Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
1. Movement pathways of individuals can be shaped by heterogeneity in the dispersal environment that separates origin and destination patches. However, effects of the dispersal environment on the phenotype (or future fitness) of dispersers is poorly known; individual experiences during dispersal may have latent effects on the performance or persistence of later life-stages.
2. We evaluated such ‘legacy effects’ for dispersing reef fish larvae using (i) otolith (ear stone) microchemistry to characterize two distinct dispersal pathways and (ii) otolith microstructure to estimate ‘larval quality’ (a composite of five measured larval phenotypes). We conducted a reciprocal transplant field experiment to evaluate selective mortality after dispersal as a function of larval quality. We conducted longitudinal sampling of natural cohorts of reef fish through to adulthood to quantify shifts in the distribution of larval quality in local populations.
3. We found the quality of dispersers to be variable and determined by their experience in the larval dispersal environment. Larval quality of successful dispersers predicted their subsequent survival after dispersal in reciprocal transplant experiments. Longitudinal sampling was consistent with short-term field experiments, and revealed that survivors to adulthood were disproportionately comprised of high quality larval dispersers.
4. Overall, our results suggest that conditions in the dispersal environment shape future fitness of individuals after successful dispersal, and that this can indirectly mediate dispersal patterns and connectivity in a metapopulation.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Many factors are known to influence the dispersal process. Dispersal is often condition dependent, meaning that the decision and/or ability to disperse may depend upon the physiological condition of dispersers, local population density and/or local availability of food or refuges from predators (Ims & Hjermann 2001; Clobert et al. 2009). Once a dispersal event has begun, its success may depend upon the ability of dispersers to (i) search and locate a suitable destination; (ii) settle into the new destination; and (iii) establish permanent residency, (Stamps 2001). The ability of organisms to detect and distinguish between candidate destinations may be influenced by their prior experiences (Davis & Stamps 2004) or present physiological condition (Stamps 2006). Similarly, the ability of dispersers to settle (which often involves a substantial transition between habitats and/or a metamorphosis of body plan; Pechenik 2006) may depend upon their physiological state and/or local environmental conditions at the destination site (e.g. species interactions, abiotic conditions, etc.; Schmitt & Holbrook 1996). Lastly, the ability of dispersers to establish themselves more permanently (i.e. ‘recruit’) into the new local population may also depend upon intrinsic (i.e. physiological condition of individuals) or extrinsic (i.e. environmental) conditions (Shima & Osenberg 2003). The successful recruitment and survival of dispersers to reproductive age is an important aspect of the dispersal process that determines demographically meaningful connections among disparate populations (Hamilton, Regetz & Warner 2008).
The decision and ability of animals to disperse is also influenced by features of the dispersal environment (i.e. the intervening landscape between the point of origin and destination; Ricketts 2001; Haynes & Cronin 2004; Murphy & Lovett-Doust 2004; Cronin 2007; Shima, Noonburg & Phillips 2010). In many ecological systems, the dispersal environment is a heterogeneous mosaic of patches that convey differential costs and benefits to dispersers. As a consequence, different dispersal environments may vary in their resistance (or permeability) to dispersers over a wider landscape, and this can shape dispersal pathways and the strengths of migratory connections between locations (Wiens 2001).
Dispersal environments may also shape the physiological condition (and future ‘fitness’) of dispersers, particularly if dispersal is energetically costly, or if experiences during dispersal can otherwise introduce variability into phenotypes of dispersers. This may be the case for the larval stages of many marine reef organisms that feed and develop for extended periods in a heterogeneous environment during their dispersal (e.g. Pechenik 2006; Hamilton, Regetz & Warner 2008; Sponaugle et al. 2009). Similarly, dispersing seeds that pass through the guts of frugivorous birds (where bird guts represent ‘dispersal environments’) may experience different rates of germination and establishment (Tewksbury et al. 2008). In such instances, dispersers may successfully move from an origin to a destination location via different pathways that vary in their costs and/or benefits. As a result of different dispersal experiences, individuals that arrive at a common destination may vary in their ability to detect and settle into their new environment (Elkin & Marshall 2007; Clobert et al. 2009).
