A. Foggo, Marine Biology and Ecology Research Centre, School of Biological Sciences, Davy 617, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK. Tel.: +44 1752232914. Fax: +44 1752232970. E-mail: AFoggo@Plymouth.ac.uk
1Dispersal is a crucial process in maintaining population structures in many organisms, and is hypothesized as a process underlying the interspecific relationship between abundance and distribution. Here we examined whether there was a link between the dispersal and developmental modes of marine macroinvertebrates and the slopes and elevations of interspecific abundance–occupancy relationships. We predicted that if within-site retention of larvae ranks in the order brooders > lecithotrophs > planktotrophs, for any given level of mean abundance, occupancy should increase in the order brooders < lecithotrophs < planktotrophs. We also predicted that propensity to form metapopulations should be greater for planktonic dispersers (i.e. lecithotrophs and planktotrophs combined) than for non-planktonic (i.e. brooders), resulting in steeper abundance–occupancy relationships for the former.
2Predictions were tested using a data set for 362 subtidal marine macroinvertebrates occurring across 446 1-km2 grid squares around the British Isles; analyses were performed on the data set as a whole and for separate phyla.
3The total data set had a Z-transformed effect size of 0·79, within the confidence intervals described by Blackburn et al. (2006; Journal of Animal Ecology, 75, 1426–1439), and was consistently present with relatively homogeneous effect size in separate analyses of polychaetes, crustaceans, molluscs and echinoderms.
4In all cases, planktonic dispersing organisms showed an abundance–occupancy relationship with greater elevation than that for non-planktonic organisms; in polychaetes the elevation of slopes was in the rank order planktotrophs > lecithotrophs > brooders. No differences between the slopes of the abundance–occupancy relationship were apparent for different dispersal modes either within, or across phyla.
5We conclude that dispersal capacity may play an important part in determining the elevation of the abundance–occupancy relationship, the corollary of low dispersal in the marine realm being greater local retention of larvae and greater local population abundance at low extents of geographical distribution.
The relationship between abundance and occupancy is one of the most extensively studied patterns in macroecology (Brown 1984; Gaston 1996; Gaston et al. 2000; Blackburn, Cassey & Gaston 2006), yet examples of studies in the marine environment are few. Blackburn et al.'s (2006) meta-analysis of effect sizes in macroecological studies of the abundance–occupancy relationship, found only 11 marine and 12 intertidal examples compared with 198 terrestrial studies. Despite this, marine systems offer the opportunity to test not only the consistency of some macroecological patterns across habitats that differ fundamentally in their characteristics and in terms of the constraints they place upon life, but also to inferentially test the hypotheses proposed to explain these patterns (Foggo, Frost & Attrill 2003; Raffaelli, Solan & Webb 2005).
Gaston (1999) and Gaston et al. (2000) highlight a variety of reasons why relationships between local abundance and range size/regional occupancy are important in ecology. Perhaps the most significant property of these patterns, however, is that they represent the physical evidence for mechanistic links between ecological processes operating at different spatial scales, as well as potentially reflecting differences in population organization; for example, the existence or absence of metapopulations (Freckleton, Noble & Webb 2006). Although a variety of mechanisms have been proposed to explain these relationships (see Gaston & Blackburn 2000), robust tests of many of these mechanisms are still lacking.
Warren & Gaston (1997) tested a suite of proposed macroecological hypotheses in microcosms, and concluded that ‘local abundance, combined with a very general (not specifically metapopulation structured) set of extinction and colonization processes’ were the probable drivers of microcosm occupancy. Gotelli & Taylor (1999) meanwhile demonstrated that species-level traits such as body size were significant in predicting species’ extinction and colonization probabilities in stream fishes. The amalgamation of the metapopulation (Hanski & Gilpin 1991; Hanski 1991; Hanski, Kouki & Halkka 1993), and vital rates (Holt et al. 1997) hypotheses, under the general heading of ‘population dynamics’ have perhaps allowed the most important advances, however. In particular research has focused on the importance of dispersal (e.g. Gonzalez et al. 1998; Gaston & Blackburn 2003) and the distribution and quality of habitat available to dispersers (e.g. Holt, Warren & Gaston 2004). The models recently proposed by Freckleton et al. (2005, 2006) both predict and describe how abundance–occupancy relationships are influenced by demographic processes in concert with variability in habitat availability and population/species-specific traits, such as colonization ability or propagule pressure.
