The haplochromine cichlid species assemblages of Lake Malawi and Victoria represent some of the most important study systems in evolutionary biology. Identifying adaptive divergence between closely-related species can provide important insights into the processes that may have contributed to these spectacular radiations. Here, we studied a pair of sympatric Lake Malawi species, Pseudotropheus fainzilberi and P. emmiltos, whose reproductive isolation depends on olfactory communication. We tested the hypothesis that these species have undergone divergent selection at MHC class II genes, which are known to contribute to olfactory-based mate choice in other taxa.
Divergent selection on functional alleles was inferred from the higher genetic divergence at putative antigen binding sites (ABS) amino acid sequences than at putatively neutrally evolving sites at intron 1, exon 2 synonymous sequences and exon 2 amino acid residues outside the putative ABS. In addition, sympatric populations of these fish species differed significantly in communities of eukaryotic parasites.
We propose that local host-parasite coevolutionary dynamics may have driven adaptive divergence in MHC alleles, influencing odor-mediated mate choice and leading to reproductive isolation. These results provide the first evidence for a novel mechanism of adaptive speciation and the first evidence of adaptive divergence at the MHC in closely related African cichlid fishes.
Citation: Blais J, Rico C, van Oosterhout C, Cable J, Turner GF, Bernatchez L (2007) MHC Adaptive Divergence between Closely Related and Sympatric African Cichlids. PLoS ONE 2(8): e734. https://doi.org/10.1371/journal.pone.0000734
Academic Editor: Richard Copley, Wellcome Trust Centre for Human Genetics, United Kingdom
Received: April 19, 2007; Accepted: July 5, 2007; Published: August 15, 2007
Copyright: © 2007 Blais et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: JB was supported by a post-graduate scholarship from the Natural Science and Engineering Research Council of Canada and Fond Québécois pour la Nature et les Technologies of Quebec. JC was funded by an Advanced Research Fellowship from the Natural and Environment Research Council (NER/J/S/2002/00706) of the UK. Field work by CvO., JB, and JC was funded by the British Ecological Society. CR was supported by the Consejo Superior de Investigacion Scientifica of Spain. GFT was funded by the Biotechnology and Biological Sciences Research Council of the UK and field work by the Natural and Environment Research Council of the UK. LB was funded by the Natural Science and Engineering Research Council of Canada (Discovery Grant Program).
Competing interests: The authors have declared that no competing interests exist.
Explaining why certain ecosystems harbor disproportionately high species richness is a hotly debated issue . Antagonistic coevolution between the sexes or between hosts and parasites is recognized as a powerful force capable of driving rapid evolutionary changes, which contributes significantly to biodiversity , . Since Haldane's  proposal on the importance of parasites resulting in genetic adaptations of the host, researchers have acknowledged their potentially important role in speciation , . More recently, the concept of cospeciation between parasites and their hosts has received increased attention . The theory of the geographic mosaic of coevolution predicts the diversification of allopatric populations of both host and parasite through three-way interactions between their genotypes and the environment .
Consistent with these models, the diversity of parasite species infecting host lineages has been shown to be positively correlated with host species richness in primates  and rodents . Some closely-related species also display striking differences in their response to infectious diseases, as exemplified by the difference between humans and chimpanzees (Pan troglodyte) in their response to infection by hepatitis B and C, HIV and Plasmodium falciparum . Parasitism has also been correlated with increased bacterial host diversity and specialization in microbial communities . Moreover, divergence in MHC genes has been associated with the adaptive radiation of African large barbs from Lake Tana, Ethiopia . Finally, genes showing signs of significant adaptive evolution among rodent and human lineages were found to be related to reproduction, olfaction and immunity , .
In vertebrates, one of the most important factors determining an individual's susceptibility and resistance (defined here as the likelihood of suffering fitness cost from exposure to a pathogen) to specific parasites is its MHC (major histocompatibility complex) genotype . Classical MHC molecules are cell surface glycoproteins whose primary role is binding and presentation of antigens to T cell receptors as part of a “self/non-self” recognition mechanism representing the first step of adaptive immune response . Antigenic peptides are anchored at specific residues commonly found to be under positive selection and called Antigen Binding Sites (ABS). MHC genes that differ at ABS produce molecules that effectively bind and present different antigens to T cells. This will in turn determine whether or not an effective immune response can be mounted against the infectious agent . The MHC thus initiates the adaptive immune defense which is why these genes are thought to be among the most important targets of the host-parasite coevolutionary arms race . Besides their central role in immunity, MHC molecules are also known for their influence, through olfactory signals, on mate choice and kin recognition, producing either assortative or disassortative associations in various species including humans , rodents ,  and fishes , .
Some of the most spectacular examples of tropical biodiversity are the adaptive radiations of East African cichlid fishes of Lake Victoria, Malawi and Tanganyika , . How these hundreds of endemic species evolved from a few founding lineages in a few million years and can coexist within single lakes remains an unresolved issue in evolutionary biology. In Lake Malawi, about 500 endemic species are believed to have diverged from a single lineage within the last 5–7 million years . Particularly intriguing is the apparent lack of clear morphological or ecological adaptive differences among many congeneric species, which are distinguished largely on the basis of colour , . Little is known about adaptive genetic differentiation among closely-related species which may lead to evolution of reproductive isolation.
The possibility that host-parasite coevolution has made a significant contribution to East African cichlid diversity has hitherto received scant attention. Nonetheless, several observations suggest that it might be worthy of consideration. Firstly, the most species-rich groups of haplochromines in Lake Victoria and Malawi (i.e. the rock dwelling Mbipi and Mbuna respectively) have highly disjointed and fragmented populations with pronounced genetic structure . This would make them prone to diverge under local coevolutionary dynamics according to the geographic mosaic theory . Secondly, population densities of rock dwelling haplochromines are typically very high, in the order of tens of thousands per km2 or more , which can favor the spread of infectious diseases . Thirdly, interspecific interactions, like parasitism, are thought to be more intense and complex in tropical ecosystems and might contribute to the higher speciation and specialization observed in these regions . Finally, parasitism has previously been shown to negatively influence male mating success through female choice in a Lake Malawi cichlid of the genus Mchenga (formerly Copadichromis) , and to be negatively correlated with the intensity of carotenoid-based male nuptial coloration and territory size in Pundamilia nyererei from Lake Victoria .
Recently, it has been demonstrated that olfactory communication plays an essential role in assortative mating between the sympatric Lake Malawi cichlids Pseudotropheus emmiltos and P. fainzilberi . Divergent natural selection can cause ecological differentiation and speciation simultaneously when its action on ecological traits affects mate choice as a by-product . Considering the documented influence of MHC genotype on odor-mediated mate choice, we hypothesized that odor differences between these species may involve differences in MHC genes accumulated under divergent host-parasite coevolution. In this context, the goal of this study was to document patterns of genetic differentiation between a pair of closely-related sympatric species at MHC class II B, as predicted if host-parasite coevolution has played a role in the divergence of the closely related P. emmiltos and P. fainzilberi. More, specifically, we predict that such divergence should be particularly prominent when comparing nonsynonymous sequence variation at the exon 2-encoded ABS of different species, as diversifying selection is thought to act most stringently at these sites. Ratios of nonsynonymous to synonymous substitutions (dN/dS) higher than unity were used to identify putative ABS, as these are indicative of positive selection. Comparisons with neutral divergence at non-coding intron or synonymous exon 2 sequences of the same gene were used to assess whether divergent selection had to be invoked to explain observed allelic compositions. We also predicted that divergence in MHC class II should be correlated with detectable differences in parasite communities currently infecting the two species in sympatry. The results show clear evidence of divergent selection at MHC class II B genes between the species. Moreover, we found significant differences in parasite communities collected on each sympatric host species.
Positive Selection and Recombination
MHC class II B intron 1 and exon 2 sequences were obtained from a total of 1022 clones, representing 74 Pseudotropheus emmiltos individuals from two populations (Luwino Reef N = 31; Mpanga Rocks N = 43) and 106 P. fainzilberi individuals from three populations (Luwino Reef N = 31; Mpanga Rocks N = 36; Chirwa Island N = 37) (Fig. 1; Fig. S1). A mean±standard deviation of 5.74±2.22 clones per individual were sequenced. The final dataset contained 419 individual exon 2 haplotypes (GenBank accession numbers: EF539902-EF540320) and a high level of polymorphism and haplotype diversity comprising 196 distinct exon 2 alleles (172 at the amino acid level) and 164 distinct intron 1 alleles (Fig. S2). The mean number of alleles per individual was 2.31 and 2.29, and ranged from one to five and one to six, for exon 2 and intron 1 respectively (Fig. S1). These data are consistent with Malaga-Trillo et al.  that concluded that the number of MHC class II loci per individual varied between one and six (and therefore that the number of alleles varied between one and 12) in Pseudotropheus zebra. Although we cannot be certain that we have detected all alleles present in each individual, our sampling nevertheless provide an unbiased and adequate estimate of population allele frequencies.
