What are the processes that give rise to new species of sea urchins? One might assume that echinoids conform to general principles that hold for all sexu- ally reproducing organisms, but what are the data that support this assumption? And given that many of these principles are still the subject of debate, which views do the echinoid data support? In this paper I attempt to summarize what is known about speciation in sea urchins and how data obtained from the class Echinoidea can address general ques- tions of speciation. The last century has seen the development of general principles of speciation meant to apply to all sexually reproducing animals (Mayr 1942 1963 1970, Dobzhansky 1937 1970, Otte & Endler 1989, Howard & Berlocher 1998), an effort that is continu- ing to the present day, (Coyne & Orr 2004, Gavrilets 2004). Very briefly, new species arise when gene flow is interrupted by a geological or oceanographic barrier. If during this period of isolation they diverge in traits important for their development or repro- duction (so they are no longer able to interbreed even when the barrier is lifted), they are converted to different biological species. This is the model of al- lopatric speciation that arose from the ?New Synthe- sis?, and was supported persuasively by Mayr and Dobzhansky. Evolution of reproductive isolation in the presence of gene flow (sympatric speciation) is considered by these authors as improbable, but ac- cording to various models (reviewed by Coyne & Orr 2004, Gavrilets 2004), it can occur if some very restrictive conditions obtain. The data on which these speciation principles are based have come primarily from insects and vertebrates. In speciation research, echinoids have played a small supporting role, but one that is not without importance. Perusal of the latest compendium (Coyne & Orr 2004) indi- cates that data on echinoids are used to address questions of allopatric vs. sympatric modes of speci- ation (p. 94), of temporal (p. 61) and gametic (pp. 63, 226, 235) reproductive isolation, of the extent of hybridization (p. 70), and of natural selection to avoid wastage of gametes in hybrids (pp. 243, 359). The importance of the echinoid data, limited as they Speciation in sea urchins H.A. Lessios Smithsonian Tropical Research Institute, Balboa, Panama pp. 91-101 In: L.G. Harris, S.A. Bottger, C.W. Walker, M.P. Lesser. Echinoderms: Durham. CRC Press, London. 2010. ABSTRACT: Data relevant to processes that give rise to new species of echinoids are reviewed. Phy- logeographic information from mitochondrial DNA is used to ask whether speciation in sea urchins fits the al- lopatric model, which predicts that if reproductive isolation accumulates steadily with time, then young, closely related species would tend to occur on two sides of a geographic barrier. The conclusion of this analy- sis is that most genera show a strong signature of allopatric speciation, but that Echinometra, Lytechinus and Strongylocentrotus also contain recently diverged sister species that are sympatric. The reason for these ex- ceptions is probably not that sympatric speciation has occurred, but rather that reproductive isolation is not a function of divergence time alone. Although postzygotic isolation (lower hybrid fitness) in echinoids is corre- lated with divergence time, prezygotic isolation seems to arise due to additional factors, which are not nec- essarily related to the time that species have remained separate. There is no correlation between gametic in- compatibility and time; bindin, one of the molecules responsible for gametic isolation, evolves under strong selection in some genera, but neutrally in others. Even though there is some evidence for reinforcement as a selective force on bindin, the differences in intensity of selection between the genera are more likely caused by intraspecific processes, such as variation in local sperm density. If so, age of species would not be a good predictor of geographic overlap, because young species may be reproductively isolated, while old species may be compatible, regardless of whether they arose sympatrically or allopatrically. The paucity of extant species in echinoids indicates that even though gametic reproductive isolation can arise rapidly, such events, leading to speciation, do not happen often. are, is that they provide a test as to whether ideas developed from the study of arthropods and verte- brates also apply to organisms with different fertili- zation systems and reproductive ecology. The two main questions to which data from echinoids have contributed are the spatial mode of speciation and the processes that give rise to reproductive isolation. 1 SPATIAL MODE OF ECHINOID SPECIATION Do sea urchins speciate only allopatrically, or is there evidence for sympatric speciation as well? The existence of biogeographic provinces in the ocean (Ekman 1953, Briggs1974) leaves no doubt that ma- jor obstacles to gene flow can cause speciation in marine organisms. Sympatric speciation can occur anywhere in a species range, independently for each event. Thus, if all speciation were sympatric, there should be no congruence in species ranges of differ- ent organisms. The question that remains open is whether some sympatric speciation might also oc- cur. Because species ranges change with time, the question is not easily resolved by simply examining modern species distributions, but, if one assumes that recently formed species are more likely to be present where they appeared, extensive overlap be- tween sister species would be evidence for sympat- ric speciation. This is the well-known ?Jordan?s rule?. In 1905 D.S. Jordan wrote: ?Given any spe- cies in a region, the nearest related species is not found in the same region nor in a remote region, but in the neighboring district separated from the first by a barrier of some sort...? (Jordan 1905). Jordan?s rule was applied to tropical echinoid genera by Mayr (1954). Mayr used the monograph of Mortensen on echinoid systematics (Mortensen 1928-1951) to illustrate that sea urchins in the sea, like birds on land, tend to speciate allopatrically. His approach was to plot the range of each species in 16 tropical shallow water sea urchin genera and ask whether there was evidence that recently separated species tended to have non-overlapping ranges. Mayr was aware that echinoid systematics were still at the stage of alpha taxonomy, and that the only available documentation of specific status was mor- phological. Lacking phylogenetic information, he was forced to deduce which genera contained young and which contained