to account for the effect of species evolutionary ages, which may add to the dispersal capabilities of species. However, if dispersal over ecological (i.e. PLD) and across evolutionary (i.e. species evolutionary age) time scales continues to show no effect on range size then an outstanding question is why? Here we collected data on PLD, evolutionary ages and range sizes of seven tropical fish families (five families were reef‐associated and two have dwell demersal habitats) to explore the independent and interactive effects of PLD and evolutionary age on range size. Separate analyses on each family showed that even after controlling for evolutionary age, PLD has an insignificant or a very small effect on range size. To shed light on why dispersal has such a limited effect on range size, we developed a global ocean circulation model to quantify the connectivity among tropical reefs relative to the potential dispersal conferred by PLD. We found that although there are several areas of great isolation in the tropical oceans, most reef habitats are within the reach of most species given their PLDs. These results suggest that the lack of habitat isolation can potentially render the constraining effect of dispersal on range size insignificant and explain why dispersal does not relate to range size in reef fishes.     

REPRODUCTIVE ECOLOGY AND BIOGEOGRAPHY OF INDO-WEST PACIFIC ANGELFISHES (PISCES: POMACANTHIDAE)

Analysis of the relationships between duration of the pelagic larval stage (as indicated by otolith microstructure), adult size, and the extent of geographic distribution for Indo-West Pacific angelfishes (Pomacanthidae) indicates that neither adult size nor larval duration significantly predicts extent of distribution, either individually or jointly in a multiple regression. These results are robust for both the family as a whole and the genus best represented in our data (Centropyge). If larval duration and adult size do have an effect, it is only jointly and at the genus level. However, larval duration and adult size do correlate significantly and negatively with one another. The operational factor seems to be egg size, which correlates positively with adult size, and negatively with duration of the pelagic larval stage. Similar correlations are evident in both marine invertebrates and at least some other coral-reef fishes, suggesting they are of widespread significance. The limited ability of either reproductive parameter to predict extent of species distribution indicates, first, that even in a group as morphologically conservative as the Indo-West Pacific pomacanthids, neither a two-fold difference between species in larval duration nor a two order of magnitude difference in female fecundity markedly affects extant distributions; and secondly, that either undescribed biological factors or historical constraints are of paramount importance. Available evidence suggests that dispersal abilities of most coral reef fishes, in fact, may be limited, despite the nearly universal occurrence of a pelagic stage in development.     

Climate-driven genetic divergence of limpets with different life histories across a southeast African marine biogeographic disjunction: different processes, same outcome

Genetic divergence among populations of marine broadcast spawners in the absence of past geological barriers presents an intriguing challenge to understanding speciation in the sea. To determine how differences in life history affect genetic divergence and demographic histories across incomplete dispersal barriers, we conducted a comparative phylogeographic study of three intertidal limpets (Siphonaria spp.) represented on either side of a biogeographic disjunction separating tropical and subtropical marine provinces in southeastern Africa. Using a combination of mitochondrial and nuclear sequence data, we identified two distinct evolutionary lineages each in both Siphonaria concinna (a planktonic disperser) and S. nigerrima (a direct developer), and panmixia in a second planktonic disperser, S. capensis. Although phylogeographic breaks were present in two species, how these became established differed depending on their life histories. In the direct developer, lack of gene flow following divergence, and demographic expansion from a small initial size in the species' subtropical population, point to a single colonisation event. In contrast, the evolutionary lineages of the planktonic disperser split into two genetic lineages with much larger initial population sizes and southward gene flow continued at least periodically, indicating that divergence in this species may have been driven by a combination of reduced larval dispersal and divergent selection. These findings help explain why the presence or absence of phylogeographic breaks often appears to be independent of species' dispersal potential.     