Similarly, the persistence and survival of dispersers to reproductive age may be a function of their previous dispersal experiences, despite successful dispersal and settlement. Latent effects (sensuPechenik 2006) of prior experiences that carry-over to affect subsequent life-stages are well known for many organisms (Goater 1994; Moran & Emlet 2001; Searcy & Sponaugle 2001; Phillips 2002; Shima & Findlay 2002; Marshall & Keough 2004). However, such legacy effects of different dispersal environments have not received substantial attention in the dispersal literature, although they are potentially important in reshaping dispersal patterns for many systems.
We evaluate the potential for legacy effects of dispersal environments to alter patterns of dispersal among a set of discrete sub-populations. We focus on the dispersive larval stages of a small reef fish, and evaluate the patterns and consequences of variation in dispersal environments and disperser phenotypes and fitness, using a combination of sampling and field experiments. Because adults in our study system are relatively sedentary, larval dispersal is the primary mode of migration that connects spatially discrete sub-populations, which collectively, comprise a metapopulation. ‘Connectivity’ in marine metapopulations is typically equated with dispersal patterns of larvae. However, our work illustrates how legacies of dispersal can bias these initial patterns of ‘dispersal connectivity’ to produce a qualitatively different pattern of ‘demographic connectivity’, and these findings have fundamental implications for metapopulation dynamics and gene flow.
Materials and methods
Study species and system
We sampled larval recruits of a small temperate reef fish (the common triplefin, Forsterygion lapillum) at six spatially discrete sub-populations near Wellington, New Zealand. F. lapillum are site attached as adults (Francis 2001) and produce dispersive larval stages that feed and develop in open water for approximately 50 days (Shima & Swearer 2009a). Larvae disperse from natal reefs and may be advected over long distances by ocean currents, or they may be retained nearby (or entrained after a period of long-distance dispersal) in coastal embayments and/or hydrodynamic features such as eddies (J.S. Shima & S.E. Swearer, unpublished data). Our previous work using environmental fingerprints recorded within the otoliths (ear stones) of recent larval recruits suggested that these six sub-populations were well-connected by dispersal (Shima & Swearer 2009b). Moreover, we concluded that larval development in a semi-enclosed embayment (Wellington harbour) resulted in increased larval growth and developmental rates (i.e. ‘high quality’ larval dispersers) relative to larval development in an adjacent open coast environment (Cook Strait; Shima & Swearer 2009b). We observed similar environmentally induced phenotypic differences among individuals regardless of their putative natal origins (i.e. we found no evidence of phenotypic variation among dispersers that might be a product of differential maternal investment and/or genetic differences among sub-populations (Shima & Swearer 2009b; Fig. 1a).
Here, we evaluate the fitness consequences of this environmentally induced phenotypic variation by (i) collecting, tagging, reciprocally transplanting, and following short-term survival of recent larval recruits, and relating this to the larval quality of focal cohorts. In addition, (ii) we sample annual recruitment and survivors to adulthood for six discrete sub-populations of common triplefin, and evaluate changes in distribution of larval experiences (inferred from otolith microchemistry) through to sexual maturation across the metapopulation.
Otolith-based measures of larval quality, dispersal history and natal origin
Detailed methods for sampling of the larval recruits of the common triplefin, F. lapillum and the microstructure and microchemistry of their otoliths are given in Shima & Swearer (2009b).
Briefly, we sampled recently settled larval recruits at six discrete locations (sub-populations), at approximately weekly intervals from December 2003 to March 2004, corresponding to the seasonal peak in recruitment for this species in the vicinity of Wellington, New Zealand (41°17′ S, 174°49′ E). Collections were made from standardized artificial substrates (Ammann 2004), deployed by us, and designed to mimic the canopies of macroalgae that are more typically used as settlement habitat by this species (McDermott & Shima 2006).
We extracted otoliths from a representative sample of larval recruits collected from each sample date and location, and prepared one sagittal otolith for LA-ICPMS analysis (to quantify trace element composition), and the remaining sagittal otolith for microstructure analysis (to quantify larval phenotypes). We used a set of statistical descriptors to characterize variation in the trace element composition of otoliths. We used clustering analyses of these statistical descriptors to identify and classify each individual fish into one of two putative ‘natal origins’, and similarly, into one of two ‘dispersal histories’. Based upon the observed chemical records from otoliths and known chemical signatures from different water masses, we inferred geographical locations for these two sets of cluster analyses (either ‘Wellington harbour’ or the adjacent ‘open coast’, as the two possible putative locations for natal origins and larval development, respectively). For each individual, we also characterized ‘larval quality’ as a composite (the first principal component score of each observation, derived from a principal components analysis) of five larval phenotypic traits (pelagic larval duration, early larval growth, late larval growth, size-at-hatch and instantaneous growth rate) reconstructed from otoliths. Extensive details of all preceding approaches are given in Shima & Swearer (2009b).