The small number of reported abundance–occupancy relationships from the marine realm (e.g. Foggo et al. 2003; Fisher & Frank 2004; Frost et al. 2004) show relatively large effect sizes compared with many of their terrestrial counterparts, leading Blackburn et al. (2006) to hypothesize that the ease of dispersal in marine systems compared, for example, with more fragmented freshwater systems may be at least partially responsible. This observation, along with the recent conclusions of Freckleton et al. (2005, 2006) make comparison of existing terrestrial patterns with marine examples a valuable exercise.
Marine macroinvertebrates are phylogenetically diverse, and provide the opportunity to test the consistency of macroecological patterns across phylogenetically disparate groups. Their populations may also be more likely to be highly connected among patches than those of terrestrial organisms, forming relatively open population structures (Camus & Lima 2002; Grimm, Reise & Strasser 2003; but also see Kinlan, Gaines & Lester 2005) or highly connected sets of multiple metapopulations (sensu Gaston 2003). Most importantly, a variety of dispersal modes and colonization abilities can be found within the same groups. Examination of consistency or variability in patterns across the different realms, with their differing states of connectivity and different dominant dispersal mechanisms and colonization potentials therefore offers a unique opportunity for insight into a variety of mechanisms proposed to explain the abundance/occupancy relationship.
Dispersal potential in marine macroinvertebrates is primarily determined by the potential for passive transport either of larvae or adults (Emlet 1995; Bradbury & Snelgrove 2001), which, in turn, has been viewed as the product of the complex consequences of variations in developmental mode, dispersing particle size and fecundity (Gaines & Bertness 1992; Emlet 1995; Scheltema 1995; Jeffery & Emlet 2003; Kinlan & Gaines 2003), which together generate what may be termed ‘propagule pressure’. There are three main types of developmental mode in marine animals: (1) planktonic, feeding larvae (planktotrophy); (2) planktonic, nonfeeding larvae (lecithotrophy); and (3) non-planktonic, brooded or direct developing larvae (brooding). All of these strategies employ essentially passive dispersal mechanisms (even species lacking planktonic larvae may be dependent upon mechanisms such as rafting for large-scale movement) (Highsmith 1985; Donald, Kennedy & Spencer 2005; Thiel & Gutow 2005), with potential for survival while suspended in the water column representing the most important determinant of long-distance dispersal potential (Lester & Ruttenberg 2005). While there has been some considerable recent challenge to the paradigm representing dispersal potential as a simple correlate of developmental mode (McQuaid 1996; Bradbury & Snelgrove 2001; Paulay & Meyer 2006), there is evidence that dispersal mode can affect distribution patterns (Jablonski 1996a). Moreover, it appears that, exceptions notwithstanding, the overall rank order of dispersability of marine invertebrates is: planktotrophs > lecithotrophs > brooders (Jeffery & Emlet 2003; Paulay & Meyer 2006).
Planktotrophic larvae tend to be the smallest, produced in greatest numbers, and usually have longer planktonic larval duration, thus in general they have greater dispersal potential (Emlet 1995; Levitan 2000; Bradbury & Snelgrove 2001; Jeffery & Emlet 2003; Paulay & Meyer 2006). Lecithotrophs meanwhile tend to be smaller as adults, with larger eggs that develop faster, and intermediate dispersal potential (Strathmann 1987; Emlet 1995; Reed et al. 2000), while brooders tend to be the smallest as adults, especially among co-occurring taxa (Strathmann & Strathmann 1982; Emlet 1995), and have the largest effective dispersing particle sizes. In general these differences allow the prediction that taxa that brood their eggs should be more narrowly or patchily distributed than those that possess lecithotrophic larvae, which in turn should be more narrowly distributed than species with planktotrophic larvae. Planktonic larvae may serve to buffer population fluctuations and reduce local extinction risks (Emlet 1995; Eckert 2003) meaning that at the macroecological scale (see Blackburn & Gaston 1998; Kinlan et al. 2005) we might expect to see a gradient of connectivity of populations, with lecithotrophs presenting typically greater population connectivity through more efficient dispersal than brooders, and planktotrophs similarly greater than lecithotrophs.
