The genomics of adaptation to climate in European great tit (Parus major) populations

Abstract The recognition that climate change is occurring at an unprecedented rate means that there is increased urgency in understanding how organisms can adapt to a changing environment. Wild great tit (Parus major) populations represent an attractive ecological model system to understand the genomics of climate adaptation. They are widely distributed across Eurasia and they have been documented to respond to climate change. We performed a Bayesian genome-environment analysis, by combining local climate data with single nucleotide polymorphisms genotype data from 20 European populations (broadly spanning the species’ continental range). We found 36 genes putatively linked to adaptation to climate. Following an enrichment analysis of biological process Gene Ontology (GO) terms, we identified over-represented terms and pathways among the candidate genes. Because many different genes and GO terms are associated with climate variables, it seems likely that climate adaptation is polygenic and genetically complex. Our findings also suggest that geographical climate adaptation has been occurring since great tits left their Southern European refugia at the end of the last ice age. Finally, we show that substantial climate-associated genetic variation remains, which will be essential for adaptation to future changes.


Figure S1 :
Figure S1: Scree plot summarising the principal components analysis of the 22 climate variables.

Figure S2 :
Figure S2: For the first four PCs: A) The contributions of climate variables to PCs (expressed as a percentage) are highlighted.B) The quality of representation of the climate variable on the PCA factor map plot is indicated by high cos2 values.The sum of cos2 values across all PCs of each variable equals one.

Figure S3 :
Figure S3: For the first four PCs: A) The contributions of populations to PCs (expressed as a percentage) are highlighted.B) The quality of representation of the population on the PCA factor map plot is indicated by high cos2 values.The sum of cos2 values across all PCs of each population equals one.

Figure S4 :
Figure S4: Scatterplot of the first two PCs of population climatic covariables

Figure S8 :
Figure S8: Manhattan plot of annotated genes associated with climate adaptation for PC4.The threshold of the decisive level of evidence (BFmc>20) is indicated along the plot, with annotated genes displayed for BFmc>20.Details as for Figure S7.

Figure S9 :
Figure S9: Allele frequencies in each population at loci associated with PC1 of climate variation (panels A-E).The title of each panel indicates the chromosome, position (bp), alleles and closest gene to each climate-associated SNP.Gene CLTRN is also termed TMEM27 in the chicken gene annotation.

Figure S10 :
Figure S10: Allele frequencies in each population at loci associated with PC2 of climate variation (panels A-O).The title of each panel indicates the chromosome, position (bp), alleles and closest gene (annotated to Parus major genes) to each climate-associated SNP.

Table S1 :
Sample sizes and locations (decimal grid reference to 2 d.p.) of the 20 study populations.

Table S2 :
Bioclimate variables extracted from WorldClimv2 at 30sec spatial resolution, (approx.1km 2 surface area).This data set has average monthly climate data for min, mean and max temp and precipitation, and uses data from 1970-2000.

Table S3 :
The number of variants associated with climate adaptation at the decisive, and strong levels and above, for raw and annotated data for PCs 1-4.The number of 'decisive' climate-associated variants also directly identified as 'outlier' variants are indicated in brackets.

Table S4 :
Annotated variants associated with climate adaptation for PC1-PC4, listed in order of genome location (chromosome and position in base pairs).Information on the identified gene ID and gene symbol are displayed, with the averaged BFmc score over the 3 runs.Variants co-identified as outliers under selection by XtX score or in a gene with a variant identified under selection are indicated by * or / respectively.

Table S5 :
Biological process GO terms that were significantly enriched among loci associated with PC1.The highest level category of each term is listed.For each term the number of annotated genes considered in the dataset (Ann), the number of annotated genes significantly associated with PC1 (Sign) and the number of expected genes to be associated with PC1 (Exp) are indicated.Statistical significance was determined using a Kolmogorov-Smirnov (KS) test statistic and topGO's default algorithm.

Table S6 :
Biological process GO terms that were significantly enriched among loci associated with PC2.Statistical significance was determined using a Kolmogorov-Smirnov (KS) test statistic and topGO's default algorithm (details as for TableS5).

Table S7 :
Candidate climate adaptation genes that were identified in significantly enriched GO term pathways associated with each of PC1-4 are presented.The "XtX Outlier" column also indicates if variants were also significant in the core model (*), or if other variants in the same gene were also significant in the core model (/).

Table S8 :
Biological process GO terms that were significantly over-represented among candidate adaptation loci for multiple climate PCs.

Table S9 :
Biological process GO terms that were significantly enriched among loci associated with PC3.Statistical significance was determined using a Kolmogorov-Smirnov (KS) test statistic and topGO's default algorithm (details as for TableS5).

Table S9 :
Biological process GO terms that were significantly enriched among loci associated with PC3.Statistical significance was determined using a Kolmogorov-Smirnov (KS) test statistic and topGO's default algorithm (details as for TableS5).(continued)

Table S10 :
Biological process GO terms that were significantly enriched among loci associated with PC4.Statistical significance was determined using a Kolmogorov-Smirnov (KS) test statistic and topGO's default algorithm (details as for TableS5).

Table S17 :
Cellular component GO term enrichment (weight01 algorithm and KS statistic) for PC3.

Table S20 :
Climate adaptation candidate genes from other studies that are closely related to candidate genes found here.