? 2003 The Society for the Study of Evolution. All rights reserved.
Evolution, 57(10), 2003, pp. 2293–2302
ADAPTIVE EVOLUTION OF BINDIN IN THE GENUS HELIOCIDARIS IS CORRELATED
WITH THE SHIFT TO DIRECT DEVELOPMENT
KIRK S. ZIGLER,1,2,3ELIZABETH C. RAFF,4,5ELLEN POPODI,4,6RUDOLF A. RAFF,4,7AND H. A. LESSIOS1,8
1Smithsonian Tropical Research Institute, Box 2072, Balboa, Panama
2Department of Biology, Duke University, Durham, North Carolina 27708
4Department of Biology and Molecular Biology Institute, Indiana University, Bloomington, Indiana 47405
fields to examine the evolution of reproductive isolation in the genus Heliocidaris. Heliocidaris tuberculata develops
indirectly via a feeding larva, whereas the only other species in the genus, H. erythrogramma, has evolved direct
development through a nonfeeding larva. We estimated the time of divergence between H. erythrogramma and H.
tuberculata from mitochondrial DNA divergence, quantified levels of gametic compatibility between the two species
in cross-fertilization assays, and examined the mode of evolution of the sperm protein bindin by sequencing multiple
alleles of the two species. Bindin is the major component of the sea urchin sperm acrosomal vesicle, and is involved
in sperm-egg attachment and fusion. Based on our analyses, we conclude that: the two species of Heliocidaris diverged
less than five million years ago, indicating that direct development can evolve rapidly in sea urchins; since their
divergence, the two species have become gametically incompatible; Heliocidaris bindin has evolved under positive
selection; and this positive selection is concentrated on the branch leading to H. erythrogramma. Three hypotheses
can explain the observed pattern of selection on bindin: (1) it is a correlated response to the evolution of direct
development in H. erythrogramma; (2) it is the result of an intraspecific process acting in H. erythrogramma but not
in H. tuberculata; or (3) it is the product of reinforcement on the species that invests more energy into each egg to
Sea urchins are widely used to study both fertilization and development. In this study we combine the two
Evolution of development, fertilization, gamete recognition, sea urchin.
Received November 11, 2002. Accepted April 25, 2003.
Studies on sea urchins have contributed greatly to our un-
derstanding of developmental biology, gamete interactions,
and the process of fertilization. The evolution of proteins
involved in gamete interactions has also received consider-
able attention (review in Vacquier 1998). One of these pro-
teins is the sea urchin sperm protein bindin, which is the
major component of the sea urchin sperm acrosomal vesicle
and is involved in sperm-egg attachment and fusion (Vac-
quier and Moy 1977; Ulrich et al. 1998). Changes in bindin
can thus contribute to the evolution of reproductive isolation
and speciation. Bindin evolves under positive selection in
some (Metz and Palumbi 1996; Biermann 1998), but not all
(Metz et al. 1998; Zigler and Lessios 2003a), genera of sea
urchins. The selective forces acting on bindin are still unclear,
but both avoidance of maladaptive hybridization (reinforce-
ment) and various intraspecific mechanisms, such as male
heterozygote advantage, nontransitive female preferences, or
interlocus conflict evolution have been suggested (Metz et
al. 1998; Palumbi 1999).
Some sea urchins develop indirectly from feeding larvae,
whereas others develop directly from nonfeeding larvae. Be-
cause of extensive similarities between the sea urchin feeding
larva and those of other echinoderms, development via a
feeding larva is thought to be the ancestral mode of devel-
opment in echinoids (Strathmann 1978). Although this an-
cestral pattern of development has been preserved in the ma-
3Present address: Friday Harbor Laboratories, 620 University
Road, Friday Harbor, Washington 98250; E-mail: ziglerk@
jority of echinoid species, direct development has evolved at
least 20 times in echinoids (Emlet 1990; Wray 1996). The
two species of the genus Heliocidaris are the best-studied
example of this transition. Heliocidaris tuberculata follows
the ancestral mode of developing from small (95 ?m di-
ameter) eggs via a feeding larva, whereas H. erythrogramma
has evolved direct development from large (430 ?m) eggs,
bypassing the planktonic feeding stage (Raff 1987). Corre-
lated with the increase in H. erythrogramma egg size are
radical differences in development between the two species,
including cleavage pattern and cell lineage, axial specifica-
tion, morphogenesis, and gene expression (e.g., Wray and
Raff 1989; Henry and Raff 1990; Henry et al. 1991; Emlet
1995; Haag et al. 1999; Raff and Sly 2000).
In addition to the more than 100-fold difference in dry
organic mass per egg (Hoegh-Guldberg and Emlet 1997),
there are other important differences in gametogenesis be-
tween the two species of Heliocidaris. The small eggs of H.
tuberculata are predominantly provisioned with vitellin pro-
tein (yolk) and triglyceride lipids, whereas the large eggs of
H. erythrogramma show a profound decrease in yolk content
and a great increase in wax ester content (Scott et al. 1990;
Byrne et al. 1999; Villinski et al. 2002). Sperm size is also
correlated with developmental mode; direct-developers gen-
erally have longer sperm heads than related indirect-devel-
opers (Eckelbarger et al. 1989; Raff et al. 1990). This holds
true in Heliocidaris, where H. tuberculata sperm have heads
5.6 ?m long and H. erythrogramma sperm heads 11 ?m long
(Raff et al. 1990).
The two Heliocidaris species coexist in intertidal and shal-
KIRK S. ZIGLER ET AL.
low subtidal rocky habitats on the southeast coast of Australia
(Mortensen 1943; Keesing 2001). It is unclear how long ago
these two species diverged, as estimates vary from five to 13
million years ago (Smith et al. 1990; McMillan et al. 1992).