We collected larval recruits of F.lapillum from a set of artificial collectors (described above) that we deployed at two separate locations known to be replenished by larval dispersers that often differed in body condition (Shima & Swearer, unpublished data). One location sampled a sub-population of F. lapillum in Wellington harbour (Shelly Bay, 41°17·8′ S, 174°49·2′ E), the other a sub-population on Wellington’s south-facing open coast (Island Bay, 41°20·8′ S, 174°46·4′ E). Collected fish were transported to the Victoria University Coastal Ecology Lab, maintained in a flow-through sea water system, fed brine shrimp ad libitum, and batch tagged using calcein (Mohler 1997). Total duration in the lab environment did not exceed 5 days for any individual used in our experiments; and individuals from each location were kept in separate aquaria.
Fish were then randomly assigned (stratified by the location of their collection) into 16 groups of five individuals. Groups were stratified by location of collection (as a proxy for variation in larval quality), because we hypothesized that individuals recruiting to the same location over the same period of time were likely to have shared similar larval developmental histories. Replicates, consisting of experimental groups of fish, were assigned randomly to treatments and reciprocally transplanted back to our two focal sub-populations, in either the harbour (at Shelly Bay) or the open coast (at Island Bay); experimental groups of fish were released on to empty artificial substrates (described in Ammann 2004) that were arrayed in a grid and separated from one another by a distance of >5 m of open water). Artificial substrates were sampled again after 7 days. Proportional survival over 1 week was evaluated using a factorial anova that included the effects of collection location (harbour or coast, a proxy for variation in larval quality) and destination location (harbour or coast) on proportional survival. This experimental design was repeated over two temporal blocks (in early February 2004 and early March 2004), and consequently, we included a blocking effect in our anova model. We estimated ‘larval quality’ from a sub-sample of the fish used in our experiments by a composite (using PCA) of five larval phenotypic traits reconstructed from otoliths (described above, and further detailed in Shima & Swearer 2009b). Spatial variation in larval quality was evaluated in a one-way anova, with temporal block as an additional effect in our model.
Forsterygion lapillum reach reproductive maturity within 6–8 months of larval recruitment (Francis 2001), and although they may live up to 3 years (Fricke 1994), most individuals in our sampled sub-populations do not survive >1·5 years (Shima & Swearer, unpublished data). We revisited the original six sub-populations (where we characterized the larval quality and developmental histories of larval recruits, described above and detailed in Shima & Swearer 2009b) and sampled the survivors of our focal recruitment season after ∼10 months. About 10–20 adults were collected haphazardly from each sub-population, by divers using hand nets. We extracted and processed sagittal otoliths (as in Shima & Swearer 2009b), and used LA-ICPMS to sample trace element composition from the larval portion of these otoliths (facilitated by marked differences in the optical densities of the larval-juvenile portions of the otolith, and a consistent elevation in Mn concentration that is concordant with settlement for this species) to obtain a set of environmental fingerprints indicative of the larval environments of surviving adults. We used the environmental fingerprints of recruits (Shima & Swearer 2009b) as a training data set in a discriminant function analysis to classify adults into one of the two observed dispersal histories (which we inferred to be indicative of Wellington harbour or the open coast respectively). We evaluated shifts in the proportional distributions of these two geographical locations of larval development, from larval recruitment to adulthood, using a Chi-Square test.
Results from our previous research (Shima & Swearer 2009b) suggest that (i) high quality larval dispersers likely develop in the harbour and not the open coast (Fig. 1a). Our new findings suggest: (ii) variation in larval quality affects juvenile survivorship irrespective of a disperser’s destination sub-population (Fig. 1b); and (iii) reproductive adults within sub-populations are comprised disproportionately of dispersers that developed within the harbour (Fig. 1c), indicating that ultimate demographic connections are strongly altered from those initially established by dispersal pathways.