Individuals of a) Pseudotropheus fainzilberi and b) P. emmiltos were collected at Mpanga Rocks, Luwino Reef, and Chirwa Island off the North Western shore of Lake Malawi. The two species are sympatric at Mpanga Rocks and Luwino Reef. Photograhs show males in full nuptial dress and are courtesy of Ad Konings.
Likelihood ratio tests comparing selective regimes affecting MHC class II B exon 2 in P. emmiltos and P. fainzilberi showed that global selection pressures affecting exon 2 were not significantly different between species, neither in terms of the proportion of sites under selection nor in terms of strength of selection (Table S1). Site-by-site analyses revealed three sites (72, 85 and 87) evolving under different selective pressures (with p<0.05) between the two species along internal branches and three sites (50, 74 and 87) at the tip of the phylogeny. However, these results were not significant after Bonferroni correction. The identification of codons under significant positive selection was therefore conducted on sequences pooled from both species.
Constraining dN≤dS revealed clear evidence of global positive selection in exon 2 since the likelihood of this model was significantly lower than the one allowing dN>dS (LRT = 648.554; d.f. = 3; p<0.001). The bivariate REL analysis identified four dN/dS rate categories across exon 2, with mean rate estimates of: 0.43, 1.11, 2.40 and 5.75. We considered the two lower rate categories as representing purifying selection and (nearly) neutral evolution respectively, and the two higher rate categories as indicating positive selection. A total of 19 out of 84 codons had an empirical Bayesian posterior probability greater than 0.95 of belonging to one of the two positive selection rate categories (Table S2). Of these, 12 corresponded with human ABS identified from crystallography  (Fig. 2), suggesting that although the functional structure of mammalian and fish MHC class II molecules are approximately similar (Fig. 3; Fig. S3), there appear to be noteworthy differences in the precise position of ABS codons. In mammals and fish, 59 and 58% respectively of MHC class II B ABS are found in the alpha-helices or in coils between adjacent helices forming the edges of the peptide binding groove. A similar proportion (52%) of positively selected sites was found in the alpha-helices of the peptide binding groove of non-classical mhc class Ib genes thought to be involved in pheromone detection in rodents .
The tertiary structure prediction was based on Psemmil_m58_c exon 2 sequence (Accession number: EF540049) of 41% homology with mouse MHC sequence gi|13399459 in the Protein Data Bank (http://www.pdb.org/pdb/home/home.do) using the 3D-jigsaw server v.2.0 (http://www.bmm.icnet.uk/~3djigsaw/). The graphical representation was created using the program Pymol v.0.99. Amino acid residues under significant positive selection in Pseudotropheus fainzilberi and P. emmiltos and corresponding to peptide binding sites in humans are highlighted in orange. Residues shown in red were under significant positive selection in P. fainzilberi and P. emmiltos but do not correspond with peptide binding sites in humans. Residues in yellow are human peptide binding sites that were not found to evolve under positive selection in cichlids. The 84 amplified exon 2 codons are numbered 1-84 on the graph. Those correspond to codons 6-89 of the mature protein.
Structure predictions were obtained from the PSIPRED server (http://bioinf.cs.ucl.ac.uk/psipred/psiform.html) for a) mouse sequence gi:13399459 and b) cichlid sequence Psemmil_m58_c (Accession number: EF540049). Mammalian and fish secondary structures are similar except for an additional short alpha-helix at position 46–49 predicted for the mouse sequence, and an additional short strand at position 74 predicted for the cichlid sequence.
We compared intron 1 and exon 2 mean genetic distances (d and dS respectively) within 14 functional clades defined using these positively selected sites (from an amino acid p-distance neighbor-joining phylogeny, Fig. S4). Significantly different mean genetic distances between intron 1 and exon 2 were detected in at least eight clades, which is evidence for intron-exon recombination  (Table S3). In three of these, mean d at intron 1 was significantly lower than mean dS at exon 2, suggesting the action of within-locus recombination and balancing selection on functional exon 2 alleles . The remaining five cases had mean d at intron 1 higher than mean dS at exon 2, suggesting inter-locus recombination . However, among those eight clades, only clade 14 had bootstrap support higher than 60% and thus results from the other seven clades must be viewed with great caution. Nevertheless, a single breakpoint recombination analysis  revealed that the most likely breakpoint was at nucleotide position 231. Including this recombination breakpoint greatly improved the phylogenetic model of intron 1 and exon 2 evolution (ΔAICc = 13001.0; ΔBIC = 3667.85). This position is located only 15 nucleotides upstream from the intron 1-exon 2 junction of the alignment, corroborating the scenario of recombination between intron 1 and exon 2.
The association index (AI) value calculated from the intron 1 phylogeny (0.69) was higher than that computed from the exon 2 phylogeny (0.61), suggesting that phylogenetic divergence was higher at exon 2 than at non-coding intron 1 (Table 1). Similarly, genetic differentiation measured by Fst value based on overall mean pairwise genetic distances was higher at exon 2 putative ABS amino acid sites (Fst = 0.097; p<0.01) than at neutrally evolving intron 1 (Fst = 0.021; p<0.01) or synonymous exon 2 alleles (Fst = 0.021; p<0.01). Both findings are contrary to what is predicted from a scenario of balancing selection with no divergent selection at exon 2. In addition, within exon 2, Fst was higher for positively selected putative ABS (Fst = 0.097; p<0.01) than for sites outside this region (Fst = 0.016; p = 0.015) (Table 1). Moreover, higher divergence at putative ABS compared to neutral alleles was observed for both heterospecific sympatric comparisons, but for none of the conspecific allopatric comparisons where functional Fst estimates at ABS were all negative and therefore technically equal to zero (Fig. 4).
Fst values (HudSon et al. 1992) from pairwise comparisons of conspecific allopatric populations (Mpanga Rocks (Mpg), Luwino Reef (Lwo), and Chirwa Island (Chw)) and between heterospecific sympatric populations of P. fainzilberi (fainz.), and P. emmiltos (emm.). Neutral Fst estimates were obtained from intron 1 p-distances, modified Nei and Gojobori synonymous distances at exon 2, and from amino acid EX distance at exon 2 sites outside the putative ABS region (dN/dS≤1) and functional Fst were obtained from EX distance at putative exon 2 ABS (dN/dS>1). Negative Fst estimates were forced to zero. Evidence of significant genetic divergence between pairs of allopatric populations was found only for neutral Fst estimates between P. emmiltos populations from Luwino Reef and Mpanga Rocks. Evidence of significant genetic divergence between pairs of P. fainzilberi and P. emmiltos sympatric populations was found only at putative ABS at Mpanga Rocks. *Significantly different from zero (p<0.05; 200 permutations)
Finally, there were 31 shared and 23 private alleles at intron 1 between P. fainzilberi and P. emmiltos, compared to 22 shared and 41 private alleles at exon 2 amino acid sequences. These raw proportions were significantly different (two-tailed Fisher's Exact Test: p = 0.017). There was no significant difference in allele frequency distributions between exon 2 and intron 1 (Fig. S5), (mean f intron 1: 0.015, mean f exon 2: 0.013, Mann-Whitney U = 1558.000, p = 0.437; intron 1 σ2: 0.246, exon 2 σ2: 0.227; Levene's test F = 0.435, p = 0.511). There was, however, a highly significant difference in shared and private allele composition as measured by Dc (Table 2). This difference was significant between intron 1 and exon 2 alleles when considering the whole exon 2 sequence, and between intron 1 and exon 2 alleles at the 19 positively selected putative ABS, but not between intron 1 and the 65 exon 2 residues outside this region (Table 2).
Comparison of Parasite Communities
Sixteen categories of parasites were identified in both species and their abundance recorded on each fish (Table 3). A Principal Component Analysis on the dataset  returned four factors explaining 84% of the original variance (Table 3). PC1 represented mainly a negative correlation between infection by acanthocephalans and stomach cysts versus infection by muscle-encysted digenean metacercaria, PC2 between infection by muscle-encysted digenean metacercaria and by stomach cysts, and PC3 between infection by ciliate Trichodina sp. and gill parasitic copepod sp. 3 versus infection by acanthocephalans. Finally, PC4 largely represented a contrast between infection by gill parasitic copepod sp. 3 and ciliate Trichodina sp.
A multivariate linear model of the relationship between these first four factors representing infecting parasite communities and host species, host sex, host standard length and collection site returned R2 values of 0.754, 0.473, 0.178 and 0.010 for PC1 to PC4 respectively. Overall, 26% of the variance in these four factors was explained by variation among collection sites, while host standard length, host sex and host species accounted for 3.6, 3.3, and 1.8% of the variance, respectively (Table 4). There were significant differences between P. fainzilberi and P. emmiltos in PC1 and PC3 (Table 4). Least square means for PC1 and PC3 were 0.263 and −0.372 for P. emmiltos and 0.027 and 0.067 for P. fainzilberi respectively. Hence, P. emmiltos tended to have more stomach cysts and muscle encysted digenean metacercaria relative to acanthocephalan infections (PC1) than P. fainzilberi. Conversely, P. emmiltos had less trichodinids and gill infecting parasitic copepod sp 3 relative to acanthocephalan infection (PC3) than P. fainzilberi.
