Genome-wide diversity and habitat underlie fine-scale phenotypic differentiation in the rainbow darter (Etheostoma caeruleum)

Abstract

Adaptation to environmental change requires that populations harbor the necessary genetic variation to respond to selection. However, dispersal-limited species with fragmented populations and reduced genetic diversity may lack this variation and are at an increased risk of local extinction. In freshwater fish species, environmental change in the form of increased stream temperatures places many cold-water species at-risk. We present a study of rainbow darters (Etheostoma caeruleum) in which we evaluated the importance of genetic variation on adaptive potential and determined responses to extreme thermal stress. We compared fine-scale patterns of morphological and thermal tolerance differentiation across eight sites, including a unique lake habitat. We also inferred contemporary population structure using genomic data and characterized the relationship between individual genetic diversity and stress tolerance. We found site-specific variation in thermal tolerance that generally matched local conditions and morphological differences associated with lake-stream divergence. We detected patterns of population structure on a highly local spatial scale that could not be explained by isolation by distance or stream connectivity. Finally, we showed that individual thermal tolerance was positively correlated with genetic variation, suggesting that sites with increased genetic diversity may be better at tolerating novel stress. Our results highlight the importance of considering intraspecific variation in understanding population vulnerability and stress response.

Keywords: Etheostomacaeruleum; RADseq; adaptive potential; genetic diversity; morphology; thermal tolerance.

Figure 1

(a) Map showing eight sample sites across three tributaries (Gull, Augusta, and Wabascon Creek) of the Kalamazoo River. (b) Population structure results from the discriminant analysis of principal components (DAPC), with discriminant functions one and two represented on the x and y axes, respectively. Each point corresponds to an individual and colors reflect each sampling site. The top right inset displays eigenvalues for each discriminant function and the bottom right inset indicates the number of principal components retained in the analysis (43)

Figure 2

Fifteen landmarks used for the geometric morphometric analysis: (1) Anterior point on the head, (2) center of the eye, (3) landmark directly above the eye, (4) point at the supraoccipital notch, anterior junction of the (5) first and (6) second dorsal fins with the dorsal midline, (7) dorsal junction of caudal tail, (8) junction of lateral line and caudal tail, (9) ventral junction of the caudal tail, (10) anterior junction of the anal fin with the ventral midline, (11) anterior junction of the pelvic fin with the body, (12) dorsal junction of the pectoral fin with the body, (13–15) landmarks along the lateral line accounting for organismal bending. (a) Landmarks placed on an actual rainbow darter. (b) Wireframe graphic of landmarks produced in MorphoJ. Landmarks used to correct for bending (13–15) are not included

Figure 3

Scores produced from discriminant function analysis (DFA) conducted between the two different habitat types (lake versus stream). Stream individuals are represented by the images in the top left, while lake individuals are represented in the bottom left. Rainbow darter images are individuals from each extreme of the discriminant function, with the wireframe graphics magnified to highlight the direction of morphological differentiation

Figure 4

Population structure inferred from ADMIXTURE for (a) each individual and (b) the population mean for the two genetic clusters (k = 2). An individual is represented by a single vertical column, with the green dotted lines demarking sites

Figure 5

(a) Correlation between site water temperature and critical thermal maximum (CT_max). Error bars indicate 95% confidence intervals with dashed error bars representing sites in which resampling was conducted in 2018. (b) Correlation between site observed heterozygosity from stacks and CT_max. Error bars indicate 95% confidence intervals