Specifically, we found that larval quality was predicted by the location of larval development irrespective of natal origin (Fig 1a, see Shima & Swearer 2009b for detailed results and interpretations). From our reciprocal transplant experiment we found that the survival rates of juvenile fish varied significantly between cohorts collected from different locations (F1,26 = 4·58, P = 0·0418), and not between destination sub-populations (F1,26 = 2·55, P = 0·1225). We found no statistical interaction between cohorts and destination sub-populations (F1,26 = 0·10, P = 0·7596). The overall strength of the responses also varied between temporal blocks (F1,26 = 11·03, P = 0·0027), although other patterns were qualitatively similar. Larval quality of cohorts varied concordantly with survival performance (Fig. 1b), also differing significantly between cohorts (F1,9 = 5·84, P = 0·0388) and temporal blocks (F1,9 = 13·45, P = 0·0052). Across our metapopulation, we sampled the distribution of dispersal histories for an annual cohort of reef fish during larval recruitment (i.e. at the end of their dispersal stage) and again at adulthood. We classified the larval histories of adults with very high posterior probabilities (mean = 96% for larval histories attributed to the harbour; 92% for those attributed to the open coast) and found that fish surviving to reproductive age were comprised disproportionately of dispersers that experienced favourable conditions (putatively in the harbour environment; Chi-Square = 14·76, P = 0·00012, Fig. 1c).
Results from our short-term reciprocal transplant experiments are consistent with longitudinal sampling of natural populations, and suggest that selective mortality operating over the lifespan of a natural fish cohort can mediate survivorship to adulthood (Fig. 2a). In short, dispersal environments can alter the phenotypes of individuals that successfully disperse, and this ‘legacy of dispersal’ can modify patterns of dispersal and the topological structure (e.g. relative strengths of connections among sub-populations) within metapopulations.
We used chemical fingerprints recorded within the otoliths of larval dispersers to infer different dispersal environments (harbour versus open coast; Shima & Swearer 2009b), and we speculate that coastal embayments and eddies in our system concentrate developing larval dispersers with both food and predators (Shima & Swearer 2009b). Such conditions may (simultaneously) maximize growth and condition of developing larvae, and selectively remove underperforming individuals (both processes may, on average, yield higher quality dispersers that successfully traverse these environments). Other work in this system (Shima & Swearer 2009a) indicates a positive correlation between the magnitude recruitment (i.e. a proxy for disperser survival) and the condition of settlers (e.g. residual body mass and larval growth history) for sub-populations within the harbour. In contrast, sub-populations along the open coast exhibited a negative relationship between recruitment and settler condition. These results are concordant with our present findings, and suggest that successful larval development in a coastal embayment promotes disperser survival and confers a future fitness advantage relative to settlers with a developmental history on the open coast. One possible outcome of these observations is that natural selection may favour dispersers that are locally retained in nearshore features over dispersers that may come from more distant sources (i.e. through offshore waters). However, the effects of natural selection on dispersal patterns must be evaluated for all individuals and their life-history stages, and an alternative hypothesis is that local retention represents a high-risk/high-reward strategy for some individuals (e.g. high concentrations of both predators and food) whereas long-distance dispersal, conversely, may be lower in both risk and reward. Regardless, our results indicate that legacy effects may contribute to the outcome of natural selection processes.
Under the postulated scenario that long-distance dispersal drives a reduction in future fitness, the quality of successful dispersers may be unevenly distributed across a dispersal kernel (Fig. 2). We speculate that (at least for some settings) long-distance dispersers may be more likely to experience food stress during their dispersal, and those individuals that successfully complete a long-distance migration may arrive in poor nutritional condition and fail to survive in the destination sub-population. Conversely, dispersers that become entrained locally in hydrodynamic features such as coastal embayments or eddies (e.g. where they may be concentrated with food resources) may ultimately recruit with improved phenotypes that facilitate their survival to reproductive age. Similar costs of dispersal (relative to natal philopatry) have been reported for terrestrial vertebrates (e.g. Soulsbury et al. 2008; Young & Monfort 2009) and plants (e.g. Tewksbury et al. 2008). In short, a legacy of dispersal environments could indirectly (via future mortality schedules) truncate the tail of a dispersal kernel and limit dispersal distances (Fig. 2b; Marshall et al. 2010). However, we note that actual effects on dispersal kernels are likely to be context-dependent. For example, Stamps (2006) rendition of the silver spoon effect suggests that longer dispersal times can lead to greater selectivity for settlement sites (e.g. high quality dispersers can afford to spend more time searching, and therefore choose better sites on average). In this ‘search hypothesis’ (sensuStamps 2006), higher quality dispersers may become overrepresented at further dispersal distances, thereby extending the realized tail of the dispersal kernel. The effect of disperser quality on the shape and form of the realized dispersal kernel may depend upon whether dispersers use a silver spoon effect to extend or shorten their dispersal durations (larval dispersers of some marine organisms—including F. lapillum—appear to reduce their larval durations and settle sooner when they are in better condition; larvae of many other species have relatively invariant developmental durations).
