Pre-zygotic isolation in the macroalgal genus Fucus from four contact zones spanning 100–10 000 years: a tale of reinforcement?

Hybrid zones provide an ideal natural experiment to study the selective forces driving evolution of reproductive barriers and speciation. If hybrid offspring are less fit than the parental species, pre-zygotic isolating barriers can evolve and strengthen in response to selection against the hybrids (reinforcement). Four contact zones between the intertidal macroalgae Fucus serratus (Fs) and Fucus distichus (Fd), characterized by varying times of sympatry and order of species introduction provide an opportunity to investigate reinforcement. We examined patterns of hybridization and reproductive isolation between Fs and Fd in: (i) northern Norway (consisting of two natural sites, 10 000 years old), (ii) the Kattegat near Denmark (Fd introduced, nineteenth century) and (iii) Iceland (Fs introduced, nineteenth century). Using 10 microsatellites and chloroplast DNA, we showed that hybridization and introgression decreased with increasing duration of sympatry. The two younger contact zones revealed 13 and 24% hybrids and several F1 individuals, in contrast to the older contact zone with 2–3% hybrids and an absence of F1s. Cross-fertilization experiments revealed that the reduction in hybridization in the oldest zone is consistent with increased gametic incompatibility.


Summary
Hybrid zones provide an ideal natural experiment to study the selective forces driving evolution of reproductive barriers and speciation. If hybrid offspring are less fit than the parental species, pre-zygotic isolating barriers can evolve and strengthen in response to selection against the hybrids (reinforcement). Four contact zones between the intertidal macroalgae Fucus serratus (Fs) and Fucus distichus (Fd), characterized by varying times of sympatry and order of species introduction provide an opportunity to investigate reinforcement. We examined patterns of hybridization and reproductive isolation between Fs and Fd in: (i) northern Norway (consisting of two natural sites, 10 000 years old), (ii) the Kattegat near Denmark (Fd introduced, nineteenth century) and (iii) Iceland (Fs introduced, nineteenth century). Using 10 microsatellites and chloroplast DNA, we showed that hybridization and introgression decreased with increasing duration of sympatry. The two younger contact zones revealed 13 and 24% hybrids and several F 1 individuals, in contrast to the older contact zone with 2-3% hybrids and an absence of F 1 s. Cross-fertilization experiments revealed that the reduction in hybridization in the oldest zone is consistent with increased gametic incompatibility.

Introduction
The reinforcement. Reinforcement is most likely to occur in areas where there is strong selection against hybrids and restricted gene flow [19], both of which are characteristic of the Fs/Fd hybrids zones [46,47]. Furthermore, as the evolution of pre-zygotic isolating barriers is expected to result in a reduction of interspecific mating through time [35], reinforcement will be stronger in older zones of sympatry. Specifically, populations in northern Norway (10 000 years of contact) should show a lower degree of hybridization and interspecific fertilization success relative to the younger hybrid zones (100 years) in the Kattegat and in Iceland.
In this study, we examined patterns of hybridization and reproductive isolation between Fs and Fd in two young and two old contact zones. Specifically, we: (i) determined the degree of hybridization and introgression using blind sampling with subsequent identification of parents and hybrids using both nuclear and chloroplast markers, and (ii) compared the degree of interspecific fertilization success within and across contact zones.

Sampling
Hybridization among Fs and Fd was examined at all contact zones: (i) Kirkenes (n = 47) and Tromsø (northern Norway) (n = 192), where both species are native and have been in sympatry for ca 10 000 years; (ii) Heimaey (Iceland) (n = 96), where F. serratus was introduced ca 100 years ago; and (iii) Blushøj (Kattegat, Denmark) (n = 286), where Fd was introduced ca 120 years ago. At both Tromsø and Blushøj, two 2 × 12 m plots were established to investigate micro-scale differences in hybridization, with the long axis perpendicular to the shoreline (beginning at the shallowest appearance of Fd and finishing at the deepest occurrence of F. serratus). Owing to physical constraints at Kirkenes and Heimaey, only one plot was established. Within each plot, a small tissue sample was excised in situ from all post-recruit (more than 10 cm) individuals without a priori species identification. Tissues were stored in silica gel crystals prior to genetic analysis. All data from Blushøj are from Coyer et al. [46].