Heliocidaris erythrogramma and H. tuberculata have partially
overlapping reproductive seasons (Laegdsgaard et al. 1991),
and their gametes can be crossed in the laboratory (Raff et
al. 1999), but the efficiency of cross fertilizations has not
In this study we present data on divergence time, gametic
compatibility, and mode of evolution of bindin between the
two species of Heliocidaris. To estimate the divergence time
between H. erythrogramma and H. tuberculata, we sequenced
part of the gene for cytochrome oxidase I (COI) from the
two species. We quantified levels of gametic compatibility
between the two species of Heliocidaris by fertilization ef-
ficiency assays. We sequenced part of the gene for bindin
from the two species to examine its relation to reproductive
isolation and the possibility that its evolution is related to
developmental mode. Previous studies of gamete recognition
protein evolution in marine invertebrates have involved spe-
cies with planktonic development; this study is the first com-
parison of gamete recognition protein evolution between spe-
cies with different developmental modes.
MATERIALS AND METHODS
Heliocidaris erythrogramma and H. tuberculata were col-
lected near Sydney, Australia. Additional H. erythrogramma
individuals were collected from Hobart, Tasmania, Australia.
Individual gonads were stored in 95% ethyl alcohol and re-
frigerated until use.
Cytochrome Oxidase I Sequencing, Phylogeny,
and Genetic Distances
DNA was extracted as described Lessios et al. (1996). A
640-bp section of the mitochondrial COI gene was amplified
and sequenced from seven individuals of H. erythrogramma
and four of H. tuberculata, using primers COIa (5?-AGTA-
GGAGGAGGAGAYCC-3?). We rooted the Heliocidaris COI
sequences with four randomly chosen COI sequences from
each of the seven Echinometra species for which COI data
were available in GenBank. Four of these species are from
the Indo-West Pacific (E. type A, E. mathaei, E. type C, and
E. oblonga; Palumbi et al. 1997) and three from the Americas
(E. viridis, E. lucunter, and E. vanbrunti; McCartney et al.
2000). Heliocidaris and Echinometra are both members of
the subfamily Echinometrinae (Smith 1988). We used Mod-
eltest version 3.06 (Posada and Crandall 1998) to identify
the simplest model of DNA evolution that best describes the
data under maximum likelihood (ML). The model identified
was that of Tamura and Nei (1993). Inclusion of site-specific
rates in place of a gamma correction or invariable sites pro-
duced a much larger likelihood value, so these rates were
included. We used PAUP* version 4.0b10 (Swofford 2001)
to construct a neighbor-joining tree based on the estimated
ML parameters (transversions ? 1, A-G ? 7.6447, C-T ?
13.0757; relative rates: first position ? 0.1547, second po-
sition ? 0.0405, third position ? 2.7962) . The tree was
bootstrapped in 1000 iterations. We tested for variation in
substitution rates between taxa by comparing the likelihoods
of the best ML tree with and without a molecular clock en-
forced (Felsenstein 1981). Heliocidaris COI sequences have
been depositedinto GenBank
Fertilization Efficiency Assays
Shed eggs were washed, and a subset was test-fertilized
with homospecific sperm to ensure that they exhibited normal
fertilization. Heliocidaris erythrogramma eggs have a thick
jelly coat; for fertilization tests of dejellied H.erythrogramma
eggs, the eggs were dejellied after washing by brief treatment
with pH 5 seawater as previously described (Raff et al. 1999).
Fresh sperm were diluted in 10-fold serial dilutions in filtered
seawater; sperm concentration was determined from para-
formaldehyde-fixed samples of sperm suspensions by count-
ing on a hemocytometer slide. Washed eggs were aliquoted
in 2 ml of filtered seawater in each well of 24-well plastic
microtiter dishes. The large, lipid-rich H. erythrogramma
eggs float, whereas the small H. tuberculata eggs sink, as is
typical for eggs from indirect-developing sea urchins. Be-
cause the size and buoyancy of eggs from the two species is
so different, no attempt was made to use the same concen-
tration of eggs in the fertilization experiments. However, the
egg suspensions of both species were sufficiently dilute for
sperm to be in vast excess, even at the lowest sperm con-
For the experiments, sperm at selected concentrations were
added to the eggs in the microtiter wells, and the eggs and
sperm were mixed by stirring. In different experiments, final
concentrations of sperm in the egg suspensions ranged from
approximately 5 ? 102to 5 ? 107sperm/ml. To determine
the percent of fertilized eggs, the zygotes were fixed by add-
ing a small amount of paraformaldehyde. Fertilization was
scored either by fixing after 12 min and counting the number
of eggs with raised fertilization envelopes, or by fixing after
2 h and counting the number of eggs that had cleaved (2-
cell or 4-cell stage). Both methods gave comparable results.
One hundred or more eggs were counted for each fertilization
Characterization of Bindin
Genomic DNA was extracted as described by Lessios et
al. (1996) from gonad tissue preserved in ethanol. Mature
bindin alleles were amplified from genomic DNA with prim-
ersHeF1 (5?-ATTAATGGTGCCCACCAACAG-3?) and
HeR1 (5?-GCGTGGCTGCTCCCTTTC-3?) based on a pre-
viously determined H. erythrogramma bindin sequence
(GenBank accession number AF530406; Zigler and Lessios
2003b) and cloned as described previously (Zigler and Les-
sios 2003a). One to four clones were sequenced per individ-
ual. Heliocidaris erythrogramma alleles were sequenced us-
ing primers HeF1, HeR1, Heout5 (5?-GGCGTAGTAAGA-
GTCCAAGATCG-3?), Heout51 (5?-CAAACCTGATGGG-
TTGCGTA-3?), HeF4 (5?-TTGACCATTAGGTCTAAGC-
CG-3?), and MB1130? (5?-TGCTSGGTGCSACSAAGATT-
ADAPTIVE EVOLUTION OF BINDIN IN HELIOCIDARIS
GA-3?). Heliocidaris tuberculata alleles were sequenced us-
ing primers HeR1, HeF1, Heout5, Htout52 (5?-CAAGGGGT-
ATCACTTG-3?), and MB1130?. This combination of prim-
ers sequenced both strands of the mature bindin and its intron.
Sequencing was performed on an ABI 377 automated se-
quencer (Applied Biosystems, Inc., Foster City, CA) and ed-
ited using Sequencher 3.1 (Gene Codes Corp., Ann Arbor,
MI). Sequences have been deposited in GenBank (accession
numbers AF530401–AF530443). We sequenced 13 mature
bindin alleles from H. erythrogramma and 15 from H. tub-
A total of 20 mutations unique to a single allele(singletons)
were observed among 28 mature bindin and intron sequences
with a combined length of 40,500 bp. Singleton mutations
may represent true differences, or they may arise from poly-
merase error during amplification or cloning. Thus, the con-
servative upper limit of sequencing error in this study is
Mature bindin sequences were aligned by eye in Se-Al (ver.