Population genetic structure and gene flow patterns between populations of the Antarctic icefish Chionodraco rastrospinosus

Aim A lack of genetic structure is predicted for Antarctic fish due to the duration of pelagic larval stages and the strength of the currents in the Southern Ocean, particularly the Antarctic Circumpolar Current. In this study we explored the population structure of the ocellated icefish, Chionodraco rastrospinosus, by means of analysing a total of 394 individuals collected at four geographical areas off the Antarctic Peninsula in the period 1996–2006. Location Elephant Island, southern South Shetlands, Joinville Island and South Orkneys in the Southern Ocean. Methods The spatio‐temporal genetic structure of Chionodraco rastrospinosus was explored using seven microsatellite loci. Existence and direction of gene flow across sampling locations were investigated using the isolation‐by‐migration procedure. Results Microsatellite data showed a lack of genetic structuring in the area studied, with no differences found at both the geographical or temporal level, and an eastward unidirectional gene flow among sites. This suggested a lack of genetic barriers for this species, attributable to larval dispersal following the Antarctic Circumpolar Current, which fits well with the predicted pattern for Antarctic fish. Re‐examination of genetic data of the closely related icefish Chaenocephalus aceratus, with similar larval duration but displaying genetically structured populations, indicated a weak but significant bidirectional gene flow. Main conclusions Our results point to a relationship that is more complex than expected between potential for dispersal and realized gene flow in the marine environment. In addition to ocean circulation and larval dispersal, other major life‐history traits might be driving connectivity, particularly larval retention.     

Ontogeny of behaviour in larvae of marine demersal fishes

The development of behaviours that are relevant to larval dispersal of marine, demersal fishes is poorly understood. This review focuses on recent work that attempts to quantify the development of swimming, orientation, vertical distribution and sensory abilities. These behaviours are developed enough to influence dispersal outcomes during most of the pelagic larval stage. Larvae swim in the ocean at speeds similar to the currents found in many locations and at 3–15 body lengths per second (BL s⁻¹), although, based on laboratory measurements, species from cold environments swim slower than those from warm environments. At least in warm-water species, larvae swim in an inertial hydrodynamic environment for most of their pelagic period. Unfed swimming endurance is >10 km from about 8–10 mm, and reaches more than 50 km before settlement in several species. Larval fishes are efficient swimmers. In most species, a large majority of larvae have orientated swimming in the ocean, but the precision of orientation does not improve with growth. Swimming direction of the larvae frequently changes ontogenetically. Vertical distribution changes ontogenetically in most species, and both ontogenetic ascents and descents are found. Development of schooling is poorly understood, but it may influence speed, orientation and vertical distribution. Sensory abilities (hearing, olfaction, vision) form early, are well developed and are able to detect cues relevant to orientation for most of the pelagic larval stage. All this indicates that the passive portion of the pelagic larval duration will be short, at least in most warm-water species, and that behaviour must be taken into account when considering dispersal, and in particular in dispersal models. Although quantitative information on the ontogeny of some behaviours is available for a relatively small number of species, more research in this field is required, especially on species from colder waters.     

Larval spatial distributions and other early life-history characteristics predict genetic differentiation in eastern Pacific blennioid fishes

In marine organisms, a pelagic larval stage increases the opportunities for long-distance dispersal and is often associated with little genetic differentiation over large geographical distances. Here we test the hypothesis that early life-history characteristics, including larval spatial distributions, affect the rates of dispersal and, therefore, the levels of genetic partitioning among three Gulf of California reef fishes: Axoclinus nigricaudus, Malacoctenus hubbsi and Ophioblennius steindachneri. These three blennioid fishes have markedly different early life histories: A. nigricaudus has a short larval duration (18 days) and develops inshore, M. hubbsi has an intermediate larval duration (24 days) and most individuals develop inshore and O. steindachneri has a long larval life (50 days) and disperses offshore. Estimates of genetic partitioning from mtDNA control region sequences differed greatly between these species and were in the same rank order as predicted by their early life-history characteristics (A. nigricaudus N(ST)=0.536, M. hubbsi N(ST)=0.261 and O. steindachneri N(ST)=0.000). These results indicate that larval strategies may be good predictors of population genetic structure in some marine fishes.     

The adaptive significance of larval dispersal in coral reef fishes

Synopsis Coral reef fishes almost universally disperse over relatively great distances during a pelagic larval phase. Barlow (1981) suggested that this dispersal is adaptive because adult fishes inhabit a patchy, uncertain environment. This reiterated an older idea that the random extinction of local populations necessarily favours dispersal, since ultimately all populations of non-dispersers will disappear. Whereas this view is based on adult survival, we emphasize a less frequent view that substantial larval dispersal may be adaptive when offspring experience patchy and unpredictable survival in the pelagic habitat. We do not address the question of why these animals ‘broadcast’ rather than ‘brood’, but suggest that species committed to pelagic offspring will be under selection to disperse siblings to spread the risk of failure among members of a cohort. Our arguments are supported by a heuristic computer simulation.