Antagonistic host-parasite interactions involving coevolutionary responses have the potential to produce rapid evolutionary changes . Genes directly implicated in these interactions are often polymorphic  and may diverge quickly if there are differences in selective pressures among populations . Such defense genes are likely candidates to show the first signs of genetic divergence in recently-speciated taxa. Whereas the importance of divergence in traits related to defense against parasites and other enemies (e.g. phytophagous insects) has long been recognized in studies of plant radiations, it has remained largely ignored in the study of adaptive radiations of animals . Here, we tested the hypothesis that two recently diverged species experienced divergent selection at important immunological genes of the MHC class II. Furthermore, we also tested for possible differentiation in parasite communities infecting the species in sympatry, which we hypothesized could be a driver of divergent evolution at the MHC.
Evidence for strong balancing and positive selection on MHC class II loci was manifested by the high polymorphism and haplotype diversity and a clear signal of positive selection across the 84 codons of exon 2. Differentiation levels expressed in the association index (AI) and Fst, as well as proportion of shared and private alleles for exon 2 of MHC class II B were contrasted with neutral divergence at intron 1 and synonymous exon 2 polymorphisms. These comparisons revealed that the observed patterns of functional allelic composition of the two species cannot be explained by divergence due to drift alone.
Four arguments support the contention that diversifying selection augmented genetic differentiation of parts of the MHC genes between the two cichlid species. Firstly, without divergent selection, genes under balancing selection are expected to show less differentiation between populations than neutral loci , . However, despite recombination between intron 1 and exon 2, divergence estimated by AI values was higher for the exon 2 phylogeny than for the intron 1 phylogeny. Secondly, neutral Fst from intron 1, exon 2 dS, or sites with dN/dS≤1 in exon 2 was about 2%, while this value was almost five times higher in putative ABS. Thirdly, this pattern of higher divergence at putative ABS was observed between both heterospecific sympatric population pairs but for none of the allopatric conspecific population pairs, suggesting that this is a species-specific adaptive divergence that persists despite potential gene flow between species in sympatry. Finally, the significantly higher proportion of private alleles at exon 2 amino acid sequences compared with non-coding intron 1, provides strong additional evidence that despite close physical linkage between intron and exon sequences, divergent selection contributed to the shaping of the functional allelic composition of the species.
Adaptive divergence at immune system genes is likely to be correlated with differences in resistance and susceptibility to parasites. Our results are consistent with this prediction as there were significant, albeit relatively small, interspecific differences for PC1 and PC3 while controlling for host sex, standard length and collection site. Apart from acanthocephalans and unidentified stomach cysts, the other parasites represented by PC1 and PC3 (namely digenean metacercaria, trichodinids and gill infecting parasitic copepod) are known to actively infect the fish from the water column. Acanthocephalans are normally acquired via the food chain. There is therefore two distinct modes of transmission involved. Thus we cannot rule out the possibility that species differences in susceptibility stems from differences in microhabitat or diet rather than from genetic differences . However, there is substantial overlap in microhabitat and both species feed by combing attached filamentous algae with their specialized bi- and tricuspid teeth . In addition, at sampling locations, males of both species have overlapping territories which they use for foraging.
A substantial amount of variance (26%) in infecting parasite communities was explained by variation among collection sites. Local variation in abundance of specific parasites can be due to transient epidemics or alternatively, to more stable variation in population densities caused by habitat or environmental differences. The latter case provides the raw material for coevolutionary divergence in genes involved in parasite resistance such as MHC between allopatric populations exposed to different parasite environments . Upon secondary contact, such differences accumulated in allopatry could be manifested by distinct infection patterns of the species such as those detected here. Alternatively, initial divergence among allopatric populations in trophic niche could have lead to different levels of exposures to parasites transmitted via the food chain such as acanthocephalans, which in turn may select for different MHC alleles. These processes of differentiation could be particularly important in species with pronounced genetic structure such as rock-dwelling haplochromines. Divergence in immune system does not imply detectable divergence in morphology and may therefore be overlooked in many animal systems. Nevertheless, divergence in immunity could have important consequences on foraging or reproductive behaviors that may in turn promote speciation or species coexistence. The phenomenon could be more widespread in animal radiations than currently appreciated. For example, divergence in MHC class II has been observed in the radiation of large barbs (cyprinid fishes) in Ethiopia's Lake Tana . In the cyprinodont Fundulus heteroclitus, divergent selection in MHC class II among populations was inferred from population specific amino acid replacement at putative ABS .
Identifying adaptive genetic differentiation between closely related species is an important step toward understanding speciation processes. Unfortunately, this critical information has been lacking in the case of African haplochromine cichlids. The impressive species diversity and endemism of cichlid species assemblages from Lake Malawi, Victoria and Tanganyika have established the systems as some of the most important models in evolutionary biology . Part of this fascination comes from the surprising ecological and morphological similarity of congeners, which led some to suggest that natural selection had little importance in recent speciation events , . The results presented here provide one of the first evidence of adaptive genetic evolution in African cichlids from the African Great Lake radiations [see also 47], . We hypothesize that this adaptive differentiation may have fuelled the speciation process as a by-product by augmenting MHC-mediated mate choice signals. Natural selection from host-parasite coevolution may thus have played a more important role in recent speciation of endemic African cichlid lineages than suggested by morphological similarities.
Sexual selection has long been regarded as an important factor in haplochromine speciation – while the importance of parasitism in sexual selection and mate choice has been discussed extensively –. Several cases of assortative MHC-based mate choice have been documented. Thus, MHC genotypes represent an ecological trait whose divergence by natural selection may directly produce reproductive isolation as a by-product. For instance, threespine stickleback females prefer to mate with males having an intermediate rather than maximal number of MHC class II alleles different from those the females possess . Female preference for intermediate MHC similarity has also been found in humans . Malagasy giant jumping rat (Hypogeomys antimena) females preferred males with MHC genotypes more similar to themselves . Possible causes of mate preferences aimed at limiting offsprings' MHC diversity include detrimental effect of excessive MHC diversity caused by: negative thymic selection of T cell lines , , cross-reactive autoimmune diseases  and reduced concentration of specific MHC-peptide ligands on the surface of antigen presenting cells for effective T cell activation . We therefore speculate that multilocus MHC genotype compatibility may exclude matings between P. fainzilberi and P. emmiltos through olfactory signals part of a mate choice rule including some limits on MHC dissimilarity of mates. This would represent a case of “one-allele” speciation mechanism  as the same mate choice alleles in both species would reduce gene flow. Behavioral studies on the precise influence of MHC genotype on mate choice in these two species may elucidate the role that this divergence has played in the evolution of reproductive isolation.
Materials and Methods
Study Sites and Field Sampling
Pseudotropheus emmiltos and P. fainzilberi are among the 17–44 currently recognized species of the P. zebra species complex (subgenus Maylandia), members of which are found all around in Lake Malawi . The two species are sympatric at Mpanga Rock and Luwino Reef off the north western coast (Fig. 1), the only two sites where P. emmiltos is known to occur . P. fainzilberi is found at several other localities both north and south of the zone of sympatry. Thus, the range of P. emmiltos is nested within the range of P. fainzilberi. Mitochondrial DNA analyses using the model of isolation with migration implemented in the program IM  revealed that the two species diverged around 30–45 000 years ago and that the most probable speciation scenario was one of allopatric divergence followed by secondary contact (Turner et al. in prep.). The most noticeable morphological difference between the taxa is the color of the dorsal fin, which is orange in P. emmiltos and blue with a black longitudinal stripe in P. fainzilberi (Fig. 1). The two species are behaviorally reproductively isolated and olfactory signals are involved in mate discrimination . Fifty individuals of each species where caught at Mpanga Rocks and Luwino Reef by SCUBA divers using monofilament gill nets. Fifty individuals of P. fainzilberi were also captured at Chirwa Island, two kilometers south of the zone of sympatry (Fig. 1).
Cloning and DNA Sequencing
MHC class II molecules are expressed on the surface of antigen presenting cells and bind antigenic peptides in a cleft formed by the α1 and β1 domains in which peptides are anchored in pockets of specific residues (ABS). Bound peptides are then presented to CD4+ T cells. Sequence polymorphisms at these sites are believed to result from positive and balancing selection , . Intron 1 along with 252 bp of exon 2 (codons 6-89 of the mature protein), coding for the β1 domain of MHC class II genes  were PCR amplified using the degenerate primer pair TU377-TU383 , which amplifies up to eight expressed putative loci in African cichlids according to comparisons of individual haplotypes obtained from genomic and cDNA clones [32; Blais et al. unpublished results]. PCR conditions consisted of 30 cycles of 30 sec at 95°, 60 sec at 56°, 60 sec at 72° in a 20 µl reaction volume containing 0.25mM of each dNTPs, 2 µl of 10× reaction buffer, 2 mM MgCl2, 0.5U of Taq DNA polymerase, and 25 ρmol of each primer.