DNA extraction and microsatellite analysis
DNA was extracted from ca 5 mg of silica-dried tissue as described in Hoarau et al. [ [46]. PCR reaction mixtures and conditions are described in Coyer et al. [46]. All genotypes were visualized on an ABI 377 or on a 3500XL automatic sequencer (Life technologies) and analysed with GENESCAN software (Life technologies).

Chloroplast marker
We used a portion of the ribulose-1,5-bisphosphate carboxylase operon (Rubisco; chloroplast) to identify the egg-producing parent for hybrids, as the parental species differed by a 33 bp indel [63]. PCR reactions and conditions are described in Coyer et al. [63] and parental identity was detected by sizing of PCR products on an ABI 377 or on a 3500XL automatic sequencer (Life technologies).

Molecular data analysis
Genetic diversity using Nei's [64] non-biased H exp , estimators of F IS and F ST (Wright [65]) as f and θ, [66], and allele frequencies were estimated using the software GENETIX 4.02 [67]. Multilocus heterozygosity (MLH) was defined as the number of heterozygous loci per individual and ranged from 0 to 10.
Admixture of microsatellite genotypes was analysed with STRUCTURE [68], which uses a Bayesian approach to identify K user-defined clusters of individuals that are genetically homogeneous. Sampled individuals were assigned either to clusters or jointly to two or more clusters if their genotypes indicated admixture. All analyses were replicated 10 times for each location independently (table 1) to ensure proper convergence of the Markov chain Monte Carlo (MCMC) with the parameters: ancestry model = admixture (to account for recent divergence and shared ancestral polymorphisms); frequency model = independent; burn-in = 1 000 000; MCMC length = 2 000 000 post-burn-in and K = 2 (to account for the parental species). The analysis with STRUCTURE produced for each individual, an admixture coefficient (I FS ) defined as the proportion of membership to the F. serratus cluster.
The power of admixture analyses to detect hybridization in Fucus was evaluated by simulation using HYBRIDLAB (v. 1.0) [69] as described previously [46]. Simulations were repeated 10 times for each location independently and these results were used to define the boundaries of the I FS admixture coefficient (95% confidence intervals) for: (i) pure F. serratus, (ii) pure F. distichus, and (iii) F 1 hybrids.

Fertilization success (pre-zygotic isolation)
Reproductive receptacles (consisting of 100s of conceptacles) from both species were collected within each of the contact zones (except Kirkenes). Receptacles were packed individually in aluminium foil and transported to the laboratory in a cooler box within 36-48 h. A reciprocal crossing design was used for Blushøj, Tromsø and Heimaey, using five Fd females, five Fs males, five Fs females and five Fd males, resulting in 25 Fd conspecific crosses, 25 Fs conspecific crosses and 2 × 25 interspecific crosses. Additionally, negative controls consisting of eggs only (no sperm added) were used for all females. All specimens were genotyped with 10 microsatellite loci and analysed as described above (e.g. STRUCTURE) to determine species identity.
Crosses were performed in sterile plastic culture plates containing 3 ml sterile seawater at 5 • C and each cross and control was replicated four times using 10 oogonia (80 eggs) in each replicate. Fs is a dioecious species, with receptacles from a mature individual possessing conceptacles containing either antheridia or oogonia. The sex of Fs individuals was determined microscopically. Fs receptacles were stored overnight at 4 • C. The addition of ice-cold sterilized seawater induced conceptacles to release eggs and sperm, which were collected and washed once in sterile seawater before further use.
As Fd is a hermaphroditic species with individual conceptacles within the receptacle containing both antheridia and oogonia, collecting eggs after release from conceptacles was inappropriate, because eggs from such preparations were potentially already fertilized by conspecific sperm. To prevent selffertilization during gamete extraction from Fd, eggs were collected with a pipette immediately after sectioning receptacles with a razor blade in order to disrupt the conceptacles and then sequentially washed (×3) in 10 ml sterile seawater to dilute any accompanying sperm and minimize/eliminate selffertilization. Sperm from Fd was also obtained by sectioning as previously described, but sections were immediately placed in a small mesh bag (50 µm mesh size) that allowed free passage of the small sperm (5-6 µm dia.), but retained the larger eggs (70-100 µm dia.). The bags were then placed in a culture dish previously inoculated with eggs for 2 h. Fertilization success was evaluated after 48 h and confirmed after one week with the development to embryos. Successes of the various combinations were compared using ANOVA and Tukey's post hoc test.