old species in part by species distributions, which led to a certain degree of circu- larity. For example, he assumed what he intended to prove when he proposed that Diadema savignyi and D. setosum, which at the time were considered to have identical geographic distributions, (and the phylogeny of which was unknown) must be old spe- cies, one of which invaded the range of the other af- ter a long period in allopatry. Molecular phylogenies have since provided the opportunity to reconstruct robust phylogenies of many of these genera and to provide approximate times of splitting between the species. Palumbi & Lessios (2005) have reviewed the degree to which these more recently obtained data support Mayr?s conclusions. In this section, I present brief summaries of these phylogenies and Figure 1. Phylogeny of the extant species of Eucidaris, based on sequences from the Cytochrome Oxidase I (COI) region of mito- chondrial DNA, and geographic distribution of the clades (Lessios et. al. 1999). The thick bar indicates the presumed break arising from the emergence of the Isthmus of Panama, 3.1 million years (my) ago. Numbers next to the nodes indicate approximate dates (in my) of most recent common ancestor, based on a calibration of COI divergence by the emergence of the Isthmus. Actual loca- tions in which the samples were collected are shown on the map. the information they can provide as to the actual bar- riers that caused speciation events in each genus. 1.1 Eucidaris In the genus Eucidaris all extant species are allo- patric, so there was never any question as to whether they speciated sympatrically. The mitochondrial DNA (mtDNA) phylogeny of Lessios et al. (1999) permits examination of whether the species as de- fined by Mortensen (1928-1951) on the basis of morphology are valid, and--by placing time con- straints on the nodes--the likely barriers that resulted in their existence (Fig. 1). The oldest speciation event between extant species of this genus was caused by the Eastern Pacific Barrier, the long stretch of deep water without stepping stones be- tween the eastern and the central Pacific, which pre- sumably isolated E. metularia from the eastern Pa- cific clade of this genus. At some earlier point, the Benguela cold water upwelling in SW Africa was also operating, so that Atlantic and Indian Ocean populations could not exchange genes with each other. Then came separation of the Atlantic E. tribu- loides from an eastern Pacific clade, as the result of the rise of the Isthmus of Panama, approximately 3 million years (my) ago. The outer islands in this re- gion, Galapagos, Isla del Coco, and Clipperton, to- gether harbor a clade of mtDNA that is reciprocally monophyletic from the coastal E. thouarsi, indicat- ing that even relatively short ocean distances can act as barriers to gene flow and be implicated in speci- ation. The alternative, that E. thouarsi and E. gala- pagensis speciated sympatrically and that the two clades sorted themselves out in different areas seems improbable, given the occurrence of a number of other endemic echinoderm species in the Galapagos (Maluf 1991), which attests to the potential isolation of this Archipelago from the mainland. There are no phylogenetic breaks in the Atlantic between popula- tions of Eucidaris from the Caribbean, Brazil, As- cension, St. Helena, and the African coast, but high FST values indicate that Eucidaris clavata in the cen- tral Atlantic islands is isolated from coastal popula- tions of E. tribuloides. All other populations ex- change genes with each other at high rates. Thus, all speciation events in Eucidaris conform to a model of allopatric speciation, and the phylogenetic breaks coincide with major phylogeographic barriers as de- fined on the basis of faunal provinces from a variety of tropical shallow water organisms (Ekman 1953, Briggs 1974). 1.2 Tripneustes Tripneustes is another genus of shallow water tropi- cal echinoid in which all the species are allopatric. The mtDNA phylogeny of Lessios et al. (2003) indi- cates that, unlike larvae of Eucidaris, those of Trip- neustes are able to cross the Eastern Pacific Barrier Figure 2. Phylogeny of the extant species of Tripneustes, based on sequences from the Cytochrome Oxidase I (COI) region of mi- tochondrial DNA, and geographic distribution of the clades (Lessios et. al. 2003). The thick bar indicates the presumed break aris- ing from the emergence of the Isthmus of Panama, 3.1 million years (my) ago. Numbers next to the nodes indicate approximate dates (in my) of most recent common ancestor, based on a calibration of COI divergence by the emergence of the Isthmus. Actual locations in which the samples were collected are shown on the map. on a regular basis (Fig. 2). There are identical mito- chondrial haplotypes of Tripneustes spread from the west coast of America to the east coast of Africa, and there is no phylogenetic distinction between the Cytochrome Oxidase I (COI) of the presumed sepa- rate species T. gratilla from the Indo-West Pacific and T. depressus from the eastern Pacific. Phylog- eny of the nuclear locus bindin is consistent with the mtDNA phylogeny (Zigler & Lessios 2003). In con- trast to the ease with which Pacific populations of Tripneustes appear to maintain connections between Pacific regions, and in contrast to the facility with which genes of Atlantic populations of Eucidaris spread from the American to the African coast, the Atlantic populations of Tripneustes appear to en- counter major obstacles to dispersal. Caribbean and Brazilian populations have been separated by the in- hospitable habitat created by the plume of the Ori- noco and the Amazon for a period long enough to have developed reciprocally monophyletic mtDNA haplotypes. The American clade is also distinct from the African one, suggesting that larvae of this genus are unable to cross the mid-Atlantic barrier. Thus, all speciation events in Tripneustes, like those in Eucidaris, can be explained with an allopatric model, but the two genera only share the Isthmus of Panama and the Benguela upwelling as common barriers to gene flow. 1.3 Arbacia Arbacia is yet another genus in which all extant spe- cies (except for some range overlap between A. stel- lata (= A. incisa), A. spatuligera, and A. dufresni on the W. coast of S. America) are allopatric. This ge- nus is found only in the Atlantic and the eastern Pa- cific. Mayr (1954) considered the species of Arbacia to be so old, that the present-day ranges give no in- dication of where the speciation events occurred, but the COI and bindin phylogeny of Metz et