PCR products were purified by ExoSAP-IT (GE healthcare, Chalfont St. Giles, U.K.) following manufacturer's instructions, ligated into pGEM®-T vector (Promega, Madison, WI, USA) and cloned using JM109 Competent Cells (Promega, Madison, WI, USA). Between eight and 30 recombinant clones per individual were amplified with primers OMNI (5′-ACA GGA AAC AGC TAT GAC CAT GAT-3′) and UNI (5′-CGA CGT TGT AAA ACG AGG CCA GT-3′) under the same PCR conditions described above, and sized on 1.5% agarose gel. Between four and 23 clones containing inserts of expected size (∼320–500 bp depending on intron 1 length) per individual were purified with ExoSAP-IT (GE healthcare, Chalfont St. Giles, U.K.), sequenced using Bigdye 2.0 dye terminator kit (Applied Biosystems, Foster City, CA, USA), cleaned with Sephadex™ and analyzed on an ABI 3100 DNA Analyzer (Applied Biosystems, Foster City, CA, USA). Sequence alignments produced by ClustalX v.1.83  were examined to detect false allelic variants due to PCR errors by comparing sequences within and among individuals. First, sequence differences of less than 3 bp when compared to multiple identical sequence replicates from different clones within individuals were considered to be PCR or cloning artifacts and removed from the alignment. Second, only sequences found in at least two independent PCR (within or among individuals) were used for subsequent analyses .
Identification of ABS
There is at present no available information from X-ray diffraction or NMR analyses about the three-dimensional structure of non-mammalian MHC. It is therefore impossible to identify with certainty which sites are involved in peptide binding. In order to identify putative antigen binding sites (ABS) within the peptide binding cleft involved in the coevolutionary arms race, we determined which codons were evolving under positive selection, i.e. having a nonsynonymous/synonymous substitution ratio (dN/dS) significantly higher than unity. We also compared the location of these with mammalian ABS identified from crystallography. The following and all subsequent sequence analyses were conducted using the HYPHY package and programming language running on a 32 processors SGI Altix3700 computer at the Bioinformatics and Computational Biology Centre of University Laval, under a maximum likelihood framework, unless otherwise stated .
According to Kosakovsky Pond et al.  and as implemented by SingleBreakpointRecomb.bf in HYPHY, we determined the most likely recombination breakpoint by computing the likelihood of neighbor-joining phylogenies  for the entire intron-exon alignment and for models fitting branch length independently for partition blocks on each side of a single break point located at each nucleotide position sequentially. Akaike information criterion with second-order correction (AICc) and Bayesian information criterion (BIC) were used to determine which was the most likely recombination breakpoint in the sequence and if including recombination improved the fit of the phylogenetic model (SingleBreakpointRecomb.bf in HYPHY). Using a method assuming a single breakpoint was shown  to perform as well or better than 14 different recombination detection methods evaluated by Posada and Crandall  and is particularly well suited for relatively short alignments like the one studied here where one is interested in evaluating the likelihood of recombination at a specific location along the sequence.
In the next sections, we explain the methodology used to examine evidence for different selection pressures: (1) positive selection, (2) balancing selection and (3) divergent selection. Here, we define positive selection as selection for improved recognition of parasites. Positive selection pressure favors nonsynonymous substitutions particularly in the ABS.
Balancing selection is thought to be the dominant force maintaining polymorphism in classical MHC genes , , . Overdominance and negative frequency dependent selection are types of balancing selection that operate within populations. In the absence of divergent selection, balancing selection is expected to prevent or delay population divergence , and some empirical studies suggested that genetic differentiation between populations is lower for functional MHC alleles than for neutrally evolving loci . Balancing selection thus maintains a high level of polymorphism within populations but reduces genetic divergence between populations.
Finally, divergent selection occurs when different selection pressures act between populations, resulting in higher genetic differentiation, particularly at dN of ABS of allopatric populations. If populations differ in local (positive) selective pressures, divergent selection may maintain different sets of alleles and result in increased differentiation . Divergent selection could thus counteract the affect of balancing selection and increase the extent of population differentiation at a level above that expected for neutral genes [reviewed in 15].
Before determining which individual codons were under positive selection, we first verified whether the selective regimes affecting each species were the same in terms of proportion of sites under selection and the strength of selection. We therefore constrained the substitution rate parameters by setting it equal for both species, and compared the fit of this model with models that could use different parameter setting for each species. The fit of the models were compared using likelihood ratio tests (LRT) with CompareDistribution.bf in HYPHY. A codon-based site-by-site fixed effect likelihood (FEL) analysis comparing selective pressures between the two species along internal branches and leaves of the phylogenetic tree was also performed (ComparesSelectivePressure.bf in HYPHY) . This analysis tests the null hypothesis that the ratio of nonsynonymous to synonymous substitutions is significantly different between the two species for each codon site using likelihood ratio tests.
Positive selection acting on exon 2 was tested by comparing a random effect likelihood (REL) dual model  where both dN and dS are free to vary and are drawn from a general discrete distribution with N = 4 rate classes but where dN is constrained to be≤dS, with a model where dN is free to be>dS . A likelihood ratio test (LRT) based on the Chi-square distribution with N–1 degrees of freedom was used to determine whether allowing dN>dS significantly improved the fit of the model. Individual codons evolving under significant positive selection were identified using a REL method where both dN and dS are free to vary and were drawn from a bivariate general discrete distribution with four dN and four dS rate categories (dNdSBivariateRates.bf in HYPHY). The model fits the parameters of this bivariate general discrete distribution of rates across the sequences given the phylogeny. Positive selection at specific codons is inferred for sites having an empirical Bayesian posterior probability≥0.95 to be in one of the dN/dS>1 categories. The empirical Bayesian approach is based on the maximum likelihood estimates of the rate distribution parameters . The results were then compared with sites of the cichlid MHC sequence corresponding to human ABS deduced from three-dimensional information.
Following Hughes , evidence of both balancing selection and recombination between exon 2 and intron 1 in each species was evaluated by comparing mean genetic distance (d) among intron 1 alleles and mean synonymous genetic distance (dS) at functional exon 2 alleles within functional clades defined using putative ABS identified by the method outlined above. Recombination between introns and exons creates incongruence between intron and exon phylogenies. This translates into significantly lower mean genetic distance among intron alleles than exon 2 alleles within exon 2 functional clades, assuming recombination occurs within locus and that balancing selection acts on exon 2 functional allelic variants . Indeed, balancing selection on the ABS can maintain synonymous sequence variation in exon 2 due to genetic hitchhiking, while recombination causes intron 1 to loose polymorphism by genetic drift. On the other hand, higher genetic distance at introns than at exon 2 within functional clades is expected when recombination occurs predominantly between loci .
We tested for the presence of divergent selection using three different approaches. First, the level of phylogenetic compartmentalization at non coding intron 1 was compared with that observed at exon 2 using the association index (AI) developed by Wang et al.  and implemented in HYPHY. An association value, d, for the tree is calculated by summation of d-values from each node, according to d = (1–f)/2n−1, where n is the number of sequences below the node and f is the frequency of sequences from the most common species in the clade. Values of d expected from the null hypothesis (i.e. no phylogenetic grouping) were calculated by random reassignment of the sequences to the different species. The association index (AI) represents the mean ratio of 100 bootstrap replicates of the association value calculated from the test sequences to those of species-reassigned controls from 200 permutations . Smaller AI values indicate higher divergence of samples (i.e. more homogenous clades). Smaller AI values are predicted for intron 1 phylogeny than for exon 2 phylogeny under a pure scenario of balancing selection acting on ABS in exon 2 involving intron-exon recombination . By contrast, divergent selection between species at exon 2 is predicted to result in relatively high AI value for intron 1.
Secondly, estimates of Fst based on pairwise genetic distances  were contrasted between neutral (non-coding or synonymous alleles) and functional alleles (i.e. alleles differing in putative ABS amino acid residues). Divergent selection can be inferred in cases where Fst estimates from functional alleles exceed those obtained from neutrally evolving alleles. We calculated neutral Fst estimates using nucleotide p-distance for intron 1, as well as modified Nei and Gojobori dS at exon 2. We then computed functional divergence at exon 2 using the amino acid experimental exchangeability distance (EX)  estimated by maximum likelihood in HYPHY, for the 65 sites with dN/dS≤1 and for the 19 sites with dN/dS>1 separately. The EX distance is based on experimental assays of the severity-of-effect of amino acid substitutions on protein function and was specifically designed to reflect functional differences in amino acid sequences while being free of mutational bias. Adaptive divergence between species was first tested using Fst values calculated from sequences of all individuals. In order to test whether observed divergence patterns were species specific and thus could be associated with putative speciation processes, we also compared neutral and functional Fst estimates obtained from the two heterospecific sympatric pairs of populations with those obtained from all pairwise comparisons of allopatric conspecific populations.