1.0, written by A. Rambaut, Dept. of Zoology, University of
Oxford, Oxford, U.K.). In the glycine-rich region 3? of the
core, 24 codons could not be unambiguously aligned between
the two Heliocidaris species and were excluded from further
analysis. Ignoring gaps introduced for alignment with the
outgroup, there are 205 alignable amino acids in the mature
bindins of the two Heliocidaris species.
Bindin Gene Genealogy
Two sequences from each of the three species of Echi-
nometra for which mature bindin sequences are available
(Metz and Palumbi 1996) were used to root the Heliocidaris
bindin gene genealogy. Some amino acids were unalignable
between Echinometra and Heliocidaris: 38 amino acids 5? of
the core and 33 amino acids 3? of the core were excluded
from further analysis for this reason. Modeltest identified the
model of Tamura and Nei (1993) with a gamma correction
as the model with the highest likelihood. We fixed the pa-
rameters to their ML estimates (transversions ? 1, A-G ?
1.8616, C-T ? 3.8848; ? ? 0.9223) and used PAUP* to
reconstruct the bindin gene genealogy and to determine boot-
strap support (1000 iterations) by the neighbor-joining meth-
Tests for Selection and Episodic Evolution
For analysis, the bindin molecule was divided into three
regions: 33 amino acids of the hotspot region of rapid evo-
lution identified in Echinometra by Metz and Palumbi (1996)
and in Strongylocentrotus by Biermann (1998); 55 amino ac-
ids of the core (i.e., the conserved area in all previously
studied bindins; Vacquier et al. 1995; Zigler and Lessios
2003b); and 117 amino acids from the rest of the molecule.
For each of these three regions, we used MEGA version 2.1
(Kumar et al. 2001) to calculate the proportion of differences
per synonymous (dS) and per nonsynonymous site (dN) be-
tween the two species of Heliocidaris by the Pamilo and
Bianchi (1993) and Li (1993) method. We tested for evidence
of positive selection in each of the three regions of the mol-
ecule, using Fisher’s exact tests on all pairwise comparisons
between sequences (Zhang et al. 1997), under Nei and Go-
jobori’s (1986) model of evolution. Additionally, we tested
for selection on the Heliocidaris mature bindin sequences
using the McDonald and Kreitman (1991) test, comparing
the ratio of amino acid replacement to silent substitutions
that are polymorphic within species to the same ratio of sub-
stitutions that are fixed between species.
To test for episodic adaptive evolution, we used a series
of ML models implemented in the program PAML (ver. 3.0,
Yang 2000). To simplify the analysis, we used only two
sequences from each species of Heliocidaris and Echinometra
for a total of 10 sequences. Within each Heliocidaris species,
two divergent alleles were chosen to cover the small range
of intraspecific variability present. We used the same align-
ment as for the reconstruction of the bindin gene genealogy.
In addition to the regions already excluded as unalignable,
we also excluded the core because of its slow rate of evolution
(Vacquier et al. 1995; Zigler and Lessios 2003b). We also
excluded all sites that had an indel in one or more sequences
because of difficulties in reconstructing ancestral states at
such sites (Yang 2000). The final dataset for this analysis
included 116 amino acids.
To determine the single best tree for the 10 sequences
included in the analysis, we reconstructed the phylogeny by
ML in PAUP* using the best model of DNA evolution iden-
tified by Modeltest. For this smaller set of data, the best model
was Kimura (1980) two-parameter distance with a gamma
correction (? ? 1.9044, ? ? 0.9074). Using this model, a
ML branch-and-bound search identified the tree with the
highest likelihood. On this tree, we compared the ratio of
nonsynonymous to synonymous substitution rates (? ? dN/
dS) on the branch leading to H. erythrogramma (?E) to the
ratio on the branch leading to H. tuberculata (?T). Using
likelihood ratio tests in PAML, we tested whether ?Eor ?T
was equal to the background rate of all other branches in the
bindin gene genealogy (?B); whether ?Eequaled ?T; and
whether ?Ewas significantly larger than one, a highly con-
servative indication of positive selection (Yang 1998; Yang
and Bielawski 2000).
Cytochrome Oxidase I Phylogeny and Timing of Divergence
between the Species of Heliocidaris
The COI phylogenetic reconstruction clearly put mito-
chondrial DNA sequences of the two Heliocidaris species in
separate clades (Fig. 1). Comparison of a tree constrained by
the molecular clock and one in which rates were free to vary
indicated that there was no significant variation in substitu-
tion rates in the COI phylogeny (2?ln L ? 46.4, df ? 37, P
? 0.10). We used a general echinoid and a specific echi-
nometrid calibration of COI divergence to date the split of
the two species of Heliocidaris. COI Kimura two-parameter
distances between congeners on opposite sides of the Isthmus
of Panama have been determined in eight genera of sea ur-
chins, including Echinometra (Lessios et al. 2001). Six of
these are in the range of 9.0–13.0%. (the other two values
are smaller and assumed to be due to more recent separation).
Assuming these six pairs of species were split by the com-
pletion of the Isthmus of Panama 3.1 million years ago
KIRK S. ZIGLER ET AL.
model with rates specific to codon position. The tree is rooted with four sequences each fromEchinometra mathaei, E. oblonga, Echinometra
type C and type A (Palumbi et al. 1997) and E. vanbrunti, E. lucunter and E. viridis (McCartney et al. 2000). Bootstrap values on branches
supported by more than 70% from a neighbor-joining bootstrap (1000 replicates) are indicated.
Maximum-likelihood tree of cytochrome oxidase I of Heliocidaris, from a heuristic search based on the Tamura and Nei (1993)
crosses between Heliocidaris erythrogramma (H. e.) and H. tuber-
culata (H. t.) gametes at different sperm concentrations. Results
from representative experiments are shown, illustrating the range
in sperm concentrations required to give effective fertilization rates
in different homologous and heterologous crosses.