al. (1998) provides little justification for this view (Fig. 3). The ancestral population of all extant species in this ge- nus appears to have been divided by the emergence of the Isthmus of Panama, though the split between the Atlantic and the Pacific clades may have oc- curred before the complete closure of the portals connecting the two oceans. The Atlantic branch was then divided by the mid-Atlantic barrier to the east- ern Atlantic-Mediterranean A. lixula and the western Atlantic A. punctulata. A. lixula then crossed the At- lantic to establish a branch in Brazil, which has re- mained cut off from its parental population for a suf- ficiently long time to be reciprocally monophyletic. Metz et al. (1998) did not include A. spatuligera in their phylogeny, but this species, ranging from Ec- uador to S. Chile, is sister to A. dufresni (Lessios, unpubl.). Thus, in the eastern Pacific there was first a speciation event separating the more northern A. incisa from the southern clade, which subsequently split into A. dufresni and A. spatuligera. Although the barriers that caused speciation events in the east- ern Pacific are not obvious (and certainly do not seem to impede gene flow in Eucidaris thouarsi or in Tripneustes depressus), there is little reason to suspect sympatric speciation in this genus, given their present-day ranges. 1.4 Diadema The genus Diadema, with more species than the previous two genera, presents a more complicated picture. MtDNA and isozyme phylogenies (Lessios et al. 2001) indicate that Diadema setosum split first from all other species of Diadema, probably during the initiation of wide fluctuations in global sea levels in the Miocene (Fig. 4). This D. setosum clade then split 3?6 million years ago into two clades, one in the northern Indian Ocean, and the other in the southern Indian Ocean and the West Pacific. There is no obvious barrier that might have isolated the two clades, but they remain allopatric. On the line- age leading to the other species of Diadema, there was an early branching event leading to the New Zealand and SE Australia endemic D. palmeri, coin- ciding with the time of cooling of the climate of New Zealand that lead to the extinction of other Figure 3. Phylogeny of the extant species of Arbacia, based on sequences from the Cytochrome Oxidase I (COI) region of mitochondrial DNA, and geographic distribution of the clades (Metz et. al. 1998). The thick bar indicates the presumed break arising from the emergence of the Isthmus of Panama, 3.1 million years (my) ago. Numbers next to the nodes indi- cate approximate dates (in my) of most recent common ances- tor, based on a calibration of COI divergence by the emer- gence of the Isthmus. Actual locations in which the samples were collected are shown on the map tropical echinoids in this region. The next lineage to separate was composed of a currently undescribed species of Diadema found at Japan and also at the Marshall Islands. The barrier that caused this split is also not obvious; this undescribed species exists sympatrically with D. savignyi at both locations, but as it split from it 4-5 my ago, it represents no chal- lenge to the allopatric model. The Eastern Pacific Barrier caused the next cladogenic event that sepa- rated D. mexicanum in the eastern Pacific, from which the Atlantic D. antillarum was subsequently split by the Isthmus of Panama. Within the Atlantic, there is a biogeographic barrier between the Carib- bean and Brazil (Lessios et al. 2001). Diadema antil- larum populations of the central Atlantic islands of Ascension and St. Helena are genetically isolated and phylogenetically derived from those of Brazil (Lessios et al. 2001). A completely separate clade in the E. Atlantic, suggests that the mid-Atlantic barrier has been effective in this genus. Apparently, the Indo-Pacific D. paucispinum and D. savignyi main- tained genetic contact with D. antillarum around the southern tip of Africa for some time after the Isth- mus of Panama was complete, but ceased to do so at the onset of Pleistocene. It is not clear how these two species separated from each other in the Pleisto- cene, but one possibility is that D. paucispinum was isolated in Hawaii or Easter Island, where it speci- ated, then invaded the rest of the central and Indo- West Pacific, as well as the Indian Ocean. D. pau- cispinum actually contains two lineages: One clade of this species is the only representative of Diadema in Easter Island and Pitcairn, but is also found in Okinawa in sympatry with two other species. A sec- ond mitochondrial clade of D. paucispinum extends from East Africa and Arabia to the Philippines and New Guinea. Presumably, these two clades are the result of water flow restrictions in the straits be- tween northern Australia and Southeast Asia during Pleistocene episodes of low sea level, and the subse- quent leakage of the Indian Ocean clade into the fringes of the western Pacific. Thus, the mtDNA phylogeography of Diadema indicates that all stages expected from a model of allopatric differentiation are present in this genus. There are anciently sepa- rated clades that now overlap in their geographic distribution, clades isolated in the periphery of the genus range that have remained in the periphery, clades that may have been isolated in the periphery but have since spread towards the center, closely re- lated clades on either side of an existing barrier, and closely related monophyletic entities on either side of an historical barrier that have subsequently crossed the former barrier line. Except for D. pau- cispinum and D. savignyi, in which hybridization may have lodged mtDNA from one species into the genome of the other, closely related clades are al- ways allopatric. Thus, the phylogenetic history and Figure 4. Phylogeny of the extant species of Diadema, based on sequences from the Cytochrome Oxidase I (COI) and the ATPase regions of mitochondrial DNA, and geographic distribution of the clades (Lessios et. al. 2001). The thick bar indicates the pre- sumed break arising from the emergence of the Isthmus of Panama. Numbers next to the nodes indicate approximate dates (in my) of most recent common ancestor, based on a calibration of COI divergence by the emergence of the Isthmus. Actual locations in which the samples were collected are shown on the map. Smaller squares indicate that only occasional haplotypes of a clade in- habit a particular area. distribution of extant species of Diadema is gener- ally (though not completely) consistent with allo- patric speciation. 