Here we examine the relationship between abundance and occupancy for a suite of subtidal marine organisms from the British Isles. We compare and contrast the slopes, fits and elevations of these relationships across taxa with differing reproductive/dispersive traits both within and across clades, and draw inferences for the importance of dispersal and population dynamics in shaping the observed patterns. Following the findings of Freckleton et al. (2005, 2006), we predicted differences in the slope and elevation of the abundance–occupancy relationships between brooders, lecithotrophs and planktotrophs. At any mean abundance level, if brooders are more likely than planktotrophs to retain larvae within sites, population persistence will be more likely due to a net positive population reproductive rate (sensu Holt et al. 1997). Populations of planktonic dispersers meanwhile require either immigration or larval retention in order to persist; and both of these would be less likely than retention of larvae in brooders. At the same time, as brooders are both less fecund and less dispersive than lecithotrophs or planktotrophs, net dispersal for any given level of mean abundance of brooders will be comparatively low, producing lower regional occupancy for any given level of mean local abundance in these taxa. Habitat filling sensu Freckleton et al. (2005) will be greatest in taxa exerting greater propagule pressure, with this pressure being consistently higher for planktonic compared with non-planktonic dispersers. Hence, for any given population size, the ratio of larvae locally retained to exported will mirror the colonization ability of the taxa, thus being in the rank order brooders > lecithotrophs > planktotrophs (Fig. 1).
Secondly, we hypothesize that as mean species’ abundances increase within a reproductive mode, the probability of formation of large open populations (habitat filling and potentially even saturation) or metapopulation structures is likely to be greater in more highly dispersive species. If the probability of forming metapopulations increases in a nonlinear fashion with increasing mean abundance, the log-transformed abundance–occupancy relationship for species forming metapopulations should be steeper than that for those not forming metapopulations. If more readily dispersed taxa form metapopulations more readily, then the slope of the abundance–occupancy relationship for planktotrophs may be steeper than that for lecithotrophs, and in turn than that for brooders (Fig. 1).
sites and samples
Abundance and occupancy (as proportion of sites occupied) data were derived from the Marine Nature Conservation Review (MNCR) database (Macdonald & Mills 1996; also see Frost et al. 2004). This database encompasses many marine and estuarine habitats across the British Isles, therefore a set of a priori criteria was used to filter data for analysis (see Foggo et al. 2003). Data selected: (1) were all from full salinity, subtidal sites; (2) had thorough taxonomic surveys of epifauna and infauna from sampling performed by quantitative equipment using multiple replicate cores; (3) were from soft sediment sites with a maximum of 20% gravel and coarser fractions in the sediment. Samples were used to generate mean densities of fauna in 446 separate 1-km OS grid squares (Fig. 2). The spatial distribution of the squares was aggregated at the scale of tens of kilometres due to the survey-based nature of the MNCR sampling; however, spatial coverage of the coastal seas of the UK was adequate to allow within-patch larval retention and between-patch dispersal to be investigated treating the sampled squares as semi-isolated patches. Further details of methodologies employed in sampling and database generation are given in Hiscock (1996).
Known introduced taxa and colonial organisms were excluded from the list. An extensive bibliographic search was then used to establish the dominant reproductive/dispersal character for as many species as possible, assigning species to three main categories: brooders, lecithotrophs (nonplanktotrophic) and planktotrophs. Data for many of the taxa in the British subtidal marine fauna are not available, thus reproductive modes were assigned based on published information regarding congenerics and confamilials, but only in cases where data indicated consistency in traits among congeners or confamilials for more than one sibling taxon. Taxa that could not be assigned to a category with such a degree of confidence were not employed in analysis. Traits were reconciled for adequate numbers of species in the phyla Polychaeta, Mollusca, Crustacea and Echinodermata to allow comparison of traits to be performed in these groups.