Fertilization efficiency of homologous and heterologous
Egg species ? sperm species % fertilization
H. e. ? H. e.
H. e. ? H. t.
Dejellied H. e. ? H. e.
3 ? 105
3 ? 103
5 ? 107
5 ? 106
5 ? 106
5 ? 104
5 ? 107
5 ? 104
4 ? 105
4 ? 102
4 ? 106
4 ? 104
Dejellied H. e. ? H. t.
H. t. ? H. t.
H. t. ? H. e.
(Coates and Obando 1996) produces an estimate of approx-
imately 3.5% divergence per million years for this region of
COI. Dividing the mean COI Kimura two-parameter diver-
gence between the two species of Heliocidaris (14.7%) by
this estimated rate suggests that H. erythrogramma and H.
tuberculata were separated approximately 4.2 million years
ago. Mean COI divergence between neotropical species of
Echinometra also produces an estimated rate of 3.5% per
million years, with a potential range (based on branch length
variation in the genus) of 24% of the mean (McCartney et
al. 2000). Thus, the range of estimates of time of splitting
between the species of Heliocidaris would be 3.7–4.7 million
years ago. McMillan et al. (1992) determined mitochondrial
DNA divergence between H. erythrogramma and H. tuber-
culata through restriction fragment length polymorphisms
(RFLP) to be 7.9%. Calibrated with Bermingham and Les-
sios’s (1993) estimated mitochondrial DNA RFLP divergence
of 1.6–2.1% per million years for sea urchins, this also pro-
duces an estimated time of 3.8–4.9 million years ago for
Heliocidaris Gametic Compatibility
Eggs of each species of Heliocidaris could be fertilized by
sperm of the other, but neither of these crosses was efficient.
Obtaining equivalent fertilization percentages of H. tuber-
culata eggs required more than an order of magnitude more
H. erythrogramma sperm than H. tuberculata sperm (Table
1). Inefficiency in the reciprocal cross was much more severe;
a H. erythrogramma sperm concentration of 3 ? 105per
milliliter fertilized 98% of H. erythrogramma eggs, whereas
150 times more H. tuberculata sperm (5 ? 107per milliliter)
only fertilized 2% of H. erythrogramma eggs.
A substantial extent of fertilization of H. erythrogramma
eggs by H. tuberculata sperm could be achieved only after
removal of the jelly coat from the eggs (Table 1). At the
same sperm concentration, H. tuberculata sperm fertilized2%
of jelly-coat-intact H. erythrogramma eggs and 87% of de-
jellied H. erythrogramma eggs. We suspect that the small
percent of fertilization of jelly-coat-intact H. erythrogramma
eggs by H. tuberculata sperm reflects fertilization of eggs
with compromised jelly coats, as we observed that in highly
concentrated sperm suspensions the egg jelly coats became
weakened, perhaps reflecting normal digestive processes of
sperm. Whereas dejellying greatly increased the ability of H.
tuberculata sperm to fertilize H. erythrogramma eggs, dejel-
lying resulted in a 10-fold decrease in the normal fertilization
efficiency of H. erythrogramma eggs by homospecific sperm.
Structure of the Heliocidaris Bindin Molecule
Heliocidaris erythrogramma bindin was described by Zig-
ler and Lessios (2003b), and H. tuberculata bindin is similar
(Fig. 2). Both contain the bindin core, the characteristic bin-
din intron, and have glycine-rich regions both 5? and 3? of
the core. Heliocidaris erythrogramma mature bindins are ei-
ther 206 or 210 amino acids long, whereas H. tuberculata
bindins range from 223 to 226 amino acids in length. The
H. erythrogramma intron is 715 nucleotides (nt) long. We
amplified mature bindin and its intron in H. tuberculata but
were unable to sequence through the middle of the intron.
Partial intron sequences from H. tuberculata are 600 nt in
Mode of Bindin Evolution in Heliocidaris
Little intraspecific amino acid variation is observed among
the Heliocidaris mature bindin alleles (Table 2). Seven of 13
H. erythrogramma alleles have identical amino acid sequenc-
es, and only six of 210 amino acid sites are polymorphic.
ADAPTIVE EVOLUTION OF BINDIN IN HELIOCIDARIS
Amino acid alignment of two mature bindin alleles per species in Heliocidaris and Indo-West Pacific Echinometra. Echinometra sequences are from Metz and Palumbi
(1996). Dots indicate identity to the first sequence, gaps are indicated by dashes. The conserved core region is shaded. Unalignable regions excluded from analysis are boxed.
KIRK S. ZIGLER ET AL.
(H. t.) mature bindin and their ratio (? ? dN/dS). We show means of pairwise comparisons calculated by the Pamilo and Bianchi (1993)
and Li (1993) methods. The hotspot and core are composed of amino acids 95–130 and 131–185 in Figure 2, respectively. The ?-value
for the hotspot is greater than the neutral expectation for all pairwise comparisons by Fisher’s exact test using the Nei and Gojobori
(1986) model of evolution (0.015 ? P ? 0.038).
Replacement (dN) and silent (dS) substitutions per site in three regions of Heliocidaris erythrogramma (H. e.) and H. tuberculata
Hotspot (33 codons)
Core (55 codons)
between H. e. and H. t.
within H. e. and H. t.
between H. e. and H. t.
within H. e. and H. t.
between H. e. and H. t.
within H. e. and H. t.
between H. e. and H. t.
within H. e. and H. t.