1.5 Lytechinus Another genus in which the distributions of extant clades are mostly, but not entirely, consistent with a model of allopatric speciation is Lytechinus. (Fig. 5). This genus is confined to the coasts of America, with only one species at the Cape Verde Islands. A combined mtDNA and bindin phylogeny by Zigler & Lessios (2004) (which does not include the Cape Verde species) has found that the deep water L. euerces is distantly related to the other species of Lytechinus and is best considered as a member of a different genus. The first split among the remaining species was probably due to the Isthmus of Panama. The eastern Pacific clade was subsequently divided into north and south clades, most likely as the result of the muddy inhospitable habitat stretching between S. Mexico and Costa Rica. There is no distinction in either COI or bindin between L. anamesus and L. pictus in the North, nor between L. panamensis and L. semitubreculatus in the South. The former pair has also been shown to represent different ecotypes of the same species by other evidence (Clark 1940, Cameron 1984). The status of L. panamensis as a separate species cannot be determined until the coast of Ecuador has been sampled (Lessios 2005). The Atlantic clade of Lytechinus challenges the allopatric model of speciation. The COI of L. williamsi and L. variegatus variegatus is very similar, and so are their isozymes; their bindin alleles, however, are re- ciprocally monophyletic, even though they differ in only 4 amino acids (Zigler & Lessios 2004). Thus, in a phylogeny of these taxa based on both bindin and COI, L. williamsi is actually nested within L. variegatus, with L. variegatus carolinus as an out- group (Fig. 5), whereas in a phylogeny based on bindin alone it is sister to the entire L. variegatus complex (Zigler & Lessios 2004). The monophyly of bindin, the existence of diagnostic morphological characters (Chesher 1968, Lessios, unpubl.) and the marked difference in adult size of the two species indicate that L. williamsi is not a juvenile form of L. variegatus, but a separate species. The difficulty for the allopatric speciation model, when Jordan?s rule is applied, is that these two closely related sister species are sympatric. It is possible that they speci- ated in allopatry, and their genetic similarities accu- mulated after secondary contact through hybridiza- tion. Although they occupy different habitats (table 4 in Lessios et al. 1984), the distance between them is not so large as to prevent cross-fertilization. How- ever, hybridization should have resulted in exchange of bindin alleles, because there is no evidence that this molecule is under selection in Lytechinus (Zigler & Lessios 2004). Thus, the possibility of sympatric speciation cannot be excluded on the basis of phylogeography in the case of the two Caribbean sister species of Lytechinus. 1.6 Echinometra Echinometra is a genus in which Jordan?s rule breaks down completely as a means of deducing the spatial mode of speciation. COI phylogenies by Palumbi et al. (1997) and by Landry et al. (2003) for the Indo-West Pacific newly discovered and as yet unnamed species and by McCartney et al. (2000) for the eastern Pacific and Atlantic species indicate a great deal of spatial overlap between sister species (Fig. 6). Bindin phylogenies (Metz and Palumbi 1996, Geyer & Palumbi 2003, McCartney & Lessios 2004) are generally congruent with the mtDNA phy- logenies. E. mathaei coexists with one of the two clades of E. oblonga from which it is separated for about one my and E. viridis coexists with E. Figure 5. Phylogeny of the extant species of Lytechinus, based on sequences from the Cytochrome Oxidase I (COI) region of mito- chondrial DNA and from bindin, and geographic distribution of the clades (Zigler & Lessios 2004). The thick bar indicates the presumed break arising from the emergence of the Isthmus of Pa- nama, 3.1 million years (my) ago. Numbers next to the nodes in- dicate approximate dates (in my) of most recent common ances- tor, based on a calibration of COI divergence by the emergence of the Isthmus. Actual locations in which the samples were collected are shown on the map. lucunter from which it diverged approximately 1.5 my ago. E. sp. A, which has been a separate species from E. mathaei and E. oblonga for about 1.2 my, is found with one of these species in just about every locality. In both the Indo-West Pacific (Tsuchiya & Nishihira 1984, Nishihira et al. 1991, Rahman & Uehara 2004) and in the Caribbean (McCartney & Lessios 2004) there is habitat separation between the sympatric species, but this separation could not be the barrier that caused interruption of gene flow and speciation, because habitat preference is an intrinsic characteristic of each species, subject to evolution, rather than an obstacle presented by the environ- ment. What is more, the only geographic barriers that can be deduced from the phylogeny are the Eastern Pacific Barrier, which accounts for the first split in the genus; the Isthmus of Panama, which separated E. vanbrunti from the two Atlantic spe- cies; and the geographic isolation of Easter Island, which presumably accounts for the speciation of the endemic E. insularis. Pleistocene sea-level fluctua- tions may have caused the separation of the rest of the clades, but it would be impossible to specify ex- actly how, where, or when. 1.7 Strongylocentrotus Yet another genus in which sympatric speciation cannot be excluded on the basis of phylogeography alone (and the only example from higher latitudes) is Strongylocentrotus (Fig. 7). MtDNA phylogenies by Biermann et al. (2003) and by Lee (2003), as well as a bindin phylogeny by Biermann (1998), indicate that this genus should also include Hemicentrotus pulcherrimus and Allocentrotus fragilis. There is a deep phylogenetic break in this genus, presumably caused by distance, between species that are limited to the western Pacific and species that are found in the eastern Pacific and the Atlantic. The geographi- cal distribution of the species within each of these major clades shows that sister species are not always allopatric. Specifically the sister species H. pulcher- rimus and S. intermedius, both inhabiting shallow water, are sympatric in the Sea of Japan; S. purpura- tus, A. fragilis, S. pallidus and S. droebachiensis overlap in the eastern Pacific. S. pallidus and S. droebachiensis are also sympatric on both shores of the Atlantic. It can be argued that sister species Figure 6. Phylogeny of the extant species of Echinometra, based on sequences from the Cytochrome Oxidase I (COI), and geo- graphic distribution of the clades (McCartney et al. 2000, Landry et al. 2003). The thick bar indicates the presumed break arising from the emergence of the Isthmus of Panama, 3.1 million years (my) ago. Numbers next to the nodes indicate approximate dates (in my) of most recent common ancestor, based on a calibration of COI divergence by the emergence of the Isthmus. Actual loca- tions in which the samples were collected are shown on the map. Smaller circles indicate that only occasional haplotypes of a clade inhabit a particular area. status is not of importance, as unknown extinctions can cause extant species to appear as sister clades, when in fact they did not speciate from each other, and that determinations on whether Jordan?s rule supports allopatric speciation should be made on the basis of divergence time alone. By this criterion, the H. pulcherrimus-S. intermedius pair does not falsify the rule, because the timing of the split exceeds 3 my. Similarly, the relationship of S. purpuratus to the tritomy of three species with which it is partially sympatric can also be excluded on the basis of the antiquity of the split. It is harder to claim, however, that phylogeography rejects the hypothesis of sym- patric speciation among A. fragilis, S. pallidus and S. droebachiensis. It is true that the three species show only minimal overlap in their bathymetric distribu- tions, but, as argued previously for habitat prefer- ence in Echinometra, depth zonation alone cannot be considered a barrier, because one still needs to ask how each species evolved to be adapted to a particu- lar depth zone. Thus, the combination of phylogeny and geo- graphic distribution of species in each genus sug- gests that many speciation events in echinoids are the result of allopatric speciation, but leaves some cases in Lytechinus, Echinometra and Strongylocen- trotus in which sympatric speciation is a possibility. Echinoids thus illustrate the limits of Jordan?s rule and of phylogeography in helping deduce the spatial mode of speciation events. An important problem with Jordan?s rule is that it is based on the assump- tion that reproductive isolation accumulates gradu- ally over time, so older species will be better iso- lated than young ones. But does this assumption hold in echinoids? 2 REPRODUCTIVE ISOLATION Jordan (1905) and Mayr (1954) tacitly assumed that reproductive isolation is the product of many small changes in many loci, and thus that speciation re- quires extended periods of geographic isolation to be completed. If, however, reproductive isolation can arise ?accidentally? independently of divergence time, then recently separated species can become re- productively incompatible in either allopatry or sympatry, and anciently separated species may re- main compatible. What do we know about the rela- tionship between time of separation and the emer- gence of reproductive isolation in echinoids? The possible reproductive barriers between species of sea urchins have been reviewed by Lessios (2007). One conclusion of this paper is that there are two repro- ductive isolation barriers in echinoids for which the relationship between species isolation and diver- Figure 7. Phylogeny of the extant species of Strongylocentrotus, based on sequences from the Cytochrome Oxidase I (COI), and geographic distribution of the clades (Biermann et al. 2003). Numbers next to the nodes indicate approximate dates (in my) of most recent common ancestor, based on a calibration of COI divergence by the emergence of the Isthmus of Panama in other gen- era. Actual locations in which the samples were collected are shown on the map. . gence time can be examined: (a) Postzygotic isola- tion in the form of reduced survivorship of hybrids or inability to back-cross to their parental species, and (b) prezygotic isolation in the form of gamete incompatibility. Figure 8 presents the relationship between time since speciation and postzygotic isolation, as deter- mined by tabulating data about hybrid crosses from different genera, and estimating their divergence times from COI differentiation. Because different studies present hybrid survival information differ- ently (only the studies on Echinometra include quantitative data), it was necessary, for the purposes of summary depiction, to construct an arbitrary ?in- dex of hybrid fitness?, as a composite measure of the survivorship of hybrid larvae produced from two crosses (one with eggs of one species and sperm from a second one, the other from the reverse cross), as well as the ability of the hybrids to produce viable offspring when back-crossed to the parentals. The data come from only four genera, Echinometra, Strongylocentrotus, Heliocidaris, and Pseudechinus. The divergence times mainly fall in two clusters: Those from the Indo-Pacific species of Echinometra and from the cross between Pseudechinus huttoni with P. albocinctus, with estimated ages of 1-2 my, and those of Pseudechinus, in which one species, P. novaezealandiae, split from the other two congeners, approximately 7 my ago. Only Heliocidaris has an intermediate divergence time. Because of multiple comparisons, the points are also not independent from each other, which precludes statistical infer- ences. Thus the available evidence is far from incon- trovertible, but it does support the idea that time is a good predictor of the degree to which species will develop developmental incompatibilities. Older spe- cies are better isolated than younger ones, and sev- eral million years are required for post-zygotic isola- tion that is measurable in captive animals and would prevent gene flow between sympatric species Prezygotic isolation, on the other hand, does not obey the same rules as hybrid inviability. Zigler et al. (2005) compiled data from 15 comparisons of congeneric sea urchin species for which data were available on gametic compatibility, COI divergence, and bindin divergence. They found that there is no correlation between compatibility among gametes of different species and COI divergence. Thus-- assuming a molecular clock--preference of gametes to combine with those of their own species does not evolve as a function of time alone. Young species, such as Echinometra mathaei and E. sp. A, separated for slightly more than a million years, have gametes incapable of fertilizing each other, whereas much older species, such as Arbacia punctulata and A. in- cisa, separated for more than 4 million years have gametes that can fertilize each other at the same rates as gametes of their own species. Zigler et al. (2005) also found that gametic in- compatibility was correlated with divergence