A Type I regression and ancova approach was adopted for analysis in the interests of comparability with previous published studies. Differences between slopes and elevations of abundance–occupancy relationships were tested using ancova with taxonomic groupings (phylum) and reproductive strategies as fixed factors. All regression analyses were verified by inspection of residuals, Cook's distances and DFITS values. Because the majority of lecithotrophic taxa were polychaetes, there was an inherent taxonomic bias in analysing data describing the strategy outside this group. Thus, within the polychaetes we analysed three strategies, but for all other analyses, and for composite analyses involving all groups, we classified taxa as having either planktonic or non-planktonic development, lumping planktotrophs and lecithotrophs.
All statistical analyses were performed using Statistica ver. 6·1 (Statsoft, Tulsa, AZ, USA). We adopted the model of Hanski & Gyllenberg (1997) to test for the abundance–occupancy relationship, where Logit(occupancy) is plotted against Log(abundance). Occupancy was calculated as the proportion of all squares at which a species occurred, and abundance as mean density per m2 across all occupied squares.
In total, 362 of the 572 taxa in our database were categorized into the three reproductive/dispersal modes. Linear regression indicated a significant overall positive relationship between abundance and occupancy (R2 = 0·437; Table 1; Fig. 3a). The Z-transformed effect size for this overall relationship (see Blackburn et al. 2006) was 0·79, higher than the values of both the marine-specific mean global means for all studies calculated by Blackburn et al. (2006). When all planktonic and non-planktonic taxa were compared ancova indicated no significant difference between the slopes of the two abundance–occupancy relationships; however, the relationship for planktonic developers was significantly higher than that for non-planktonic (Table 1, Fig. 3b). There was no significant interaction between taxonomic identity and dispersal mode in ancova (F3,353 = 1·12, P = 0·340), and no statistical differences between the slopes of relationships for different dispersal mechanisms in separate phyla. Abundance–occupancy relationships were consistently higher for planktonic than non-planktonic dispersers irrespective of taxon (Fig. 3c–f), with significant differences between dispersal modes in polychaetes and echinoderms (Table 1, Fig. 3c,f). ancovas and subsequent linear regressions for the separate taxa indicated that all the separate phyla with the exception of benthic developing echinoderms (Fig. 3f) had significant positive abundance–occupancy relationships (Table 1).
Table 1. Results of type I linear regressions of log-abundance (independent) vs. log occupancy (dependent) in all taxa, and ancova tests for difference between the different dispersal modes in that taxonomic grouping, both before and after categorization by dispersal mechanism, and in all taxonomic subgroups categorized by dispersal mechanism. β0 = slope, β1 = intercept
ancova (dispersal mode)
ancova (dispersal mode)
ancova (dispersal mode)
ancova (dispersal mode)
ancova (dispersal mode)
The presence of a significant overall, and taxonomically consistent, abundance–occupancy relationship supports previous observations of this general ecological pattern in marine systems (Foggo et al. 2003; Frost et al. 2004; Goodwin, Dulvey & Reynolds 2005). The observed slopes of the regressions are slightly steeper than those previously observed for invertebrates in the marine realm; however, the R2 values produced are within the range previously described both in the marine and terrestrial environments (see Gaston 1996; Foggo et al. 2003; Frost et al. 2004). The most significant result of this study is the consistently higher fits of the abundance–occupancy relationship for taxa with planktonic dispersal. In particular the polychaetes demonstrate a series of relationships entirely consistent with our first prediction, that dispersal limitation and larval retention might be a significant factor in structuring the abundance–occupancy relationships in marine macroinvertebrates. There was little evidence, however, to suggest that organisms with greater dispersal potential might be establishing metapopulations; none of the slopes for different dispersal modes were statistically different.
Our results are consistent with a prominent role for dispersal in shaping the observed patterns, in accordance with recent theoretical contributions (see Freckleton et al. 2005).