Rest of molecule (117 codons)
Total (205 codons)
from a heuristic search based on the Tamura and Nei (1993) model
with a gamma correction. The tree is rooted with the six Echino-
metra sequences shown in Figure 2. Bootstrap values on branches
supported by more than 70% from a neighbor-joining bootstrap
(1000 replicates) are indicated. We indicate when both alleles were
sequenced from a single individual (e.g., Sydney 15-1 and Sydney
Maximum-likelihood tree of Heliocidaris mature bindin,
One indel of four residues is shared by three alleles of H.
erythrogramma. Among 15 H. tuberculata alleles there are
five singleton amino acid changes and four small (one or two
residue) indels. All the observed indels are located in the
The bindin gene genealogy separates the H. erythrogramma
and H. tuberculata alleles (Fig. 3). As in bindins of other sea
urchin genera (Palumbi and Metz 1996; Metz et al. 1998;
Biermann 1998; Zigler and Lessios 2003a), nonsynonymous
changes are accumulating slowly in the core of Heliocidaris
bindin and more rapidly in the hotspot just 5? of the core
(Table 2). In Heliocidaris, there is greater than a 10-fold
difference in the rate of nonsynonymous change between
these two adjacent regions of the molecule. Rates of syn-
onymous change follow the opposite pattern: there are no
synonymous changes observed in the hotspot, whereas nearly
20% of the synonymous sites are substituted in the core re-
gion between the two species. In the hotspot, there is a sig-
nificant excess of nonsynonymous to synonymous changes
(12–14 to 0, depending on the specific pairwise comparison).
McDonald-Kreitman (1991) tests for the entire mature bin-
din did not reveal any evidence of selection in Heliocidaris.
Twenty-six replacement and 17 silent differences were fixed
between the species, and nine replacement and seven silent
differences were polymorphic within the species (Fisher’s
exact test, P ? 0.50). When the core region (amino acids
131–185 in Fig. 2) was excluded, there were 24 replacement
and eight silent fixed differences, as compared to seven re-
placement and five silent polymorphic differences (Fisher’s
exact test, P ? 0.24). The small number of intraspecific dif-
ferences within each species (Table 2) makes it difficult to
identify selection using this statistical test.
Amino Acid Replacements along the Lineages of
Heliocidaris erythrogramma and H. tuberculata
We tested for shifts in the ratio of amino acid replacement
to silent substitutions (?) on the branches leading to H. er-
ythrogramma and H. tuberculata compared to the background
rate (?B) and to each other. We estimated the log likelihood
of six models on the 10-taxon tree composed of two se-
quences from each of the species of Heliocidaris and from
each of the species of Echinometra sequenced by Metz and
Palumbi (1996; see Fig. 4). The number of estimated param-
eters in each model ranged from 18 to 20, including 16 branch
lengths, one transition/transversion ratio, and one to three ?-
values (Table 3). We then compared various nested models
to test four hypotheses (Table 4). The ratio of nonsynony-
mous to synonymous substitution on the branch leading to
H. erythrogramma (?E) is significantly greater than the back-
ground ratio (?B), whereas the ratio on the branch leading
to H. tuberculata (?T) is not. Additionally, ?Eis significantly
greater than ?T. Finally, ?Eis significantly greater than one.
When ?E, ?T, and ?Bare all allowed to vary (model C in
Table 3) the estimated ?-values are ?, 0.32, and 0.30, re-
spectively. Under this model, 16.6 nonsynonymous changes
ADAPTIVE EVOLUTION OF BINDIN IN HELIOCIDARIS
as an outgroup. T marks the branch leading to H. tuberculata, E marks the branch leading to H. erythrogramma. Reconstructed nonsy-
nonymous changes on branches E and T under model C (see Table 3) are indicated by filled boxes, synonymous changes by unfilled
Maximum-likelihood assignment of nucleotide changes along the branches of the Heliocidaris bindin tree, using Echinometra
synonymous to synonymous substitution (?) of bindin on the lin-
eages leading to Heliocidaris erythrogramma (E) and H. tuberculata
(T). Branches E and T are indicated in Figure 4; B is the background
rate for all other branches.
Maximum-likelihood models estimating the ratio of non-
Model Free ?
Estimates of ?
mous to synonymous substitution (?) on branches E and T by like-
lihood-ratio test. Branches E and T are indicated in Figure 4, B is
the background rate for all other branches. Null (H0) and alternative
(H1) model details and log likelihoods are in Table 3.
Testing hypotheses regarding the ratio of nonsynony-
2(ln L1? ln L0)
occurred on the branch of H. erythrogramma (and 0.1 syn-
onymous change), whereas on the branch of H. tuberculata
6.0 nonsynonymous changes (and 6.4 synonymous changes)
have occurred (Fig. 4). Two lines of evidence suggest these
results are not influenced by the choice of Echinometra as
the outgroup for this analysis. First, the accuracy of the in-
ferred ancestral amino acid sequences is calculated by PAML
to be 99.4% for all the ancestral nodes and 98.6% for the
common ancestor of the two species of Heliocidaris. Second,
similar patterns of statistical significance were obtained when
the long branch connecting the Heliocidaris and Echinometra
sequences was allowed to have its own ? value in each of
the models in Table 3.
There is also evidence for a concentration of change in
bindin on the branch leading to H. erythrogramma in the core
and the indels that were not included in the previous analysis.
In the core there are two amino acid differences between H.
erythrogramma and H. tuberculata (at residues 172 and 179
in Fig. 2). In both cases H. tuberculata has the same residue
as all of the outgroup sequences, which suggests that both
changes occurred on the branch leading to H. erythrogramma.
This is the first case of amino acid variation between con-
geners. In the four genera of echinoids for which bindin se-
quences have been published (Metz and Palumbi 1996; Bier-
mann 1998; Metz et al. 1998; Zigler and Lessios 2003a) core
amino acids remain the same within a genus. Additionally,
several small indels have arisen on the branch leading to H.
erythrogramma (at residues 30–33, 206–209, and 264–265 in
We found that the two species of Heliocidaris diverged
less than five million years ago. Our estimated time of di-
vergence is considerably shorter than the 10–13 million years
ago estimated by DNA-DNA hybridization methods (Smith
et al. 1990) but more in line with the five to eight million
years ago derived from mitochondrial RFLP divergence by
McMillan et al. (1992). These results suggest that the evo-
lution of nonfeeding development and the differences in ami-
no acid sequence of bindin in Heliocidaris have occurred
within the past 5 million years. There are only two species
of Heliocidaris, but a subspecies, H. erythrogramma armi-
gera, is found on the south and west coast of Australia (Mor-
tensen 1943). McMillan et al. (1992) estimate that, based on
RFLP divergence, H. erythrogramma armigera diverged from
H. erythrogramma erythrogramma 1 million years ago. Be-
cause H. erythrogramma armigera individuals also develop
directly from large eggs (R. A. Raff, pers. obs.) , this suggests
that the transition from indirect to direct development in H.
erythrogramma occurred between five million years ago and
one million years ago.
