1 0 5 10 15 20 25 30 35 40 45 50 52 53 55 60 65 70 75 80 85 90 95 100 105 High connectivity among habitats precludes the relationship between dispersal and range size in tropical reef fi shes Camilo Mora , Eric A. Treml , Jason Roberts , Kate Crosby , Denis Roy and Derek P. Tittensor C. Mora ( moracamilo@hotmail.com ), K. Crosby, D. Roy and D. P. Tittensor, Dept of Biology, Dalhousie Univ., Halifax, NS, B3H 4J1, Canada. ��� E. A. Treml, Th e Ecology Centre, Univ. of Queensland, St. Lucia, QLD 4072, Australia. ��� J. Roberts, Nicholas School of the Environment, Duke Univ., Durham, NC 27708, USA. Th e hypothesis that pelagic larval duration (PLD) infl uences 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 eff ect 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 eff ect on range size then an outstanding question is why? Here we collected data on PLD, evolutionary ages and range sizes of seven tropical fi sh families (fi ve families were reef-associated and two have demersal habits) to explore the independent and interactive eff ects of PLD and evolutionary age on range size. Independent analyses on each family showed that even after controlling for evolutionary age, PLD has an insignifi cant or a very small eff ect on range size. To shed light on why dispersal has such a limited eff ect 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. Th ese results suggest that the lack of habitat isolation can potentially render the constraining eff ect of dispersal on range size insignifi cant and explain why dispersal does not relate to range size in reef fi shes. Identifying factors shaping the geographical distribution of species is a fundamental aim of macroecology (Gaston 2003). For tropical reef fi shes, the role of dispersal ability, quantifi ed as the duration of the pelagic period (i.e. pelagic larval duration PLD), has received considerable atten- tion as a driver of range size because as adults these species are mostly sedentary (Th resher and Brothers 1985, 1989, Th resher et al. 1989, Wellington and Victor 1989, Zapata and Herron 2002, Goodwin et al. 2005, Lester and Ruttenberg 2005, Lester et al. 2007). Th e rationale is that species with longer PLDs can disperse over longer distances, resulting in larger geographical ranges compared to species with shorter PLDs. Although dispersal distances can be aff ected by addi- tional factors such as currents and larval behaviors (reviewed in Mora and Sale 2002), the use of PLD as a quantitative measure of dispersal potential in reef fi shes has been pre- viously supported (Riginos and Victor 2001, Shanks et al. 2003, Siegel et al. 2003, Lester and Ruttenberg 2005, Lester et al. 2007). Interestingly, however, numerous studies explicitly comparing PLD to range size have failed to dem- onstrate a signifi cant or important relationship between the two (Th resher and Brothers 1985, 1989, Tresher et al. 1989, Wellington and Victor 1989, Victor and Wellington 2000, Jones et al. 2002, Zapata and Herron 2002, Goodwin et al. 2005, Lester and Ruttenberg 2005, Lester et al. 2007). Although recent meta-analyses, including all available data on PLDs, have concluded that dispersal is not a general determinant of range size (Lester and Ruttenberg 2005, Lester et al. 2007), the nullifi cation of the hypothesis that PLD infl uences range size may be premature due to the failure of previous studies to account for species evolutionary ages (i.e. older species ought to have had more time to disperse and if dispersal infl uences range size then older species would be expected to have larger ranges than younger species). Species evolutionary ages may also confound the eff ects of PLD because species with large PLDs may have small geographic ranges if they are relatively young alternatively, species with short PLDs could potentially have larger ranges if they have been dispersing over longer evolutionary time scales. Th is situation calls for analyzing the combined eff ects of PLD and species ages on range size. Consequently, if even after accounting for evolutionary age, PLD does not aff ect range size, then a critical question is why? In this study, we gathered data on PLD, evolutionary age and range sizes of tropical reef and demersal fi shes to evalu- ate the relative eff ect and interaction strength of evolution- ary (i.e. species age) and ecological (i.e. PLD) dispersal on range size. We found that even after controlling for species age, PLD has a very small or statistically insignifi cant eff ect on range size. To gain further insight as to why dispersal does Ecography 000: 001���008, 2011 doi: 10.1111/j.1600-0587.2011.06874.x �� 2011 Th e Authors. Journal compilation �� 2011 Ecography Subject Editor: Th ierry Oberdorff . Accepted 18 March 2011 ECOG_A_006874.indd 1 ECOG_A_006874.indd 1 4/21/2011 8:37:41 PM 4/21/2011 8:37:41 PM

3 0 5 10 15 20 25 30 35 40 45 50 55 60 61 65 70 75 80 85 90 95 100 105 110 115 121 inter-annual variability in currents, we performed four dis- persal simulations per year for the entire time span avail- able in the HYCOM NCODA dataset (i.e. 2004���2009, for a total of 24 simulations). Th e time period for which hydrodynamic data was available encompasses enough vari- ability in episodic events known to aff ect water circulation (Supplementary material Appendix 1 Fig. A1), suggesting that our results refl ect the generality of larval connectivity in the tropical ocean. It is clear, however, that a longer time frame would likely increase the chances of infrequent dispersal events making our results more conservative. We released larvae at the midpoint of each season (i.e. 2 February, 5 May, 6 August, and 11 November) and simulated dispersal for 100 d, a duration encompassing the majority of PLDs in tropical reef fi shes (Supplementary material). After completing the 24 simulations, we created a single asym- metric dispersal matrix giving the minimum dispersal time between all pairs of