Third, we compared haplotype compositions of exon 2 amino acid alleles with non-coding intron 1 alleles. In genes such as classical MHC where haplotype diversity is typically very high, most haplotypes may appear only once in the sample and methods based on haplotype frequencies, such as Gst , have little or no power , . To circumvent this problem, we contrasted the relative proportion of shared and private alleles at intron 1 and exon 2 amino acid sequences. Divergent selection for alternative sets of functional exon 2 alleles predicts that the proportion of private alleles should be higher for exon 2 amino acid sequences than for non coding intron 1 sequences. Alleles that were observed only once were omitted as these cannot be shared by both species. We then tallied the number of ‘private alleles’ (i.e. alleles that were observed at least twice in only one of the two species) and used Fisher's exact test on observed proportions of shared and private alleles at intron 1 and exon 2. We also quantified the relative proportion of private and shared alleles by assigning allelic distinctness scores Dc to each allele defined as allele frequency f in the case of private alleles, and as-(1-fs) in the case of shared alleles, where fs is the frequency of the allele in the species where it is most common. Thus, according to this index, Dc is bounded by −1 and 1 (private alleles have positive scores and shared alleles have negative scores), and more weight is given to common private alleles and to rare shared alleles, reflecting the likelihood of sampling each category. We tested for a significant difference between intron 1 and exon 2 in the composition of private and shared alleles as measured by Dc using the non-parametric Mann-Whitney U test.
Parasite Screening and Analysis
Heterospecific populations in sympatry were predicted to show different parasite communities due to independent host-parasite coevolution (i.e. a prediction consistent with divergent selection). We therefore dissected a total of 81 fish captured at Mpanga Rocks (P. fainzilberi N = 16; P. emmiltos N = 18), Luwino reef (P. fainzilberi N = 18; P. emmiltos N = 17) and Chirwa Island (P. fainzilberi N = 12) by SCUBA diving using gill nets in September 2005. The fish were examined for all eukaryote parasites within 24 h of collection. Parasites incidence and abundance were recorded by killing the fish in an overdose of 0.2% MS222 and examining the dissected organs under a stereo-microscope with cold light illumination. The resulting parasite community dataset was subjected to a linear transformation according to , where yij denotes the observation for parasite type j, on host i and yi+ is the sum of observations for host i [eq. 12 in 38]. As a result of this transformation, a principal component analysis (PCA) performed on the transformed data is based on Hellinger distances, suitable for community data comparisons, rather than Euclidian distances . Thus, to reduce the dimensionality of the data, the minimum number of factors required to explain at least 80% of the original variance were retained from a PCA on the transformed dataset. These factors were then used in a multivariate general linear model to test for differences in relative abundance and composition of parasite communities infecting the two host species. The model included host species as categorical predictor, as well as collection site, host sex and host standard length as additional categorical and continuous covariates. Confidence intervals around model coefficients were obtained from 1000 bootstrap replicates of the original dataset in S-Plus.
List of intron 1 and exon 2 alleles found in each sampled individual. Each individual fish identifier is given in the leftmost column followed by its species, population and its alleles. Alleles are numbered 1-164 and 1-196 for intron 1 and exon 2 alleles respectively. Individual sequences were deposited in the GenBank database under accession number EF539902-EF540320.
(0.09 MB XLS)
GenBank accession numbers for intron 1 and exon 2 alleles. Accession numbers for sequences corresponding to intron 1 and exon 2 alleles as numbered and listed in figure S1.
(0.10 MB XLS)
Structural Alignment of Cichlid MHC Class II Exon 2 Amino Acid Sequences to Protein Data Bases. The HHpred server (http://toolkit.tuebingen.mpg.de/hhpred) using a profil hidden Markov model was used to align cichlid sequences to the 18 protein family alignment data bases currently available on the server. Q ss_pred: query secondary structure as predicted by PSIPRED; Q ss_conf: PSIPRED confidence values (0–9); Q query_name: query sequence; Q Consensus: query alignment consensus sequence; −quality of colum-column match (very bad = ; bad −; neutral .; good +; very good |; T Consensus: template alignment consensus sequence; T templ_name: template sequence; T ss_dssp: template secondary structure as determined by DSSP; T ss_pred: template secondary structure as predicted by PSIPRED; T ss_conf: PSIPRED confidence values (0–9); H: alpha-helix; E: extended beta sheet; C: coil; S: bend; B : beta brigded residue; T : turn; G : 3_10 helix residue.
(0.16 MB TXT)
Neighbor-Joining Phylogeny of MHC Class II B Functional Alleles. Neighbor-joining phylogeny of the 130 distinct alleles at 19 positively selected putative ABS amino acid sequences based on p-distances. Node bootstrap support (from 500 replicates) of more than 50% are indicated. Labeled clades were those used in intron 1 d and exon 2 dS comparisons.
(0.05 MB EPS)
Exon 2 and Intron 1 Allele Frequency Distributions in Pseudotropheus fainzilberi and P. emmiltos. Histograms of allele frequencies showing similar distributions for exon 2 amino acid sequences and intron 1 sequences. Private and shared alleles between the two species are also plotted separately. All alleles with an absolute frequency≥2 were included.
(0.07 MB EPS)
Likelihood ratio tests comparing selective regimes affecting MHC class II B exon 2 in Pseudotropheus fainzilberi and P. emmiltos. Likelihood values under each of the alternative hypotheses were obtained from fitting models where species specific parameters were free to vary independently. Selective regimes were not significantly different between the species indicating that functional constraints were similar and therefore that selective pressures at individual sites could be inferred from the combined datasets.
(0.03 MB DOC)
Empirical Bayesian posterior probabilities of belonging to each of the four substitution rate categories defined by a bivariate REL analysis. The selective pressures associated with each categories were defined as follow: purifying selection: dN/dS = 0.432; (nearly) neutral evolution: dN/dS = 1.107; positive selection dN/dS = 2.404, dN/dS = 5.746. Probabilities are reported for each of the 19 codons under positive selection of MHC class II β exon 2 in Pseudotropheus fainzilberi and P. emmiltos. Codon position numbers are those of the mature protein. Probabilities higher than 5% are highlighted in bold.
(0.04 MB DOC)
Mean d, dS, and dN within clades of a neighbor joining tree built using amino acid p-distances at 19 positively selected putative ABS codons of MHC class II β. dS and dN were computed using the modified Nei and Gojobori method with a transition transversion ration of 1.26 (estimated from the data). d at intron 1 was measured using the Jukes and Cantor distance. Standard errors (SE) were computed from 500 bootstrap replicates. All calculations were done in MEGA v.3.1. Significantly (α = 0.05) different mean d in intron 1 and mean dS in exon 2 are marked in bold.
(0.06 MB DOC)
We thank Jérôme Laroche from the Centre de Bio-Informatique et de Biologie Computationnelle de l'Université Laval for bioinformatics support. We are also grateful to Ad Konings for the photographs of specimens.
Conceived and designed the experiments: LB GT JB CR. Performed the experiments: JB. Analyzed the data: JB. Contributed reagents/materials/analysis tools: CR. Wrote the paper: LB GT JB CR Cv JC. Other: Provided parasitological expertise in parasite collection and identification: JC. Contributed theoretical and analytical discussions on interpretation of MHC evolution: Cv.