Cross-fertilization experiments revealed that there is a
small likelihood of hybrid production between the two He-
liocidaris species under normal environmental conditions,
even though there is a part of the year during which both
species produce gametes (Laegdsgaard et al. 1991). The ma-
jor barrier to cross-fertilization between the two species is
gametic incompatibility, with especially strong incompati-
bility between H. erythrogramma eggs and H. tuberculata
sperm. Another factor that may contribute to isolation be-
tween the species is the different buoyancy of the eggs. Eggs
of H. tuberculata females, like those of other indirect-de-
KIRK S. ZIGLER ET AL.
veloping sea urchin species, sink as they are extruded. In still
water, H. tuberculata eggs remain layered over the female or
sink to the substrate. In contrast, eggs of H. erythrogramma
females immediately spiral upward toward the water surface
as they are shed. The buoyancy and behavior of Heliocidaris
sperm have not been characterized, but the difference in egg
buoyancy implies that H. erythrogramma sperm in the wild
might have to float or swim upward to achieve most efficient
fertilization, whereas upward movement by H. tuberculata
sperm might decrease fertilization efficiency.
Postzygotic isolation further reduces the possibility of gene
flow between the two species. Embryos derived from the
cross between H. tuberculata eggs and H. erythrogramma
sperm arrest at gastrulation (Raff et al. 1999). Successful
fertilization in the reciprocal cross would be most likely for
H. erythrogramma eggs with damaged or missing jelly coats.
Embryos from dejellied H. erythrogramma eggs are more
fragile than normal embryos. Of those that survive past meta-
morphosis, no hybrids lived past the juvenile adult stage,
even though we can routinely raise nonhybrid individuals
past this stage.
The evolution of bindin in Heliocidaris is characterized by
positive selection in the hotspot region, low intraspecific var-
iability, and a burst of selection on the branch leading to H.
erythrogramma. The evidence for positive selection in the
hotspot region is due to the combination of an increased rate
of nonsynonymous substitution and a decreased rate of syn-
onymous substitution relative to the rest of the molecule. A
similar pattern is seen in the bindins of Echinometra and
Strongylocentrotus, where there is evidence of positive se-
lection, and in Tripneustes, where there is not (Metz and
Palumbi 1996; Biermann 1998; Zigler and Lessios 2003a).
The cause of this pattern is unclear. There is no evidence of
codon usage bias in the bindins of any of these genera, so it
is unlikely that indirect selection on synonymous substitu-
tions is contributing to this effect (Zigler and Lessios
Three hypotheses could explain the observation of adaptive
evolution of bindin along the branch leading to H. erythro-
gramma: (1) a correlated response to the evolution of direct
development in H. erythrogramma; (2) an intraspecific pro-
cess acting within H. erythrogramma but not H. tuberculata;
and (3) selection against hybridization (reinforcement).
That adaptive evolution of bindin has occurred along the
H. erythrogramma lineage suggests that it may be correlated
with the evolution of direct development. The associated
changes in gametogenesis leading to large egg and large
sperm may have required correlated changes in H. erythro-
gramma bindin. Although bindin is the first molecule found
to be under episodic selection along a lineage where direct
development has evolved, it is undoubtedly just one of many
proteins that have changed rapidly in concert with this de-
velopmental transition. There is, however, no obvious con-
nection between the evolution of a gamete recognition protein
and the mode of development. Determining whether the se-
lection on bindin is a result of the shift to direct development
will require examining how bindin has evolved along other
lineages where direct development has arisen.
A second hypothesis is that some intraspecific process that
promotes evolution under positive selection is occurring in
H. erythrogramma but not H. tuberculata. Palumbi (1999)
suggested that an intraspecific process such as male hetero-
zygote advantage, nontransitive female preferences, or in-
terlocus conflict evolution (Rice and Holland 1997) may ex-
plain the high intraspecific diversity observed in Echinome-
tra. Given the low intraspecific diversity in Heliocidaris, male
heterozygote advantage or nontransitive female preferences
are unlikely to apply, but interlocus conflict evolution could
still account for the selection on H. erythrogramma bindin.
It is not clear, however, why this sort of selection should be
present in one, but not the other, species.
A third potential explanation for the selection on Helio-
cidaris bindin is the avoidance of maladaptive hybridization
(Dobzhansky 1940). The evolution of bindin has been studied
in five genera of sea urchins. Selection on bindin has been
reported in Echinometra (Metz and Palumbi 1996) and Stron-
gylocentrotus (Biermann 1998), which have multiple sym-
patric species, and now Heliocidaris, whose two species have
partially overlapping ranges. No positive selection has been
observed in the bindins of Arbacia (Metz et al. 1998) or
Tripneustes (Zigler and Lessios 2003a), in which extant spe-
cies are all allopatric. The observed positive selection on
bindin of H. erythrogramma may track changes in the bindin
receptors of the egg necessary to prevent fertilization by
sperm of H. tuberculata. Consistent with this idea, we ob-
served stronger gametic incompatibility between H. erythro-
gramma eggs and H. tuberculata sperm than in the reciprocal
cross. H. erythrogramma females produce far fewerlargeeggs
than H. tuberculata, so each egg lost to hybridization is more
costly to a female of H. erythrogramma than to a female of
H. tuberculata. The low intraspecific diversity of bindin in
Heliocidaris is also consistent with directional selection
caused by reinforcement. This is a pattern different than that
of Echinometra or Strongylocentrotus, in which intraspecific
variability in bindin is high and in which there are high dN/
dSratios between alleles of the same species, which is not
consistent with expectations of reinforcement (Zigler and
The partially overlapping geographic distributions of the
Heliocidaris species permit a test of the importance of re-
inforcement in the evolution of bindin (Noor 1997). Their
ranges overlap on the southeast Australian coast, but H. er-
ythrogramma is spread along the south coast to southwest
Australia, whereas H. tuberculata is distributed up the east
coast of Australia and off northern New Zealand (Mortensen
1943). There is extensive differentiation between local pop-
ulations of H. erythrogramma, presumably due to the limited
dispersal abilities of their direct-developing larvae(McMillan
et al. 1992). If the selection on H. erythrogramma bindin
arises from the need to avoid cross-fertilization, and if gene
flow is sufficiently restricted to keep conspecific populations
from becoming homogeneous, there should be fewer changes
in H. erythrogramma bindins in west and south Australia,
where they do not have to contend with H. tuberculata. Fer-
tilization assays should also show gametes of H. erythro-
gramma from south and west Australia to be more compatible
with gametes from H. tuberculata than those of H. erythro-
gramma from eastern Australia. The two alternative hypoth-
eses (shift in developmental mode and interlocus conflict
evolution) should operate regardless of patterns of allopatry
ADAPTIVE EVOLUTION OF BINDIN IN HELIOCIDARIS
or sympatry. Thus, there is a need to study bindin in H.