habitat patches obtained over the entire set of simulations. We then compared these times to the database of PLDs to calculate the fraction of species whose PLD was shorter than the time it took to cross the diff erent gaps between patches, thereby highlighting where biogeo- graphical barriers to the dispersal of reef fi shes are likely to exist in the world ��� s tropical reefs. Results We collected data on PLD for 547 species in 67 families from 137 sources (Supplementary material). Only 13 families had data on PLD for more than 10 species (i.e. Pomacentridae, Labridae, Pomacanthidae, Gobiidae, Acanthuridae, Chaeto- dontidae, Lutjanidae, Anguillidae, Serranidae, Sparidae, Blenniidae, Scombridae and Pleuronectidae). Of these fami- lies, we found representative phylogenies for seven families [i.e. Pomacentridae (Cooper et al. 2009), Labridae (Cowman et al. 2009), Pomacanthidae (Bellwood et al. 2004), Chaeto- dontidae (Bellwood et al. 2010), Anguillidae (Minegishi et al. 2005), Serranidae (Craig and Hastings 2007) and Spari- dae (Chiba et al. 2009)]. Most of these families comprise reef-associated species with the exception of Anguillidae and Sparidae, which are mostly dermersal (Supplementary mate- rial). We maintained data on all seven families as the hypoth- esis that dispersal infl uences ranges size is relevant to all families with benthic habitat associations. In combination, we obtained data on species PLDs and evolutionary ages for 183 species in the seven families. For the species in these seven families, we obtained 98, 385 biogeographical records, all of which were verifi ed visually against their reported dis- tribution in general reef fi sh books (see Methods). All of the seven families analyzed have relatively large body sizes sug- gesting that these species have a high sighting probability. Additionally, fi sh assemblages on reefs are the best sampled in the world ��� s oceans (Mora et al. 2008) and for at least the Pacifi c and Indian Oceans we used published checklists of species, which are known to be relatively complete. Th ese facts combined suggest that the records we obtained are very likely to provide a reliable indication on the distribution of the analyzed species. We found that neither the main eff ects of PLD or evolutionary age nor their interaction were important determinants of species geographical ranges in the seven independently analyzed families (Fig. 1). We kept the analy- ses for each family independent because the lack of phylo- genetic data on the relatives among families and consistency on the genetic markers among studies prevents the concat- enation of all families into a single phylogenetic ��� supertree ��� . Th e analyses on each family show that geographical ranges were equally variable in size along the entire series of PLDs and species evolutionary ages (Fig. 1). Only Pomacentrids showed a signifi cant but weak relationship with PLD, which explained 4% of the variance in range sizes (Fig. 1). In the other six families, the relationships between range size and PLD and evolutionary age and their interaction were statistically non-signifi cant (Fig. 1). Th e world ��� s tropical oceans have several regions of great isolation for reefs, such as the region between the central and eastern Pacifi c, the central Atlantic, and the central Indian Ocean, where distances to the nearest potential reef habitat can be as large as 1700 km (Fig. 2a). When con- sidering isolation under hydrodynamic conditions many of these distances can be traversed over relatively short time periods. Only among few reefs were there travel times 100 d (Fig. 2b) this was due to a combination of long distances of separation and slow current speeds (i.e. the eastern and southern Pacifi c and the south Indian Ocean) (Fig. 2b). For the majority of tropical oceans, travel times between reef patches were on the order of 30���100 d (Fig. 2c). Among the 547 species for which we collected data on PLD, the average PLD was 42.6 d (stdev 41.5) with several species having PLDs of over 100 d (Supplementary material). As result, when comparing the duration of the pelagic larval stage to the times to cross between reef patches, we found that few species are fi ltered by isolation across the tropi- cal oceans. Only two geographical locations could poten- tially fi lter reef fi sh species through dispersal limitations at PLDs 70���80 d ( ��� 88% of species fi ltered). One occurred along the fringes between the central and eastern Pacifi c (i.e. the Eastern Pacifi c Barrier) and the other between the western and eastern Atlantic (Fig. 2c, d). Other than these two gaps, which are known to defi ne diff erent biogeo- graphical regions with distinctive fi sh faunas (MacPherson 2002, Mora and Robertson 2005a, b, Floeter et al. 2008), most of the habitats in the ocean are within the travel time conferred by the PLD of most reef fi shes (Fig. 2d). Note that this pattern on the isolation by dispersal of reef fi shes is highly conservative because we considered average PLD and average daily currents at a 25 km 2 resolution. If maximum PLDs and higher resolution currents were used, the levels of isolation may have been smaller given the greater current speeds and dispersal potential. Similarly, we did not consider the potential extreme events of dispersal that some species may experience through their associa- tion with fl oating objects (Mora et al. 2001, Castro et al. 2002). Note: we did not use maximum PLD in our analy- ses because the availability of such data was variable among sources furthermore, among studies sampling size to esti- mate PLDs was often small and variable making maximum PLD not a very reliable metric for comparison among spe- cies that is, extreme events of dispersal are expected to be rare and therefore extensive sampling is required to reliably quantify them. ECOG_A_006874.indd 3 ECOG_A_006874.indd 3 4/21/2011 8:37:41 PM 4/21/2011 8:37:41 PM