in bindin. Bindin is a molecule that covers the acro- some process of the sperm, binds with a receptor on the vitelline layer, then fuses with the egg membrane to permit transfer of sperm DNA into the egg. Al- though it is not yet clear what modifications of the bindin and the bindin receptor molecules are neces- sary to confer incompatibility between gametes, the correlation with time indicates that study of these molecules can shed light on some of the factors in- volved in the evolution of pre-zygotic isolation be- tween sea urchin species. Very little is known about the bindin receptor, because it is a large molecule difficult to sequence (Kamei & Glabe 2003), but there is a modest accumulation of comparative facts about bindin. These are reviewed in Lessios (2007) and will be presented here only briefly. In addition to the correlation between gametic in- compatibility and bindin divergence, we also know that in Echinometra most amino acid replacements accumulate in the species with eggs which do not permit fertilization by heterospecific sperm, and that three genera with sympatric species have bindins that show evidence of positive selection, whereas Figure 8. Relationship between fitness of interspecific hybrids and time since parental species split from each other. The hy- brid fitness index is an arbitrary value, based on information provided in the original publications regarding the survivor- ship of F1 hybrids and their ability to back-cross to the par- ents. See Lessios (2007) for a tabulation of the data, and for the sources of COI differentiation, used here to estimate di- vergence time. Explanation of symbols: 1: Echinometra Sp. A vs. E. oblonga (Aslan & Uehara, 1997); 2: E. mathaei vs. E. sp. A (Rahman et al. 2005); 3: E. sp. C vs. E. mathaei (Rah- man & Uehara 2004); 4: Strongylocentrotus pallidus vs. S. droebachiensis (Strathmann 1981); 5: Pseudechinus huttoni vs. P. albocinctus (McClary & Sewell 2003); 6: Heliocidaris tuberculata vs. H. erythrogramma (Raff et al. 1999); 7: Pseu- dechinus novaezealandiae vs. P. albocinctus; 8: P. novaezea- landiae vs. P. huttoni (McClary & Sewell 2003); 9: Echi- nometra sp. A vs. E. sp. C (Rahman et al. 2001). three other genera with no sympatric species have bindins that do not share these features (Metz & Palumbi 1996, Biermann 1998, Metz et al. 1998, Zigler et al. 2003, Zigler & Lessios 2003 2004, McCartney & Lessios, 2004). There is, therefore, selection on bindin to track changes that occur in the egg receptor, and one may well think that this selec- tion is avoidance of hybrid production (reinforce- ment). Corroborating this hypothesis is the geo- graphical pattern of character displacement found by Geyer & Palumbi (2003) in Echinometra oblonga, which has bindin alleles different than those of E. sp. C where the two species coexist, but similar to those of E. sp. C where it found alone. Not all evi- dence, however, is consistent with the hypothesis that the selective force acting on bindin is rein- forcement. In both Echinometra and Strongylocentrotus an excess of amino acid replacement over silent substi- tutions between alleles of the same species suggests that there is positive selection not just for divergence between species, but also for intraspecific polymor- phism. Selection to avoid hybridization with other species cannot generate adaptive differences be- tween alleles of the same species. Further evidence against the hypothesis that reinforcement drives the evolution of echinoid bindin comes from failures to find a pattern of character displacement in either the Atlantic Echinometra, or the Australian Heliocidaris (Lessios 2007). In both cases, bindin alleles of popu- lations that are not sympatric with a congener and do not receive gene flow from the area of overlap are similar to the those in the area of overlap. It is thus possible that the pattern of reinforcement suggested by the comparison of the mode of evolution of bindin in different genera does not result from selec- tive pressures created by the challenge of sympatric species. Instead, it may be a secondary effect of bindin divergence that has accumulated between species by intraspecific forces. If only species with bindins that have diverged for other reasons are able to coexist sympatrically without merging, then we would also expect to see a pattern of correlation be- tween range overlap and bindin divergence. The intraspecific forces of selection that may drive bindins to diverge in some species but not oth- ers are likely to be related to sexual selection (West- Eberhard 1983) and inter-locus sexual conflict (Rice 1998, Gavrilets 2000). In a mating system such as bindin and its receptor, frequencies of different al- leles can become rapidly predominant in different populations, because of linkage disequilibrium. When bindin alleles are preferred by different egg receptor alleles in each generation, there will be an over-representation of offspring in which male and female alleles are matched. If different combinations become predominant in different geographically iso- lated populations, they will lead to reproductive iso- lation. Assortative mating in bindin has been dem- onstrated experimentally in Echinometra mathaei by Palumbi (1999). Levitan and Ferrell (2006) have shown in Strongylocentrotus franciscanus that re- productive success of bindin alleles depends on local sperm density. Eggs of females that carry rare bindin alleles are more likely to be fertilized than those car- rying common alleles when sperm density is high, but the reverse is true at limiting sperm densities. If there is linkage disequilibrium between bindin and receptor loci, different combinations of alleles will predominate in each species, depending on sperm density, which is affected by habitat specialization and prevailing oceanographic conditions. In Echi- nometra (McCartney & Lessios 2004) and also in Strongylocentrotus (Biermann 1998, Debenham et al. 2000, Levitan & Ferrell 2006) species that are found in shallow water in high point densities ?and are thus likely to experience high sperm density dur- ing spawning events? are also species that show the strongest selection on bindin. We do not know enough about other molecules that may be involved in species recognition between sperm and egg (Biermann et al. 2004, Neill & Vacquier 2004, Mah et al. 2005, Kaupp et al. 2006) to determine whether similar selection forces as those of bindin may be acting on them. But if what we do know about bindin is indicative of how gametic isolation evolves in sea urchins, it is not surprising that prezygotic