erythrogramma populations from south and west Australia,
particularly populations of the subspecies H. erythrogramma
armigera, which has presumably remained free of gene flow
from the area of overlap for one million years (McMillan et
We thank H. Pedersen and K. Wilson for collecting tissue
samples of Heliocidaris and A. Caldero ´n, L. Caldero ´n, and
E. Archie for assistance in the laboratory. The manuscript
was improved by comments from R. Collin, C. Cunningham,
D. McClay, G. Wray, an anonymous reviewer, and the as-
sociate editor. This work was supported by National Science
Foundation and Smithsonian predoctoral fellowships to KSZ
and the Duke University Department of Zoology, the Smith-
sonian Molecular Evolution Program, and National Science
Foundation and National Institutes of Health grants to RAR.
Bermingham, E., and H. A. Lessios. 1993. Rate variation of protein
and mitochondrial DNA evolution as revealed by sea urchins
separated by the Isthmus of Panama. Proc. Nat. Acad. Sci. USA
Biermann, C. H. 1998. The molecular evolution of sperm bindin in
six species of sea urchins (Echinoida: Strongylocentrotidae).
Mol. Biol. Evol. 15(12):1761–1771.
Byrne, M., J. T. Villinski, P. Cisternas, E. Popodi, and R. A. Raff.
1999. Maternal factors and the evolution of developmental
mode: evolution of oogenesis in Heliocidaris erythrogramma.
Dev. Genes Evol. 209:275–283.
Coates, A. G., and J. A. Obando. 1996. The geologic evolution of
the Central American Isthmus. Pp. 21–56 in J. B. C. Jackson,
A. G. Coates, and A. Budd, eds. Evolution and environment in
tropical America. Univ. of Chicago Press, Chicago.
Dobzhansky, T. 1940. Speciation as a stage of evolutionary diver-
gence. Am. Nat. 74:312–321.
Eckelbarger, K. J., C. M. Young, and J. L. Cameron. 1989. Modified
sperm in echinoderms from the bathyal and abyssal zones in the
deep sea. Pp. 67–74 in J. S. Ryland and P. A. Tyler, eds. Re-
production, genetics and distributions of marine organisms. Ol-
sen and Olsen, Fredensborg, Denmark.
Emlet, R. B. 1990. World patterns of developmental mode in echi-
noid echinoderms. Pp. 329–335 in M. Hoshi and O. Yamashita,
eds. Advances in invertebrate reproduction. Vol. 5, Elsevier,
———. 1995. Larval spicules, cilia, and symmetry as remnants of
indirect development in the direct developing sea urchin Helio-
cidaris erythrogramma. Dev. Biol. 167:405–415.
Felsenstein, J. 1981. Evolutionary trees and DNA sequences: a max-
imum likelihood approach. J. Mol. Evol. 17:368–376.
Haag, E. H., B. J. Sly, and R. A. Raff. 1999. Apextrin, a novel
extracellular protein involved in adaptive evolution of larval
ectoderm in the direct-developing sea urchin Heliocidaris ery-
throgramma. Dev. Biol. 211:77–87.
Henry, J. J., and R. A. Raff. 1990. Evolutionary change in the
process of dorsoventral axis determination in the direct devel-
oping sea urchin, Heliocidaris erythrogramma. Dev. Biol. 141:
Henry, J. J., G. A. Wray, and R. A. Raff. 1991. Mechanism of an
alternate type of echinoderm blastula formation: the wrinkled
blastula of the sea urchin Heliocidaris erythrogramma. Dev.
Growth Differ. 33:317–328.
Hoegh-Guldberg, O., and R. B. Emlet. 1997. Energy use during the
development of a lecithotrophic and a planktonic echinoid. Biol.
Keesing, J. K. 2001. The ecology of Heliocidaris erythrogramma.
Pp. 261–270 in J. M. Lawrence, ed. Edible sea urchins: biology
and ecology. Elsevier, Amsterdam.
Kimura, M. 1980. A simple method for estimating evolutionary
rates of base substitution through comparative studies of nucle-
otide sequences. J. Mol. Evol. 16:111–120.
Kumar, S., K. Tamura, I. B. Jakobsen, and M. Nei. 2001. MEGA
2: molecular evolutionary genetics analysis software. Bioinfor-
Laegdsgaard, P., M. Byrne, and D. T. Anderson. 1991. Reproduction
of sympatric populations of Heliocidaris erythrogramma and H.
tuberculata (Echinoidea) in New South Wales. Mar. Biol. 110:
Lessios, H. A., B. D. Kessing, G. M. Wellington, and A. Graybeal.
1996. Indo-Pacific echinoids in the tropical east Pacific. Coral
Lessios, H. A., B. D. Kessing, and J. S. Pearse. 2001. Population
structure and speciation in tropical seas: phylogeography of the
sea urchin Diadema. Evolution 55(5):955–975.
Li, W.-H. 1993. Unbiased estimation of the rates of synonymous
and nonsynonymous substitution. J. Mol. Evol. 36:96–99.
McCartney, M. A., G. Keller, and H. A. Lessios. 2000. Dispersal
barriers in tropical oceans and speciation in Atlantic and eastern
Pacific sea urchins of the genus Echinometra. Mol. Ecol. 9:
McDonald, J. H., and M. Kreitman. 1991. Adaptive protein evo-
lution at the Adh locus in Drosophila. Nature 351:652–654.
