Genomics of Experimental Adaptive Radiation in the Cryptic Coloration of Feather Lice

Abstract

A major challenge faced by living organisms is adaptation to novel environments. This process is poorly understood because monitoring genetic changes in natural populations is difficult. One way to simplify the task is to focus on organisms that can be studied in captivity under conditions that remain largely natural. Feather lice (Insecta, Phthiraptera, Ischnocera) are host-specific parasites of birds that live, feed, and breed solely on feathers. Birds defend themselves against lice, which damage feathers, by killing them with their beaks during bouts of preening. In response, feather lice have evolved background-matching cryptic coloration to help them avoid preening. We experimentally manipulated the color backgrounds of host-specific pigeon lice (Columbicola columbae) by confining them to different colored breeds of rock pigeon (Columba livia) over a period of 4 years (ca. 60 louse generations). Over the course of the experiment, we sampled lice from pigeons every 6 months for genomic resequencing and then calculated allele frequency differences and trajectories to identify putative genomic sites under selection. We documented many loci that changed in response to selection for color. Most loci putatively under selection were unshared among replicate populations of lice, indicating that independent adaptation of distinct lineages to the same novel environment resulted in similar phenotypes driven by different genotypes.

Keywords: adaptive radiation; allele frequency trajectory; body color; experimental evolution; genomics; pigeon louse.

Fig. 1.

Examples of cryptic coloration among feather lice in the genus columbicola. The Australian pied imperialpigeon, Ducula bicolor (a), is parasitized by C. wolffhuegeli (b); the rock pigeon, Columba livia (c), is parasitized by C. columbae (d); and the chestnut-quilled rock pigeon, Petrophassa rufipennis (e), is parasitized by C. masoni (f). Photos by: (a) JJ Harrison, Wikimedia Commons, CC BY-SA 4.0; (b) SEB; (c) pxfuel.com/en/free-photo-qffys; (d and f) SMV and J. Altuna; (e) Nimzee, https://inaturalist.ala.org.au/photos/123753242.

Fig. 2.

Experimental evolution of lice on hosts of varying color shows adaptation toward the host's color. a) Feather lice (Columbicola columbae) were transferred to 3 different rock pigeon color morphs: white (left), black (right), and gray (wild-type controls, middle). b) Half of the pigeons could preen normally, while the other half were fitted with poultry “bits” to impair their preening ability. (C–D) Evolution of feather lice on different colored rock pigeons over a 4-year period (ca. 60 louse generations). The y-axis shows changes in the mean (±SE) luminosity (brightness) of lice on white and black rock pigeons, relative to lice on gray rock pigeon controls (set to zero). Different lower-case letters indicate statistically significant differences. c) On birds with normal preening, the relative luminosity of lice on white pigeons increased rapidly (LMM, P < 0.0001); the relative luminosity of lice on black pigeons decreased, but more slowly (LMM, P = 0.001). d) Relative luminosity did not change significantly over time on white or black pigeons with impaired preening (LMM, P ≥ 0.34 in both cases). (Panels B–D redrawn from Bush et al. 2019).

Fig. 3.

Population structure of experimental populations. Analysis with popvae shows that lice within a population—i.e. lice on birds inhabiting the same aviary (same replicate, host color, and preening regime)—are generally more genetically similar to one another, and that populations in the same replicate (Table 1) are more closely related to one another, than they are to lice from the same treatment in other replicates. The lice here were all drawn from populations at the 36-month time point and individually sequenced.

Fig. 4.

GWAS analyses identified different genomic regions under putative selection in different treatments and replicates. We used pFST at month 36 of the experiment to identify regions of the genome under selection. (a–b) We contrasted an experimental population—lice on a) white pigeons or b) black pigeons—with lice from gray controls. The top 0.1% of sites (upper dotted line) are shown in red and are above the higher dotted line. FDR-corrected −log(p) values are plotted, with the FDR-corrected significance threshold of P = 0.05 near the bottom of each plot (lower dotted line). We calculated pFST in each of the 4 replicates individually (R1–R4), and in a combination of all replicates (all pFST). In addition, we used the Cochran-Mantel-Haenszel (CMH) test to identify sites with replicated allele frequency change across replicate populations of the same treatment. Purple bars represent regions where the normally-preening louse population is above the 0.01% statistical threshold, while the impaired-preening louse population on the same color birds is not above 0.01%.

Fig. 5.

Most selected sites are private to each treatment and replicate. This plot shows the average number of putative selected sites that are private (x-axis = 1) versus shared between replicates (x-axis ≥ 2). While a few putative selected sites were found in 2 replicates, none were shared by 3 or 4 replicates.

Fig. 6.

Selected alleles increased in frequency over time. We identified the most significant SNPs by pFST in each of our putatively selected loci (65 loci for white, 26 for black), then plotted allele frequencies over time. a) Allele frequency change in lice from populations exposed to preening-mediated selection (normally preening birds). The heavier red line is the average of all plotted allele frequencies. b) Allele frequency change at the same sites, but in lice not exposed to preening-mediated selection (impaired preening birds); these populations showed no change in allele frequencies over time. c) Allele frequency change at randomly chosen sites in louse populations on normally preening birds; these sites have more rare minor alleles than A or B, in keeping with a neutral site frequency spectrum outside of selected regions. d) Slopes of lines of best fit for the same trajectories in A–C, with the midline representing the median, the box representing the interquartile range, and the whiskers representing 1.5 times the interquartile range. Different lower-case letters indicate statistically significant differences (P < 0.01, Tukey's HSD test).

Fig. 7.

Populations of lice experiencing selection by preening have higher rates of allelic fixation (solid lines) than populations of lice on impaired preening birds (dotted lines). Lines depict cumulative fixations in the polymorphic loci shown in Fig. 6 (65 loci for lice on white pigeons, 26 for lice on black pigeons). An allele was assumed to be fixed if its frequency reached 100% and remained at 100% for the remainder of the experiment. Fixations at the putatively selected sites in unselected populations of lice on impaired preening birds were exceedingly rare.