iso- lation between species is not a simple function of time. 3 CONCLUSIONS How do new species of sea urchins arise? The exist- ing data are not adequate to provide a complete an- swer to this question (a statement that holds true for any group of organisms), but we do know a great deal more than what was available to Mayr in 1954. The combined information of phylogeny and species distribution has made it possible to ask whether young species tend to be still separated by extrinsic barriers, as the allopatric model would predict if re- productive isolation requires long periods of time to develop. We have phylogeographic data for only shallow water species and mostly for tropical ones, but the answers they have provided will likely be applicable to all species of echinoids: The majority of speciation events appear to be allopatric, but there are some cases of closely related sympatric sister species that raise the possibility of sympatric speci- ation. Indeed, if reproductive isolation evolved only as a function of time, the young sympatric species of Echinometra, Strongylocentrotus and Lytechinus would be evidence that sympatric speciation does occur in echinoids. Data on postzygotic isolation limited as they are, suggest that developmental in- compatibilities in hybrids between species increase steadily with the time that the parental species have been separated. The same, however, is not true for prezygotic isolation. Gametic compatibility and bindin divergence can appear rapidly in some spe- cies and can be absent in others, despite much longer times spent in allopatry. This makes Jordan?s rule an inadequate tool for assessing the possibility of sym- patric speciation, because it does not allow the falsi- fication of the null hypothesis of allopatric speci- ation. There is, however, good reason for doubting that new species can appear in sympatry through mutations that will alter gametic compatibility. Indi- viduals with gametes that cannot combine with those of the opposite sex will be rapidly eliminated from a population, unless an extrinsic barrier happens to isolate them in a small population in which some other deviants are present. Only under very special conditions, stipulated by models reviewed in Coyne & Orr (2004) and Gavrilets (2004), can different species arise sympatrically through sexual selection in the absence of a barrier. That the emergence of prezygotic isolation can be rapid in geographically separated populations due to conditions such as sperm density, does not necessarily mean that it will also happen often. There are, after all, less than 900 described extant species in the entire class Echinoidea, fewer than in many families of more speciose groups, despite the very solid foundation of echinoid alpha taxonomy. Barring an unknown reason for ex- traordinary rates of echinoid extinction, the paucity of extant sea urchin species suggests that speciation is a rare event in this class of animals. 4 ACKNOWLEDGMENTS I thank Larry Harris for the invitation to present a plenary lecture in the 12th International Echinoderm Conference, and L. Geyer for comments on the manuscript. REFERENCES Aslan, L. M. & Uehara, T. 1997. Hybridization and F1 back- crosses between two closely related tropical species of sea urchins (genus Echinometra) in Okinawa. Invertebrate Re- production & Development. 31: 319-324. Biermann, C. H. 1998. The molecular evolution of sperm bindin in six species of sea urchins (Echinoida : Strongylo- centrotidae). Molecular Biology and Evolution. 15: 1761- 1771. Biermann, C. H., Kessing, B. D. & Palumbi, S. R. 2003. Phy- logeny and development of marine model species: Strongy- locentrotid sea urchins. Evolution & Development. 5: 360- 371. Biermann, C. H., Marks, J. A., VilelaSilva, A. C. E. S., Castro, M. O. & Mourao, P. A. S. 2004. Carbohydrate-based spe- cies recognition in sea urchin fertilization: another avenue for speciation? Evolution & Development. 6: 353-361. Briggs, J. C. 1974. Marine zoogeography. New York: McGraw-Hill. Cameron, R. A. 1984. Two species of Lytechinus (Toxopneustidae: Echionidea: Echinodermata) are com- pletely cross-fertile. Bull. South. Calif. Acad. Sci. 83: 154- 157. Chesher, R. H. 1968. Lytechinus williamsi, a new sea urchin from Panama. Breviora. 305: 1-13. Clark, H. L. 1940. Eastern Pacific Expeditions of the New York Zoological Society. XXI. Notes on Echinoderms from the west Coast of central America. Zoologica: New York Zoological Society. 25: 331-352. Coyne, J. A. & Orr, H. A. 2004. Speciation. Sunderland, Mass: Sinauer. Debenham, P., Brzezinski, M. A. & Foltz, K. R. 2000. Evalua- tion of sequence variation and selection in the bindin locus of the red sea urchin, Strongylocentrotus franciscanus. Journal of Molecular Evolution. 51: 481-490. Dobzhansky, T. 1937. Genetics and the origin of species. New York: Columbia Univ. Press. Dobzhansky, T. 1970. Genetics of the evolutionary process. New York: Columbia Univ. Press. Ekman, S. 1953. Zoogeography of the sea. London: Sidgwick and Jackson Ltd. Gavrilets, S. 2000. Rapid evolution of reproductive barriers driven by sexual conflict. Nature. 403: 886-889. Gavrilets, S. 2004. Fitness landscapes and the origin of spe- cies. Princeton, NJ: Princeton University Press. Geyer, L. B. & Palumbi, S. R. 2003. Reproductive character displacement and the genetics of gamete recognition in tropical sea urchins. Evolution. 57: 1049-1060. Howard, D. J. & Berlocher, S. (eds.) 1998. Endless forms: spe- cies and speciation. Oxford: Oxford Univ. Press. Jordan, D. S. 1905. The origin of species through isolation. Science. 22: 545-562. Kamei, N. & Glabe, C. G. 2003. The species-specific egg re- ceptor for sea urchin sperm adhesion is EBR1, a novel ADAMTS protein. Genes & Development. 17: 2502-2507. Kaupp, U. B., Hildebrand, E. & Weyand, I. 2006. Sperm chemotaxis in marine invertebrates: molecules and mecha- nisms. Journal of Cellular Physiology. 208: 487-494. Landry, C., Geyer, L. B., Arakaki, Y., Uehara, T. & Palumbi, S. R. 2003. Recent speciation in the Indo-West Pacific: rapid evolution of gamete recognition and sperm morphol- ogy in cryptic species of sea urchin. Proceedings of the Royal Society of London Series B -Biological Sciences. 270: 1839-1847. Lee, Y. H. 2003. Molecular phylogenies and divergence times of sea urchin species of Strongylocentrotidae, Echinoida. Molecular Biology and Evolution. 20: 1211-1221. Lessios, H. A., Cubit, J. D., Robertson, D. R., Shulman, M. J., Parker, M. R., Garrity, S. D. & Levings, S. C. 1984. Mass mortality of Diadema antillarum on the Caribbean coast of Panama. Coral Reefs. 