- 1. Willig MR, Kaufman DM, Stevens RD (2003) Latitudinal gradients of biodiversity: pattern, process, scale, and synthesis. Ann Rev Ecol Evol Syst 34: 273–309.MR WilligDM KaufmanRD Stevens2003Latitudinal gradients of biodiversity: pattern, process, scale, and synthesis.Ann Rev Ecol Evol Syst34273309
- 2. Buckling A, Rainey PB (2002) The role of parasites in sympatric and allopatric host diversification. Nature 420: 496–499.A. BucklingPB Rainey2002The role of parasites in sympatric and allopatric host diversification.Nature420496499
- 3. Summers K, McKeon S, Sellars J, Keusenkothen M, Morris J, et al. (2003) Parasitic exploitation as an engine of diversity. Biol Rev 78: 639–675.K. SummersS. McKeonJ. SellarsM. KeusenkothenJ. Morris2003Parasitic exploitation as an engine of diversity.Biol Rev78639675
- 4. Haldane JBS (1949) Disease and evolution. La Ricerca Scientifica Suppl A 19: 68–76.JBS Haldane1949Disease and evolution.La Ricerca Scientifica Suppl A196876
- 5. Lederberg J (1999) J. B. S. Haldane (1949) on infectious disease and evolution. Genetics 153: 1–3.J. Lederberg1999J. B. S. Haldane (1949) on infectious disease and evolution.Genetics15313
- 6. Turelli M, Barton NH, Coyne JA (2001) Theory and speciation. Trends Ecol Evol 16: 330–343.M. TurelliNH BartonJA Coyne2001Theory and speciation.Trends Ecol Evol16330343
- 7. Switzer WM, Salemi M, Shanmugam V, Gao F, Cong M-e, et al. (2005) Ancient co-speciation of simian foamy viruses and primates. Nature 434: 376–380.WM SwitzerM. SalemiV. ShanmugamF. GaoM-e Cong2005Ancient co-speciation of simian foamy viruses and primates.Nature434376380
- 8. Thompson JN (2005) The Geographic Mosaic of Coevolution. Chicago: The University of Chicago Press. JN Thompson2005The Geographic Mosaic of Coevolution.ChicagoThe University of Chicago Press
- 9. Nunn CL, Altizer S, Sechrest W, Jones KE, Barton RA, et al. (2004) Parasites and the evolutionary diversification of primate clades. Am Nat 164: S90–S103.CL NunnS. AltizerW. SechrestKE JonesRA Barton2004Parasites and the evolutionary diversification of primate clades.Am Nat164S90S103
- 10. Krasnov BR, Shenbrot GI, Khokhlova IS, Degen AA (2004) Relationship between host diversity and parasite diversity: flea assemblages on small mammals. J Biogeogr 31: 1857–1866.BR KrasnovGI ShenbrotIS KhokhlovaAA Degen2004Relationship between host diversity and parasite diversity: flea assemblages on small mammals.J Biogeogr3118571866
- 11. Anzai T, Shiina T, Kimura N, Yanagiya K, Kohara S, et al. (2003) Comparative sequencing of human and chimpanzee MHC class I regions unveils insertions/deletions as the major path to genomic divergence. Proc Natl Acad Sci USA 100: 7708–7713.T. AnzaiT. ShiinaN. KimuraK. YanagiyaS. Kohara2003Comparative sequencing of human and chimpanzee MHC class I regions unveils insertions/deletions as the major path to genomic divergence.Proc Natl Acad Sci USA10077087713
- 12. Kruiswijk CP, Hermsen T, Heerwaarden Jv, Dixon B, Savelkoul HFJ, et al. (2005) Major histocompatibility genes in the Lake Tana African large barb species flock: evidence for complete partitioning of class II B, but not class I, genes among different species. Immunog 56: 894–908.CP KruiswijkT. HermsenJv HeerwaardenB. DixonHFJ Savelkoul2005Major histocompatibility genes in the Lake Tana African large barb species flock: evidence for complete partitioning of class II B, but not class I, genes among different species.Immunog56894908
- 13. Waterston RH, Lindblad-Toh K, Birney E, Rogers J, Abril JF, et al. (2002) initial sequencing and comparative analysis of the mouse genome. Nature 420: 520–562.RH WaterstonK. Lindblad-TohE. BirneyJ. RogersJF Abril2002initial sequencing and comparative analysis of the mouse genome.Nature420520562
- 14. Gibbs RA, Weinstock GM, et al. (2004) Genome sequence of the Brown Norway rat yields insights into mammalian evolution. Nature 428: 493–521.RA GibbsGM Weinstock2004Genome sequence of the Brown Norway rat yields insights into mammalian evolution.Nature428493521
- 15. Bernatchez L, Landry C (2003) MHC studies in nonmodel vertebrates: what have we learned about natural selection in 15 years? J Evol Biol 16: 363–377.L. BernatchezC. Landry2003MHC studies in nonmodel vertebrates: what have we learned about natural selection in 15 years?J Evol Biol16363377
- 16. Bjorkman PJ, Saper MA, Samraoui B, Bennett WS, Strominger JL, et al. (1987) The foreign antigen binding site and T cell recognition regions of class I histocompatibility antigens. Nature 329: 512–518.PJ BjorkmanMA SaperB. SamraouiWS BennettJL Strominger1987The foreign antigen binding site and T cell recognition regions of class I histocompatibility antigens.Nature329512518
- 17. Jacob S, McClintock MK, Zelano B, Ober C (2002) Paternally inherited HLA alleles are associated with women's choice of male odor. Nat Gen 30: 175–179.S. JacobMK McClintockB. ZelanoC. Ober2002Paternally inherited HLA alleles are associated with women's choice of male odor.Nat Gen30175179
- 18. Carroll LS, Penn DJ, Potts WK (2002) Discrimination of MHC-derived odors by untrained mice is consistent with divergence in peptide-binding region residues. Proc Natl Acad Sci USA 99: 2187–2192.LS CarrollDJ PennWK Potts2002Discrimination of MHC-derived odors by untrained mice is consistent with divergence in peptide-binding region residues.Proc Natl Acad Sci USA9921872192
- 19. Sommer S (2005) Major histocompatibility complex and mate choice in a monogamous rodent. Behav Ecol Sociobiol 58: 181–189.S. Sommer2005Major histocompatibility complex and mate choice in a monogamous rodent.Behav Ecol Sociobiol58181189
- 20. Landry C, Garant D, Duchesne P, Bernatchez L (2001) “Good genes as heterozygosity”: MHC and mate choice in Atlantic salmon (Salmo salar) Proc R Soc Lond B 268: 1279–1285.C. LandryD. GarantP. DuchesneL. Bernatchez2001“Good genes as heterozygosity”: MHC and mate choice in Atlantic salmon (Salmo salar) Proc R Soc Lond B26812791285
- 21. Milinski M, Griffiths S, Wegner KM, Reusch TBH, Haas-Assenbaum A, et al. (2005) Mate choice decisions of stickleback females predictably modified by MHC peptide ligands. Proc Natl Acad Sci USA 102: 4414–4418.M. MilinskiS. GriffithsKM WegnerTBH ReuschA. Haas-Assenbaum2005Mate choice decisions of stickleback females predictably modified by MHC peptide ligands.Proc Natl Acad Sci USA10244144418
- 22. Kocher TD (2004) Adaptive evolution and explosive speciation: The cichlid fish model. Nat Rev Gen 5: 288–298.TD Kocher2004Adaptive evolution and explosive speciation: The cichlid fish model.Nat Rev Gen5288298
- 23. Genner MJ, Turner GF (2005) The mbuna cichlids of Lake Malawi: a model for rapid speciation and adaptive radiation. Fish and Fisheries 6: 1–34.MJ GennerGF Turner2005The mbuna cichlids of Lake Malawi: a model for rapid speciation and adaptive radiation.Fish and Fisheries6134
- 24. Ribbink AJ, Marsh BA, Marsh AC, Ribbink AC, Sharp BJ (1983) A preliminary survey of the cichlid fishes of rocky habitats in Lake Malawi. S Afr J Zool 18: 147–310.AJ RibbinkBA MarshAC MarshAC RibbinkBJ Sharp1983A preliminary survey of the cichlid fishes of rocky habitats in Lake Malawi.S Afr J Zool18147310
- 25. Turner GF (1999) Explosive speciation of African cichlid fishes. In: Maguran AE, May RM, editors. Evolution of Biological Diversity. Oxford: Oxford University Press. pp. 113–129.GF Turner1999Explosive speciation of African cichlid fishes.AE MaguranRM MayEvolution of Biological DiversityOxfordOxford University Press113129
- 26. Rico C, Turner GF (2002) Mol Ecol 11: 1585–1590.C. RicoGF Turner2002Mol Ecol1115851590
- 27. Fenton A, Fairbairn JP, Norman R, Hudson PJ (2002) Parasite transmission: reconciling theory and reality. J Anim Ecol 71: 893–905.A. FentonJP FairbairnR. NormanPJ Hudson2002Parasite transmission: reconciling theory and reality.J Anim Ecol71893905
- 28. Taylor MI, Turner GF, Robinson RL, Stauffer JR JR (1998) Sexual selection, parasites and bower height skew in a bower-building cichlid fish. Anim Behav 56: 379–384.MI TaylorGF TurnerRL RobinsonJR Stauffer JR1998Sexual selection, parasites and bower height skew in a bower-building cichlid fish.Anim Behav56379384