Molecular characterization of the hybrid zones
The number of hybrids did not differ between plots within Tromsø and within Blushøj (data not shown); hence, data from both plots were subsequently pooled.  1 and figure 2). By contrast, 8.3% and 3.7% F 1 hybrids, and 23.9% and 12.9% introgressed individuals were found in the youngest contact zones (Heimaey and Blushøj, respectively). All F 1 and almost all hybrids (53 out of 68) were characterized by an Fd chloroplast (figure 2). In both Tromsø and Blushøj, several individuals with an Fs microsatellite genotype possessed an Fd chloroplast, but the reverse was never found.

Fertilization success
Conspecific fertilization successes (both Fd × Fd and Fs × Fs) did not differ significantly among the three contact zones examined (Kirkenes not tested) (67-49%, p = 0.1199) (figure 3). In the young contact zones at Heimaey and Blushøj, interspecific fertilization successes involving Fd egg and Fs sperm were comparable to the conspecific success (38% and 43%, respectively), but the interspecific fertilization successes involving Fs egg and Fd sperm were significantly lower than the conspecific crosses (24% and 10%, respectively, p < 0.001) (figure 3). In sharp contrast, both interspecific crosses showed significantly (p < 0.001) lower success (less than 1.1%) in the oldest contact zone at Tromsø (Kirkenes not tested).