3: 173-182. Lessios, H. A., Kessing, B. D., Robertson, D. R. & Paulay, G. 1999. Phylogeography of the pantropical sea urchin Euci- daris in relation to land barriers and ocean currents. Evolu- tion. 53: 806-817. Lessios, H. A., Kessing, B. D. & Pearse, J. S. 2001. Population structure and speciation in tropical seas: Global phy- logeography of the sea urchin Diadema. Evolution. 55: 955-975. Lessios, H. A., Kane, J. & Robertson, D. R. 2003. Phy- logeography of the pantropical sea urchin Tripneustes: Contrasting patterns of population structure between oceans. Evolution. 57: 2026-2036. Lessios, H. A. 2005. Echinoids of the Pacific Waters of Pa- nama: Status of knowledge and new records. Revista de Biologia Tropical. 53: 147-170. Lessios, H. A. 2007. Reproductive ecology and reproductive isolation in sea urchins. Bulletin of Marine Science. in press. Levitan, D. R. & Ferrell, D. L. 2006. Selection on gamete rec- ognition proteins depends on sex, density, and genotype frequency. Science. 312: 267-269. Mah, S. A., Swanson, W. J. & Vacquier, V. D. 2005. Positive selection in the carbohydrate recognition domains of sea urchin sperm receptor for egg jelly (SuREJ) proteins. Mo- lecular Biology and Evolution. 22: 533-541. Maluf, L. Y. 1991. Echinoderm fauna of the Galapagos Is- lands. In M. J. James (eds), Galapagos marine inverte- brates: 345-367. New York: Plenum Press. Mayr, E. 1942. Systematics and the origin of species. New York: Columbia Univ. Press. Mayr, E. 1954. Geographic speciation in tropical echinoids. Evolution. 8: 1-18. Mayr, E. 1963. Animal species and evolution. Cambridge, Mass.: Harvard University Press. Mayr, E. 1970. Populations, species, and evolution. Cam- bridge, Massachusetts: Belknnap Press. McCartney, M. A., Keller, G. & Lessios, H. A. 2000. Dispersal barriers in tropical oceans and speciation of Atlantic and eastern Pacific Echinometra sea urchins. Molecular Ecol- ogy. 9: 1391-1400. McCartney, M. A. & Lessios, H. A. 2004. Adaptive evolution of sperm bindin tracks egg incompatibility in neotropical sea urchins of the genus Echinometra. Molecular Biology and Evolution. 21: 732-745. McClary, D. J. & Sewell, M. A. 2003. Hybridization in the sea: gametic and developmental constraints on fertilization in sympatric species of Pseudechinus (Echinodermata: Echin- oidea). Journal of Experimental Marine Biology and Ecol- ogy. 284: 51-70. Metz, E. C. & Palumbi, S. R. 1996. Positive selection and se- quence rearrangements generate extensive polymorphism in the gamete recognition protein bindin. Molecular Biology and Evolution. 13: 397-406. Metz, E. C., G?mez-Guti?rrez, G. & Vacquier, D. 1998. Mito- chondrial DNA and bindin gene sequence evolution among allopatric species of the sea urchin genus Arbacia. Molecu- lar Biology and Evolution. 15: 185-195. Mortensen, T. 1928-1951. A monograph of the Echinoidea. Copenhagen: C.A. Reitzel. Neill, A. T. & Vacquier, V. D. 2004. Ligands and receptors mediating signal transduction in sea urchin spermatozoa. Reproduction. 127: 141-149. Nishihira, M., Sato, Y., Arakaki, Y. & Tsuchiya, M. 1991. Ecological distribution and habitat preference of four types of the sea urchin Echinometra mathaei on the Okinawan coral reefs. In T. Yanagisawa, I. Yasumasu, C. Oguro, N. Suzuki &T. Motokawa (eds), Biology of Echinodermata: 91-104. Balkema. Otte, D. & Endler, J. A. 1989. Speciation and its consequences. Sunderland, Mass: Sinauer. Palumbi, S. R., Grabowsky, G., Duda, T., Geyer, L. & Tachino, N. 1997. Speciation and population genetic struc- ture in tropical Pacific sea urchins. Evolution. 51: 1506- 1517. Palumbi, S. R. 1999. All males are not created equal: fertility differences depend on gamete recognition polymorphisms in sea urchins. Proceedings of the National Academy of Sciences US. 96: 12632-12637. Palumbi, S. R. & Lessios, H. A. 2005. Evolutionary animation: How do molecular phylogenies compare to Mayr's recon- struction of speciation patterns in the sea? Proceedings of the National Academy of Sciences of the United States of America. 102, Suppl. 1: 6566-6572. Raff, E. C., Popodi, E. M., Sly, B. J., Turner, F. R., Villinski, J. T. & Raff, R. A. 1999. A novel ontogenetic pathway in hy- brid embryos between species with different modes of de- velopment. Development. 126: 1937-1945. Rahman, M. A., Uehara, T. & Pearse, J. S. 2001. Hybrids of two closely related tropical sea urchins (Genus Echi- nometra): Evidence against postzygotic isolating mecha- nisms. Biological Bulletin. 200: 97-106. Rahman, M. A. & Uehara, T. 2004. Interspecific hybridization and backcrosses between two sibling species of Pacific sea urchins (Genus Echinometra) on Okinawan intertidal Reefs. Zoological Studies. 43: 93-111. Rahman, M. A., Uehara, T. & Pearse, J. S. 2004. Experimental hybridization between two recently diverged species of tropical sea urchins, Echinometra mathaei and Echinometra oblonga. Invertebrate Reproduction & Development. 45: 1- 14. Rahman, M. A., Uehara, T. & Lawrence, J. M. 2005. Growth and heterosis of hybrids of two closely related species of Pacific sea urchins (Genus Echinometra) in Okinawa. Aquaculture. 245: 121-133. Rice, W. R. 1998. Intergenomic conflict, interlocus antagonis- tic coevolution, and evolution of reproductive isolation. In D. J. Howard &S. H. Berlocher (eds), Endless forms. Spe- cies and speciation: 261?270. New York: Oxford Univer- sity Press. Strathmann, R. R. 1981. On barriers to hybridization between Strongylocentrotus droebachiensis (O.F. Muller) and S. pallidus (G.O. Sars). Journal of Experimental Marine Biol- ogy and Ecology. 55: 39-47. Tsuchiya, M. & Nishihira, M. 1984. Ecological distribution of two types of the sea-urchin, Echinometra mathaei (Blain- ville), on Okinawan reef flat. Galexea. 3: 131-143. West-Eberhard, M. J. 1983. Sexual selection, social competi- tion, and speciation. Quarterly Review of Biology. 58: 155- 183. Zigler, K. S. & Lessios, H. A. 2003. Evolution of bindin in the pantropical sea urchin Tripneustes: Comparisons to bindin of other genera. Molecular Biology and Evolution. 20: 220- 231. Zigler, K. S., Raff, E. C., Popodi, E., Raff, R. A. & Lessios, H. A. 2003. Adaptive evolution of bindin in the genus Helio- cidaris is correlated with the shift to direct development. Evolution. 57: 2293-2302. Zigler, K. S. & Lessios, H. A. 2004. Speciation on the coasts of the new world: Phylogeography and the evolution of bindin in the sea urchin genus Lytechinus. Evolution. 58: 1225- 1241. Zigler, K. S., McCartney, M. A., Levitan, D. R. & Lessios, H. A. 2005. Sea urchin bindin divergence predicts gamete compatibility. Evolution. 59: 2399-2404.