Divergent selection on opsins drives incipient speciation in Lake Victoria cichlids

Abstract

Divergent natural selection acting on ecological traits, which also affect mate choice, is a key element of ecological speciation theory, but has not previously been demonstrated at the molecular gene level to our knowledge. Here we demonstrate parallel evolution in two cichlid genera under strong divergent selection in a gene that affects both. Strong divergent natural selection fixed opsin proteins with different predicted light absorbance properties at opposite ends of an environmental gradient. By expressing them and measuring absorbance, we show that the reciprocal fixation adapts populations to divergent light environments. The divergent evolution of the visual system coincides with divergence in male breeding coloration, consistent with incipient ecological by-product speciation.

Figure 1. Maps and Frequencies of LWS Alleles

(A) The study area in southern Lake Victoria. Arabic numerals indicate stations at which cichlids were collected. The Secchi disk water transparency (cm) at each station is shown in parentheses. The stations are: 1, Buyago Rocks; 2, Marumbi Island; 3, Matumbi Island; 4, Luanso Island; 5, Python (Nyamatala) Islands; 6, Kissenda Island; 7, Hippo Island; 8, Juma Island; 9, Bwiru Point; 10, Makobe Island; 11, Igombe Island; 12, Ruti Island; 13, Senga Point; 14, Sozihe Islands; 15, Namatembi Island; 16, Nyamatala (Nyameruguyu) Island; 17, Gabalema Islands; 18, north end of Luanso Bay. Where the local name differs from that used in previous publications, the local name is indicated in parenthesis. (B) A representation of the microhabitat distribution of the studied species by water depth and between, as opposed to outside, the rocky boulders. Photos show males of the blue morph in nuptial coloration. “Affected” and “Non-affected” indicate the species with the LWS allele frequencies that are strongly affected and not strongly affected by variation in water transparency, respectively. (C) LWS allele group frequencies in the populations of N. greenwoodi/omnicaeruleus. (D) LWS allele group frequencies in the populations of M. mbipi. (E) LWS allele group frequencies in the populations of N. rufocaudalis. (F) LWS allele group frequencies in the populations of P. pundamilia. In (B–E), Arabic numerals correspond to those in (A). The size of a pie indicates the number of haplotypes sequenced: N. greenwoodi: n = 58 at station 2; n = 10 at 3; n = 14 at 5; n = 10 at 7; n = 14 at 9; n = 22 at 11; n = 8 at 13; n = 8 at 14; n = 50 at 15, and N. omnicaeruleus, n = 40 at 10. M. mbipi: n = 36 at station 5; n = 32 at 7; n = 30 at 10; n = 10 at 11; n = 2 at 13. N. rufocaudalis: n = 42 at station 5; n = 16 at 6; n = 10 at 10; n = 10 at 16; n = 6 at 17; n = 4 at 18. P. pundamilia: n = 2 at station 1; n = 10 at 2; n = 4 at 3; n = 8 at 4; n = 18 at 5; n = 8 at 6; n = 6 at 8; n = 6 at 9; n = 10 at 10; n = 6 at 11; n = 8 at 12. The color of the sections of the pie indicates the frequency of allele groups L (red), M1 (yellow), M2 (green), H (blue), M3 (orange), P (blue-green), and other alleles (black). The amino acid differences among allele groups are shown for every species in the corresponding white panels.

Figure 2. Detection of Selection Pressure on the LWS Gene

The genome structure of the LWS gene and its flanking region and (A) sliding-window analysis of F_ST between Marumbi (Mr., station 2: N. greenwoodi), Namatembi (Nm., station 15: N. greenwoodi), and Makobe (Ma., station 10: N. omnicaeruleus) populations (Marumbi versus Makobe, Marumbi versus Namatembi, and Makobe versus Namatembi indicated by black, blue, and red lines, respectively). (B) Sliding-window analysis of silent polymorphism (π_s) in Marumbi (Mr: light blue), Namatembi (Nm: red), and Makobe (Ma: yellow), and silent divergence (k_s) between Marumbi, Namatembi, and Makobe populations (Marumbi versus Namatembi, Marumbi versus Makobe, and Makobe versus Namatembi indicated by blue, black, and gray lines, respectively). π _s and k _s were calculated for segments of 700 bp in 25-bp intervals. The solid lines under the genome structure of LWS indicate the three regions used in HKA tests (Table S2). The LWS gene region is defined as the sequence between initiation codon and stop codon (2205 bp) of LWS. The up- and downstream regions are defined as the 5′ and 3′ ends of sequences of the same length as the LWS gene region.

Figure 3. Absorption Spectra of the LWS Pigments Evaluated by the Dark–Light Difference Spectra

The LWS pigments were reconstituted from (A) H allele with A1 retinal, (B) L allele with A1 retinal, (C) H allele with A2 retinal, and (D) L allele with A2 retinal. The λ_max values are indicated with their standard errors.

Figure 4. The Relationship between the Frequency of LWS Alleles, Male Nuptial Color Morphs, and Water Transparency

(A) The relationship between the frequency of long wavelength–sensitive allele group L at the LWS locus (black), yellow-red male nuptial color (red), and water transparency (Secchi disk [cm]) in N. greenwoodi. (B) The relationship between the frequency of allele group M3 at the LWS locus (black), yellow male nuptial color (red), and water transparency in M. mbipi.