If the general trend for planktonic dispersers to be more widespread at any given level of mean abundance holds true, then the primary reason for this is likely to be the total propagule pressures exerted by the species. Species possessing different dispersal modes generate different total propagule pressures as a result of variation in fecundity and propagule size and hence transport potential. Counteracting this is the effect of propagule dilution, which increases with distance transported, and the benefits of maternal provisioning (McEdward & Miner 2003) or parental care (Thiel 1998) in lecithotrophs and brooded species. To disentangle the relative effects of these traits would require knowledge about the mean fecundities, propagule sizes, survival probabilities and dispersal potential of the species; such data are not available for adequate numbers of taxa (Rundle, Bilton & Foggo 2007). However, reconciling the roles of these two determinants of dispersal ability is simplified if the assumption that the probability of within-population larval retention is greater in brooders is shown to be unfounded. A growing number of publications points to the probability that planktonic dispersers may in fact show significant local larval retention despite their potential for long-distance larval transport (Todd 1998; Strathmann et al. 2002; Swearer et al. 2002). In this case, if both widespread dispersal and local larval retention are possible over ecologically significant time-scales in both planktonic and non-planktonic, then the main determinant of distribution at any mean level of abundance is simply likely to be fecundity of the organisms determining the probability of successful emigration events.
In many instances, fecundity may be a correlate of body size; larger species within each dispersal mode might produce more offspring, and planktonic dispersers at a given abundance level be larger and hence more fecund than non-planktonic dispersers. In a recent review, we demonstrated that, for many of the polychaetes studied here, the average size of planktotrophic dispersers is greater than that of non-planktonic forms (Rundle et al. 2007), although planktotrophic taxa were not significantly larger than lecithotrophs despite showing a more elevated abundance–occupancy relationship in the present study. In many clades the relationship between body size and fecundity may be more straightforward (Brown 1995; Jablonski 1996b), and the consequences of differences in size, potentially mediated by effects of fecundity, have been shown in particular in studies of marine bivalves. Roy, Jablonski & Valentine (2001) demonstrated that larger bodied bivalves have tended to show more dynamic geographical ranges in response to Pleistocene climate shifts, and the same authors later demonstrated that invasive marine bivalves also tended to be larger bodied (Roy, Jablonski & Valentine 2002). Both of these conclusions are consistent with our results and the proposition that fecundity determining dispersal probability is the primary determinant of the abundance–occupancy relationship in many marine macroinvertebrates. This is not to say that the relationship between size and fecundity within a dispersive-taxonomic group of species is always simple; the generalization that larger species produce more offspring and that dispersal probability is therefore greater in these taxa may be unfounded if survival probability of neonates and juveniles is not positively correlated with parental size through variation in maternal provisioning (Chaparro et al. 1999; McEdward & Miner 2003) or parental care (Thiel 1998). Moreover there is evidence that egg size (and hence cost), fecundity and adult body size may not be simply correlated in some taxa (e.g. Mchugh 1993; Collin 2003), casting doubt over the role of adult body size in determining number of dispersing propagules and hence the form of the abundance–occupancy relationship.
The significance of consistent differences in elevation of the abundance–occupancy relationship may be greatest for the monitoring and conservation of marine taxa, and the design of marine reserve networks (Palumbi 2003; Carr et al. 2003; Shanks, Grantham & Carr 2003; Guichard et al. 2004). If formation of linked population structures and exploitation of available habitat is indeed facilitated by dispersal as our result suggests, then the corollary of this is that benthic direct developing organisms will be most likely to suffer regional level population declines from influences such as habitat degradation and loss, which eliminate local populations. While other local populations might persist through internal recruitment under scenarios of habitat loss, they might be less able to shift their ranges in face of rapidly fluctuating conditions in coastal seas resulting from climate change. Marine reserve structures for such organisms may therefore need to be highly linked, while planktonic dispersers would be able to exploit less spatially contiguous habitats. Only better data concerning distribution, density, body size, fecundity, propagule pressure, and pre- and post-dispersal survival of the taxa concerned can facilitate more informed analyses of population processes in marine macroinvertebrates, and better strategic planning for their conservation.
The authors wish to thank D. Connor and the Marine Habitats Team at the Joint Nature Conservation Committee in Peterborough for permission to access and use these data and Matt Frost for all his help regarding the MNCR. We wish to thank Alex Fraser, Shelley Taylor and Alex MacDonald for help with determining reproductive modes, and express our thanks to colleagues in the Plymouth Marine Institute for valuable conversations, especially J.I. Spicer, M.J. Attrill, J.D.D. Bishop, S.J. Hawkins and M.B. Jones. The constructive criticism of two anonymous referees was instrumental in considerable improvements in the final submission.