- 29. Maan ME, van der Spoel M, Jimenez PQ, van Alphen JJM, Seehausen O (2006) Fitness correlates of male coloration in a Lake Victoria cichlid fish. Behav Ecol 17: 691–699.ME MaanM. van der SpoelPQ JimenezJJM van AlphenO. Seehausen2006Fitness correlates of male coloration in a Lake Victoria cichlid fish.Behav Ecol17691699
- 30. Plenderleith M, van Oosterhout C, Robinson RL, Turner GF (2005) Female preference for conspecific males based on olfactory cues in a Lake Malawi cichlid fish. Biol Lett 1: 411–414.M. PlenderleithC. van OosterhoutRL RobinsonGF Turner2005Female preference for conspecific males based on olfactory cues in a Lake Malawi cichlid fish.Biol Lett1411414
- 31. Schluter D (2001) Ecology and the origin of species. Trends Ecol Evol 16: 372–380.D. Schluter2001Ecology and the origin of species.Trends Ecol Evol16372380
- 32. Málaga-Trillo E, Zaleska-Rutczynska Z, McAndrew B, Vincek V, Figueroa F, et al. (1998) Linkage Relationship and Haplotype Polymorphism Among Cichlid Mhc Class II B Loci. Genetics 149: 1527–1537.E. Málaga-TrilloZ. Zaleska-RutczynskaB. McAndrewV. VincekF. Figueroa1998Linkage Relationship and Haplotype Polymorphism Among Cichlid Mhc Class II B Loci.Genetics14915271537
- 33. Ono H, Klein D, Vincek V, Figueroa F, O'hUigin C, et al. (1992) Major histocompatibility complex class II genes of zebrafish. Proc Natl Acad Sci USA 89: 11886–11890.H. OnoD. KleinV. VincekF. FigueroaC. O'hUigin1992Major histocompatibility complex class II genes of zebrafish.Proc Natl Acad Sci USA891188611890
- 34. Emes RD, Beatson SA, Ponting CP, Goodstadt L (2004) Evolution and comparative genomics of odorant- and pheromone-associated genes in rodents. Genome Research 14: 591–602.RD EmesSA BeatsonCP PontingL. Goodstadt2004Evolution and comparative genomics of odorant- and pheromone-associated genes in rodents.Genome Research14591602
- 35. Hughes AL (2000) Evolution of introns and exons of class II major histocompatibility complex genes of vertebrates. Immunog 51: 473–486.AL Hughes2000Evolution of introns and exons of class II major histocompatibility complex genes of vertebrates.Immunog51473486
- 36. Reusch TBH, Langefors Å (2005) Inter- and Intralocus recombination drive MHC class IIB gene diversification in a teleost, the three-spined stickleback Gasterosteus aculeatus J Mol Evol 61: 531–541.TBH ReuschÅ. Langefors2005Inter- and Intralocus recombination drive MHC class IIB gene diversification in a teleost, the three-spined stickleback Gasterosteus aculeatusJ Mol Evol61531541
- 37. Kosakovsky Pond SL, Posada D, Gravenor MB, Woelk CH, Frost SDW (2006) Automated phylogenetic detection of recombination using a genetic algorithm. Mol Biol Evol 23: 1891–1901.SL Kosakovsky PondD. PosadaMB GravenorCH WoelkSDW Frost2006Automated phylogenetic detection of recombination using a genetic algorithm.Mol Biol Evol2318911901
- 38. Legendre P, Gallagher ED (2001) Ecologically meaningful transformations for ordination of species data. Oecologia 129: 271–280.P. LegendreED Gallagher2001Ecologically meaningful transformations for ordination of species data.Oecologia129271280
- 39. Frank SA (1994) Recognition and polymorphism in host-parasite genetics. Philos Trans R Soc Lond B Biol Sci 346: 283–293.SA Frank1994Recognition and polymorphism in host-parasite genetics.Philos Trans R Soc Lond B Biol Sci346283293
- 40. Schluter D (2000) The Ecology of Adaptive Radiation;. In: May RM, Harvey PH, editors. New York: Oxford University Press. D. Schluter2000The Ecology of Adaptive Radiation;RM MayPH HarveyNew YorkOxford University Press
- 41. Schierup MH, Vekemans X, Charlesworth D (2000) The effect of hitch-hiking on genes linked to a balanced polymorphism in a subdivided population. Gen Res 76: 63–73.MH SchierupX. VekemansD. Charlesworth2000The effect of hitch-hiking on genes linked to a balanced polymorphism in a subdivided population.Gen Res766373
- 42. van Oosterhout C, Joyce DA, Cummings SM, Blais J, Barson NJ, et al. (2006) Balancing selection, random genetic drift, and genetic variation at the major histocompatibility complex in two wild populations of guppies (Poecilia reticulata). Evolution 60: 2562–2574.C. van OosterhoutDA JoyceSM CummingsJ. BlaisNJ Barson2006Balancing selection, random genetic drift, and genetic variation at the major histocompatibility complex in two wild populations of guppies (Poecilia reticulata).Evolution6025622574
- 43. Skarstein F, Folstad I, Rønning HP (2005) Spawning colouration, parasites and habitat selection in Salvelinus alpinus: initiating speciation by sexual selection? J Fish Biol 67: 969–980.F. SkarsteinI. FolstadHP Rønning2005Spawning colouration, parasites and habitat selection in Salvelinus alpinus: initiating speciation by sexual selection?J Fish Biol67969980
- 44. Cohen S (2002) Strong positive selection and habitat-specific amino acid substitution patterns in Mhc from an estuarine fish under intense pollution stress. Mol Biol Evol 10: 1870–1880.S. Cohen2002Strong positive selection and habitat-specific amino acid substitution patterns in Mhc from an estuarine fish under intense pollution stress.Mol Biol Evol1018701880
- 45. Kornfield I, Smith PF (2000) African cichlid fishes: Model systems for evolutionary biology. Ann Rev Ecol Syst 31: 163–196.I. KornfieldPF Smith2000African cichlid fishes: Model systems for evolutionary biology.Ann Rev Ecol Syst31163196
- 46. Deutsch JC (1997) Colour diversification in Malawi cichlids: evidence for adaptation, reinforcement or sexual selection? Biol J Linn Soc 62: 1–14.JC Deutsch1997Colour diversification in Malawi cichlids: evidence for adaptation, reinforcement or sexual selection?Biol J Linn Soc62114
- 47. Albertson RC, Markert JA, Danley PD, Kocher TD (1999) Phylogeny of a rapidly evolving clade: The cichlid fishes of Lake Malawi, East Africa. Proc Natl Acad Sci USA 96: 5107–5110.RC AlbertsonJA MarkertPD DanleyTD Kocher1999Phylogeny of a rapidly evolving clade: The cichlid fishes of Lake Malawi, East Africa.Proc Natl Acad Sci USA9651075110
- 48. Terai Y, Seehausen O, Sasaki T, Takahashi K, Mizoiri S, et al. (2006) Divergent selection on opsins drives incipient speciation in Lake Victoria cichlids. PLoS Biology 4: Y. TeraiO. SeehausenT. SasakiK. TakahashiS. Mizoiri2006Divergent selection on opsins drives incipient speciation in Lake Victoria cichlids.PLoS Biology4
- 49. Dominey WJ (1984) Effects of sexual selection and life history on speciation: Species flocks in African cichlids and Hawaiian Drosophila. In: Echelle AA, Kornfield I, editors. Evolution of fish species flocks. Orono, Maine: University of Maine at Orono Press. pp. 231–249.WJ Dominey1984Effects of sexual selection and life history on speciation: Species flocks in African cichlids and Hawaiian Drosophila.AA EchelleI. KornfieldEvolution of fish species flocksOrono, MaineUniversity of Maine at Orono Press231249
- 50. Seehausen O, van Alphen JJM (1999) Can sympatric speciation by disruptive sexual selection explain rapid evolution of cichlid diversity in Lake Victoria? Ecol Lett 2: 262–271.O. SeehausenJJM van Alphen1999Can sympatric speciation by disruptive sexual selection explain rapid evolution of cichlid diversity in Lake Victoria?Ecol Lett2262271
- 51. Allender CJ, Seehausen O, Knight ME, Turner GF, Maclean N (2003) Divergent selection during speciation of Lake Malawi cichlid fishes inferred from parallel radiations in nuptial coloration. Proc Natl Acad Sci USA 100: 14074–14079.CJ AllenderO. SeehausenME KnightGF TurnerN. Maclean2003Divergent selection during speciation of Lake Malawi cichlid fishes inferred from parallel radiations in nuptial coloration.Proc Natl Acad Sci USA1001407414079
- 52. Hamilton WD, Zuk M (1982) Heritable true fitness and bright birds: a role for parasites? Science 218: 384–387.WD HamiltonM. Zuk1982Heritable true fitness and bright birds: a role for parasites?Science218384387
- 53. Hamilton WD, Axelrod R, Tanese R (1990) Sexual reproduction as an adaptation to resist parasites. Proc Natl Acad Sci USA 87: 3566–3573.WD HamiltonR. AxelrodR. Tanese1990Sexual reproduction as an adaptation to resist parasites.Proc Natl Acad Sci USA8735663573
- 54. Von Schantz T, Wittzell H, Goransson G, Grahn M, Persson K (1996) MHC genotype and male ornamentation: Genetic evidence for the Hamilton-Zuk model. Proc R Soc Lond B 263: 265–271.T. Von SchantzH. WittzellG. GoranssonM. GrahnK. Persson1996MHC genotype and male ornamentation: Genetic evidence for the Hamilton-Zuk model.Proc R Soc Lond B263265271