Discussion
Natural selection can drive speciation by 'reinforcing' those mechanisms preventing hybridization when hybrids are maladaptive. As maladaptive hybrids can be viewed as energetically expensive 'mistakes', pre-zygotic mechanisms may evolve to prevent hybridization. These mechanisms range from preventing mating behaviour (through enhancement of species recognition systems) to preventing gamete recognition (fertilization). In particular, pheromones and gamete surface receptor polysaccharides have been reported from a wide variety of eukaryotic species (see [33,35] and references therein) including the brown algae (fucoserratene in F. vesiculosus and F. serratus) [70]. However, these recognition molecules have primarily been studied in the context of gamete attraction rather than as possible pre-zygotic isolation mechanisms. The pattern of greater pre-zygotic isolation in sympatry relative to allopatry is one of the main signatures of reinforcement and has been termed reproductive character displacement' [19,36]. However, establishing an empirical case for reinforcement remains difficult because isolation is not necessarily due to a single isolating factor, but a consequence of several pre-and (possibly) post-zygotic barriers and their interactions. Nevertheless, a number of criteria have been proposed to demonstrate reinforcement [18,33] including heterospecific mating (gene flow between parental species), presence of maladaptive hybrids, frequency of hybridization between populations across time, reproductive character displacement, and the absence of alternative explanations (e.g. ecological divergence via local selection or temporal isolation). The Fd × Fs hybrid zones we examined meet some of the reinforcement criteria. First, hybridization is occurring or has occurred in all contact zones. Both younger contact zones show a high percentage of hybrids (13-24%) and several F 1 individuals. Despite low contemporary hybridization in the older contact zone, there is evidence that hybridization has occurred in the past, as ca 3% of the individuals genotyped in the 'old' contact zones carry alleles from the other species (table 1 and figure 2). Furthermore, one individual with a pure Fs nuclear genotype and a Fd chloroplast DNA was found in Tromsø, resulting from past introgression.
Second, hybridization between Fs and Fd appears to be maladaptive. the Blushøj F 1 hybrids showed a lower fitness compared with the parental species, with selection coefficients against hybrids ca 80% [46] and reproductive F 1 hybrids showed reduced fertility, mostly owing to lower egg quality (e.g. fewer and more variably sized eggs per receptacle). Additionally, laboratory crosses with eggs from F 1 hybrids were significantly less successful both in Blushøj [47] and Heimaey where lower egg quality was also observed (J. A. Coyer and G. Hoarau 2002, unpublished data). Survivorship and fertility of F 1 hybrids were not assessed at Tromsø and Kirkenes as no F 1 hybrids were found in the field. In Fucus, as in several other species, reduction of fertility and potential sterility appear before hybrid inviability [12,71,72].
Third, patterns observed in the Fd × Fs hybrid zones are consistent with the presence of reproductive character displacement. In contrast to virtually all previous studies of reinforcement, which have compared populations of plants and animals in regions of sympatry to those in allopatry, we examined sympatric populations differing in the time of sympatry (100 to 10 000 years), as reinforcement is expected to result in a reduction of interspecific mating through time ( [35] and references therein). Our molecular data suggest that hybridization has decreased with the time of contact. Although based on only two locations per age, both of the younger contact zones revealed a high percentage of hybrids (13-24%) and the presence of F 1 individuals. By contrast, hybridization was very low (2-3%) and no F 1 individuals were found in either of the old contact zones in northern Norway (table 1 and figure 2). Our cross-fertilization experiments further suggested that the reduction in hybridization observed in Tromsø was the result of increased gametic incompatibilities. Whereas interspecific fertilization success (Fd egg and Fs sperm) was comparable to conspecific crosses in the younger contact zones, interspecific fertilization successes decreased to virtually zero in Tromsø (figure 3). While our results cannot provide conclusive evidence for reinforcement (i.e. increased isolation in sympatry compared with allopatry using the temporal comparison) because of low replicates, they are consistent with patterns that would be expected over such a period of time.
In this context, we reviewed a new contact zone in Bergen Harbour, where Fd was introduced to Fs within the past 10 years, where we expected to see many hybrids. Unexpectedly, a genetic survey conducted in 2008 found no hybrids (G. Hoarau and J. A. Coyer 2008, unpublished data). At least three aspects of the Bergen population, however, can temper the contradiction. First, the source of Fd to Bergen Harbour is unknown and to date, impossible to trace (G. Hoarau and J. A. Coyer 2008, unpublished data). The source of Fd in Bergen could be northern Norway and subsequently, a population already displaying strong reinforcement. Second, the phenological overlap in Bergen Harbour is reduced (K. Sjøtun 2008, personal communication) compared with all other zones of contact, thus limiting the temporal window for hybridization. This difference in observed phenology at Bergen, in fact, suggests a northern origin of Fd [73]. Finally, F. distichus is an order of magnitude more abundant than F. serratus in Bergen Harbour (J. A. Coyer, K. Sjøtun and G. Hoarau 2008, unpublished data), thus further limiting the likelihood of hybridization (which requires threshold densities of gametes). These results further highlight the difficulties in definitively proving reinforcement. Nevertheless, we consider the temporal approach involving contact zones compelling (especially, the older ones).