More generally, legacies of dispersal environments can alter the pattern of connections between sub-populations that were initially established by successful dispersal events. Metapopulation connectivity underlies the dynamics, evolution and successful management of many species. For marine metapopulations, the potential direct effects of the dispersal environment on connectivity (e.g. via hydrodynamically mediated diffusion, advection and/or retention of larval dispersers by currents and eddy structures) are apparent, and the implications of these have been comparatively well studied (e.g. Gaines, Gaylord & Largier 2003; Largier 2003; Cowen, Paris & Srinivasan 2006; Siegel et al. 2008). Nonetheless, successful dispersal is only the first prerequisite for demographic connectivity, and our results suggest that patterns of connectivity may be transformed by legacies of dispersal environments.
Our results also highlight a potential for successful dispersers (but not necessarily successful ‘recruiters’) to persist in a destination population for a period of time although they may ultimately fail to contribute to future reproductive events. Although these individuals do not directly contribute to reproductive output of a sub-population, their presence (even for a short period of time) may influence local sub-population dynamics and hence the fates of other dispersers (e.g. through density-dependent interactions and/or predator attraction). Post-dispersal interactions that involve these ‘living-dead’ represent another mechanism whereby the legacy of dispersal environment may re-shape initial dispersal patterns and demographic connections between sub-populations.
The role of heterogeneous landscapes in dispersal patterns and processes is a subject of much active research (Ricketts 2001; Haynes & Cronin 2004; Murphy & Lovett-Doust 2004; Cronin 2007; Shima, Noonburg & Phillips 2010). Heterogeneity in the dispersal environment can alter the decision of individuals to disperse (Wiens 2001; Clobert et al. 2009)—or for marine larvae, the probability that they are advected away from their natal site by local hydrodynamic patterns—and it can shape the pathways and directions of movement (Haddad 1999; Wiens 2001). Pathways with higher costs relative to benefits (e.g. those with less food for dispersers, or with less shelter from predators) may directly affect demographic connections between patches because dispersers may avoid these or else incur a higher probability of mortality along the way. Our work is suggestive of how different dispersal pathways may contribute to latent variation in the fitness of successful dispersers, which indirectly affects these connections via processes acting later in the post-dispersal life history of individuals.
Another potentially important way in which the phenotypes of dispersers may influence dispersal and connectivity patterns in marine systems is through differences in the vertical distribution of larval dispersers. Many marine species exhibit diel vertical migratory behaviour in response to foraging opportunities and predation risk (Pearre. 2003) as surface waters usually contain more prey but also more predators. Because individual dispersers may differ in their gut fullness, hungry dispersers may trade-off increased predation risk for increased foraging time (Kristiansen et al. 2009). As even small differences in vertical position can have large effects on dispersal distances (Vikebøet al. 2005), larval condition may correlate strongly with both the ‘speed’ of dispersal pathways as well as their risk. Such variation among dispersers in their vertical positioning will likely lead to either (i) increased dispersal of poor conditioned larvae and thus an increased potential for latent effects after settlement, or (ii) increased mortality of poor conditioned larvae during dispersal. In both scenarios, the interaction between the dispersal landscape and the phenotype of dispersers will lead to reduced connectivity among sub-populations.
We suggest that a more integrated view of dispersal and connectivity, one that evaluates both the direct and indirect effects of dispersal landscapes, may ultimately improve our understanding of population dynamics and evolution, as well as the effectiveness of management and conservation efforts of fragmented populations.
Funding was provided by two grants from the Royal Society of New Zealand’s Marsden fund, and from Victoria University of Wellington. We acknowledge logistic support from the Victoria University Coastal Ecology Lab (VUCEL), and the research and technical assistance from the following people: J. Allen, B. Dudley, J. Ford, M. Forsyth, S. Geange, V. Hernaman, L. Liggins, J. Long, C. McDermott, D. McNaughtan and R. Williamson. We also wish to acknowledge the helpful comments of two anonymous reviewers, which enabled us to greatly improve the quality of our manuscript.