McMillan, W. O., R. A. Raff, and S. R. Palumbi. 1992. Population
genetic consequences of developmental evolution in sea urchins
(genus Heliocidaris). Evolution 46(5):1299–1312.
Metz, E. C., and S. R. Palumbi. 1996. Positive selection and se-
quence rearrangements generate extensive polymorphism in the
gamete recognition protein bindin. Mol. Biol. Evol. 13(2):
Metz, E. C., G. Gomez-Gutierez, and V. D. Vacquier. 1998. Mi-
tochondrial DNA and bindin gene sequence evolution among
allopatric species of the sea urchin genus Arbacia. Mol. Biol.
Mortensen, T. 1943. A monograph of the Echinoidea v. III.3 Ca-
marodonta II. Echinidae, Strongylocentrotidae, Parasaleniidae,
Echinometridae. C. A. Reitzel, Copenhagen.
Nei, M., and T. Gojobori. 1986. Simple methods for estimating the
numbers of synonymous and nonsynonymous nucleotide sub-
stitutions. Mol. Biol. Evol. 3(5):418–426.
Noor, M. A. F. 1997. How often does sympatry affect sexual iso-
lation in Drosophila? Am. Nat. 149(6):1156–1163.
Palumbi, S. R. 1999. All males are not created equal: fertility dif-
ferences depend on gamete recognition polymorphisms in sea
urchins. Proc. Nat. Acad. Sci. USA 96(22):12632–12637.
Palumbi, S. R., G. Grabowsky, T. Duda, L. Geyer, and N. Tachino.
1997. Speciation and population genetic structure in tropical
Pacific sea urchins. Evolution 51(5):1506–1517.
Pamilo, P., and N. O. Bianchi. 1993. Evolution of the Zfx and Zfy
genes: rates and interdependence between the genes. Mol. Biol.
Posada, D., and K. A. Crandall. 1998. Modeltest: testing the model
of DNA substitution. Bioinformatics 14(9):817–818.
Raff, E. C., E. M. Popodi, B. J. Sly, F. R. Turner, J. T. Villinski,
and R. A. Raff. 1999. A novel ontogenetic pathway in hybrid
embryos between species with different modes of development.
Raff, R. A. 1987. Constraint, flexibility, and phylogenetic history
in the evolution of direct development in sea urchins. Dev. Biol.
Raff, R. A., and B. J. Sly. 2000. Modularity and dissociation in the
evolution of gene expression territories in development. Evol.
Raff, R. A., L. Herlands, V. B. Morris, and J. Healy. 1990. Evo-
lutionary modification of echinoid sperm correlates with devel-
opmental mode. Dev. Growth. Differ. 32(3):283–291.
Rice, W. R., and B. Holland. 1997. The enemies within: intergen-
omic conflict, interlocus contest evolution (ICE) and the intra-
specific red queen. Behav. Ecol. Sociobol. 41:1–10.
Scott, L. B., W. J. Lennarz, R. A. Raff, and G. A. Wray. 1990. The
2302 Download full-text
KIRK S. ZIGLER ET AL.
‘‘lecithotrophic’’ sea urchin Heliocidaris erythrogramma lacks
typical yolk platelets and yolk glycoproteins. Dev. Biol. 138:
Smith, A. B. 1988. Phylogenetic relationship, divergence times, and
rates of molecular evolution for Camarodont sea urchins. Mol.
Biol. Evol. 5(4):345–365.
Smith, M. J., J. D. G. Boom, and R. A. Raff. 1990. Single copy
DNA distance between two congeneric sea urchin species ex-
hibiting radically different modes of development. Mol. Biol.
Strathmann, R. R. 1978. The evolution and loss of feeding larval
stages of marine invertebrates. Evolution 32:894–906.
Swofford, D. L. 2001. PAUP*: phylogenetic analysis using parsi-
mony (* and other methods). Ver. 4. Sinauer Associates, Sun-
Tamura, K., and M. Nei. 1993. Estimating the number of nucleotide
substitutions in the control region of mitochondrial DNA in hu-
mans and chimpanzees. Mol. Biol. Evol. 10:512–526.
Ulrich, A. S., M. Otter, C. G. Glabe, and D. Hoekstra. 1998. Mem-
brane fusion is induced by a distinct peptide sequence of the sea
urchin fertilization protein bindin. J. Biol. Chem. 273(27):
Vacquier, V. D. 1998. Evolution of gamete recognition proteins.
Vacquier, V. D., and G. W. Moy. 1977. Isolation of bindin: the
protein responsible for adhesion of sperm to sea urchin eggs.
Proc. Nat. Acad. Sci. USA 74(6):2456–2460.
Vacquier, V. D., W. J. Swanson, and M. E. Hellberg. 1995. What
have we learned about sea urchin sperm bindin? Develop.
Growth Differ. 37:1–10.
Villinski, J. T., J. C. Villinski, M. Byrne, and R. A. Raff. 2002.
Convergent maternal provisioning and life-history evolution in
echinoderms. Evolution 56(9):1764–1775.
Wray, G. A. 1996. Parallel evolution of nonfeeding larvae in echi-
noids. Syst. Biol. 45(3):308–322.
Wray, G. A., and R. A. Raff. 1989. Evolutionary modification of
cell lineage in the direct-developing sea urchin Heliocidaris er-
ythrogramma. Dev. Biol. 132:458–470.
Yang, Z. 1998. Likelihood ratio tests for detecting positive selection
and application to primate lysozyme evolution. Mol. Biol. Evol.
———. 2000. Phylogenetic analysis by maximum likelihood
(PAML). Ver. 3.0. Univ. College London, London.
Yang, Z., and J. P. Bielawski. 2000. Statistical methods for detecting
molecular adaptation. Trends in Ecol. Evol. 15(12):496–503.
Zhang, J., S. Kumar, and M. Nei. 1997. Small-sample tests of ep-
isodic evolution: a case study of primate lysozymes. Mol. Biol.
Zigler, K. S., and H. A. Lessios. 2003a. Evolution of bindin in the
pantropical sea urchin Tripneustes: comparisons to bindin of
other genera. Mol. Biol. Evol. 20(2):220–231.
———. 2003b. 250 million years of bindin evolution. Biol. Bull.
Corresponding Editor: G. Wallis