Ecological divergence caused by strong local selection is also a potentially contributing mechanism, especially when gamete recognition systems are weak (reviewed in [74], see also [75,76]). Although Fd and Fs can differ in position on the intertidal shore, with Fd occurring slightly higher on the shore than Fs, wide areas of intermixing were the rule at all four contact zones and our sampling occurred within the areas of intermixing (table 1). Furthermore, no pattern of microhabitat utilization was detected in a detailed spatial analysis of the parental species and F 1 hybrids at Blushøj: mature plants were intermingled and F 1 hybrids occurred throughout the intermingled area rather than clustered around either parental species or themselves [46]. Changes in the mating system, such as increased/decreased self-fertilization in plants, can be another mechanism for reproductive character displacement (reviewed in [36]). Selfing may thus promote prezygotic isolation. In the hermaphroditic Fd, high inbreeding coefficients (up to 10× higher than Fs) have been routinely found in broad surveys (Coyer et al. [59]), whereas in the dioecious Fs, this is seldom the case (Coyer et al. [47]). However, no evidence for mating system change was observed, as inbreeding coefficients (F IS ) for Fd were not significantly different among the contact zones (table 1) and increased levels of self-fertilization did not occur. As shown in this study, as well as Coyer et al. [46], hybridization is asymmetrical with Fd providing the egg most of the time ( figure 3). To our knowledge, no model of hybrid zones has considered a mixed mating system and little more can be said at the present time. What we do know, however, is that hybrids are almost never formed between Fucus species with the same mating systems.
We hypothesized that sexual trait(s) accounting for reinforcement in Fucus would most likely occur at the level of gamete attraction and/or recognition. Gamete recognition proteins/genes in marine organisms have been investigated mainly in marine invertebrates (mussels, snails and sea urchins) [76,77]. In these systems, positive selection (as assessed by high sequence divergence and more nonsynonymous than synonymous substitutions) for sperm-egg interaction proteins has been found in closely related and sympatric species of sea urchins [78,79], but not for species in allopatry [80]. Sympatry among closely related species also seems to be associated with increased sperm specificity. In sea urchins, for example, crosses between allopatric species pairs require substantially less sperm for fertilization than crosses involving sympatric pairs [81]. The sperm binding protein fss27 in Fucus has long been known and shares a number of chemical properties with sea urchin bindin gamete recognition protein [82][83][84]. Recognition between eggs and sperm has been hypothesized to be based upon a receptor mechanism involving ligands (oligosaccharide side-chains of egg surface glycoproteins) and complementary binding proteins [84]. However, the protein and its ligand remain to be sequenced.
Another important evolutionary mechanism potentially occurring at the level of gamete attraction/recognition is sexual conflict [85,86]. For example, an overabundance of sperm increases the risk of polyspermy (multiple sperm entry), which will prevent subsequent development of embryos. Under these conditions, male and female gametes have different priorities: males for fast entry into eggs, females for prevention of polyspermy and slow sperm entry. Any mutation that makes it more difficult to fertilize eggs can concomitantly lower the effective concentration of sperm, thereby allowing eggs to block polyspermy. Thus, a conflict over fertilization rates arises in which females are selected for lower egg-sperm affinity, whereas males are selected for higher affinity. This affinity 'arms race' could lead to divergences in the gamete recognition system and thus to reproductive character displacement, independent from reinforcement [76,87].
Fucus species display a fast and sodium-dependent block to polyspermy [88,89]. Consequently, polyspermy rates in natural populations are higher in brackish (F. vesiculosus [90]) relative to fully marine habitats (F. ceranoides [91], F. distichus [92]). Three of the contact zones examined in this study (Tromsø, Kirkenes and Heimaey) were fully marine and although the Blushøj site could vary between near fully marine (30 psu) to semi-brackish (20 psu) over a time scale of hours to days, polyspermy in F. vesiculosus most commonly occurs at less than 6.5 psu [90]. Consequently, polyspermy is unlikely to drive sexual conflict among Fd and Fs gametes in the contact zones we examined. An additional outcome of divergence from sexual conflict is lower conspecific fertilization success (reviewed in [36]), which was not observed in our study. Therefore, the most parsimonious explanation for the pattern we observed in Fd × Fs hybrid zones is probable reinforcement of pre-zygotic isolation mechanisms.

Conclusion
The genus Fucus, with the combination of contemporary radiation (e.g. F. radicans within the past 400-2000 years; [93], secondary contact zones (e.g. Coyer [46]), extensive hybridization (e.g. [46,50]), and a well-supported phylogenetic [40,42,44] and phylogeographic framework [18,19], e.g. F. serratus [41], e.g. F. distichus [58], e.g. F. vesiculosus and F. spiralis [43]) provides evolutionary biologists with a system phylogenetically distant from, but complementary to, plants and animals. While the results of our study of Fd × Fs hybrid zones do not prove reinforcement, they are consistent with reinforcement of prezygotic isolation as shown by the decreasing rates of hybridization and interspecific fertilization success with increasing time of sympatry. Evidence for reinforcement will be most compelling when observed in species pairs with a well-corroborated historical biogeographical framework, as it the case in Fucus (e.g. [21]).
Data accessibility. The supporting data can be found in the electronic supplementary material.