- 55. Westneat DF, Birkhead TR (1998) Alternative hypotheses linking the immune system and mate choice for good genes. Proc R Soc Lond B 265: 1065–1073.DF WestneatTR Birkhead1998Alternative hypotheses linking the immune system and mate choice for good genes.Proc R Soc Lond B26510651073
- 56. Getty T (2002) Signaling health versus parasites. Am Nat 159: 363–371.T. Getty2002Signaling health versus parasites.Am Nat159363371
- 57. Moore SL, Wilson K (2002) Parasites as a viability cost of sexual selection in natural populations of mammals. Science 297: 2015–2018.SL MooreK. Wilson2002Parasites as a viability cost of sexual selection in natural populations of mammals.Science29720152018
- 58. Garamszegi LZ, Moller AP, Erritzoe J (2003) The evolution of immune defense and song complexity in birds. Evolution 57: 905–912.LZ GaramszegiAP MollerJ. Erritzoe2003The evolution of immune defense and song complexity in birds.Evolution57905912
- 59. Blais J, Rico C, Bernatchez L (2004) Nonlinear effects of female mate choice in wild threespine sticklebacks. Evolution 58: 2498–2510.J. BlaisC. RicoL. Bernatchez2004Nonlinear effects of female mate choice in wild threespine sticklebacks.Evolution5824982510
- 60. Reusch TBH, Haberli MA, Aeschlimann PB, Milinski M (2001) Female sticklebacks count alleles in a strategy of sexual selection explaining MHC polymorphism. Nature 414: 300–302.TBH ReuschMA HaberliPB AeschlimannM. Milinski2001Female sticklebacks count alleles in a strategy of sexual selection explaining MHC polymorphism.Nature414300302
- 61. Nowak MA, Tarczy-Hornoch K, Austyn JM (1992) The optimal number of major histocompatibility complex molecules in an individual. Proc Natl Acad Sci USA 89: 10896–10899.MA NowakK. Tarczy-HornochJM Austyn1992The optimal number of major histocompatibility complex molecules in an individual.Proc Natl Acad Sci USA891089610899
- 62. Borghans JAM, Noest AJ, De Boer RJ (2003) Thymic selection does not limit the individual MHC diversity. Eur J Immunol 33: 3353–3358.JAM BorghansAJ NoestRJ De Boer2003Thymic selection does not limit the individual MHC diversity.Eur J Immunol3333533358
- 63. van den Berg HA, Rand DA (2003) Antigen presentation on MHC molecules as a diversity filter that enhances immune efficacy. J Theor Biol 224: 249–267.HA van den BergDA Rand2003Antigen presentation on MHC molecules as a diversity filter that enhances immune efficacy.J Theor Biol224249267
- 64. Felsenstein J (1981) Skepticism towards Santa Rosalia, or why are there so few kinds of animals? Evolution 35: 124–138.J. Felsenstein1981Skepticism towards Santa Rosalia, or why are there so few kinds of animals?Evolution35124138
- 65. Konings A (2001) Malawi cichlids in their natural habitat. El Paso, TX: Cichlid Press. A. Konings2001Malawi cichlids in their natural habitat.El Paso, TXCichlid Press352
- 66. Nielsen R, Wakeley J (2001) Distinguishing migration from isolation: a Markov chain Monte Carlo approach. Genetics 158: 885–896.R. NielsenJ. Wakeley2001Distinguishing migration from isolation: a Markov chain Monte Carlo approach.Genetics158885896
- 67. Hughes AL, Nei M (1988) Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection. Nature 335: 167–170.AL HughesM. Nei1988Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection.Nature335167170
- 68. Yang Z, Swanson WJ (2002) Codon-substitution models to detect adaptive evolution that account for heterogeneous selective pressures among site classes. Mol Biol Evol 19: 49–57.Z. YangWJ Swanson2002Codon-substitution models to detect adaptive evolution that account for heterogeneous selective pressures among site classes.Mol Biol Evol194957
- 69. Ono H, O'hUigin C, Tichy H, Klein D (1993) Major-histocompatibility-complex variation in two species of cichlid fishes from Lake Malawi. Mol Biol Evol 10: 1060–1072.H. OnoC. O'hUiginH. TichyD. Klein1993Major-histocompatibility-complex variation in two species of cichlid fishes from Lake Malawi.Mol Biol Evol1010601072
- 70. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG (1997) The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 24: 4876–4882.JD ThompsonTJ GibsonF. PlewniakF. JeanmouginDG Higgins1997The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools.Nucleic Acids Res2448764882
- 71. Saitou N, Nei M (1987) The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4: 406–425.N. SaitouM. Nei1987The neighbor-joining method: a new method for reconstructing phylogenetic trees.Mol Biol Evol4406425
- 72. Posada D, Crandall KA (2001) Evaluation of methods for detecting recombination from DNA sequences: computer simulations. Proc Natl Acad Sci USA 98: 13757–13762.D. PosadaKA Crandall2001Evaluation of methods for detecting recombination from DNA sequences: computer simulations.Proc Natl Acad Sci USA981375713762
- 73. Parham P, Ohta T (1996) Population biology of antigen presentation by MHC class I molecules. Science 272: 67–74.P. ParhamT. Ohta1996Population biology of antigen presentation by MHC class I molecules.Science2726774
- 74. Aguilar A, Roemer G, Debenham S, Binns M, Garcelon D, et al. (2004) High MHC diversity maintained by balancing selection in an otherwise genetically monomorphic mammal. Proc Natl Acad Sci USA 101: 3490–3494.A. AguilarG. RoemerS. DebenhamM. BinnsD. Garcelon2004High MHC diversity maintained by balancing selection in an otherwise genetically monomorphic mammal.Proc Natl Acad Sci USA10134903494
- 75. Kosakovsky Pond SL, Frost SDW, Grossman Z, Gravenor MB, Richman DD, et al. (2006) Adaptation to Different Human Populations by HIV-1 Revealed by Codon-Based Analyses. Plos Computational Biology 2: e62.SL Kosakovsky PondSDW FrostZ. GrossmanMB GravenorDD Richman2006Adaptation to Different Human Populations by HIV-1 Revealed by Codon-Based Analyses.Plos Computational Biology2e62
- 76. Kosakovsky Pond SL, Muse SV (2005) Site-to-site variation in synonymous substitution rates. Mol Biol Evol 22: 2375–2385.SL Kosakovsky PondSV Muse2005Site-to-site variation in synonymous substitution rates.Mol Biol Evol2223752385
- 77. Sorhannus U, Kosakovsky Pond SL (2006) Evidence for positive selection on a sexual reproduction gene in the diatom genus Thalassiosira (Bacillariophyta). J Mol Evol 63: 231–239.U. SorhannusSL Kosakovsky Pond2006Evidence for positive selection on a sexual reproduction gene in the diatom genus Thalassiosira (Bacillariophyta).J Mol Evol63231239
- 78. Kosakovsky Pond SL, Frost SDW (2005) Not so different after all: A comparison of methods for detecting amino acid sites under selection. Mol Biol Evol 22: 1208–1222.SL Kosakovsky PondSDW Frost2005Not so different after all: A comparison of methods for detecting amino acid sites under selection.Mol Biol Evol2212081222
- 79. Wang TH, Donaldson YK, Brettle RP, Bell JE, Simmonds P (2001) Identification of shared populations of human immunodeficiency virus type 1 infecting microglia and tissue macrophages outside the central nervous system. J Virol 75: 11686–11699.TH WangYK DonaldsonRP BrettleJE BellP. Simmonds2001Identification of shared populations of human immunodeficiency virus type 1 infecting microglia and tissue macrophages outside the central nervous system.J Virol751168611699
- 80. Hudson R, Slatkin M, Maddison W (1992) Estimation of levels of gene flow from DNA sequence data. Genetics 132: 583–589.R. HudsonM. SlatkinW. Maddison1992Estimation of levels of gene flow from DNA sequence data.Genetics132583589
- 81. Yampolsky LY, Stoltzfus A (2005) The Exchangeability of Amino Acids in Proteins Genetics 170: 1459–1472.LY YampolskyA. Stoltzfus2005The Exchangeability of Amino Acids in Proteins Genetics17014591472
- 82. Nei M (1973) Analysis of gene diversity in subdivided populations. Proc Natl Acad Sci USA 70: 3321–3323.M. Nei1973Analysis of gene diversity in subdivided populations.Proc Natl Acad Sci USA7033213323
- 83. Hedrick PW (1999) Highly variable loci and their interpretation in evolution and conservation. Evolution 53: 313–318.PW Hedrick1999Highly variable loci and their interpretation in evolution and conservation.Evolution53313318
- 84. Hudson RR (2000) A new statistic for detecting genetic differentiation. Genetics 155: 2011–2014.RR Hudson2000A new statistic for detecting genetic differentiation.Genetics15520112014