One of the biggest challenges with genomic offset approaches is the difficulty in validating the relationships between genomic offset and fitness. We investigate the relationship between genomic offset and current fitness in the yellow warbler, using telomere length as a proxy for fitness. The yellow warbler is a migratory songbird that breeds in various habitats throughout North America and is an excellent system for this study because of the seminal work of Bay et al. (2018) that identified patterns of genomic offset across the species range. We measure telomere length in yellow warbler populations occupying regions with high and low genomic offset across the species range. We predict that if yellow warbler populations are locally adapted to climate on the breeding grounds and future offset is indicative of recent climate change impacts, a significant negative relationship will exist between future predicted genomic offset and current telomere length. Additionally, we will investigate the specific aspects of climate contributing to fitness loss and population declines in the yellow warbler system. We focus on precipitation because previous research indicates precipitation changes are linked to genomic and morphological signals of local adaptation in this species (Bay et al. 2018, Bay et al. 2021). We predict that the correlation between past population decline and future genomic offset previously identified is due to a correlation between past and future precipitation change.

We used restriction-site associated DNA sequencing (RAD-Seq) data from Bay et al. (2018) on 229 individuals from 39 locations across the yellow warbler breeding range. To estimate genomic offset, we then ran gradient forest (Fitzpatrick et al. 2021) on a subset of 1694 unlinked candidate single-nucleotide polymorphisms (SNPs) that were significantly associated with BIOCLIM variables based on LFMM analysis from the previous paper. We built a gradient forest model with average monthly precipitation values for the months of May through July as environmental response variables and the candidate SNPs as predictors. Precipitation data were obtained from the CRU-TS 4.06 dataset (Harris et al., 2020), downscaled with WorldClim 2.1 (Fick and Hijmans, 2017). We then used the predict function within gradient forest to weight the environmental response variables for both current and future predicted climates at 10,000 random locations across the yellow warbler breeding range. We then interpolated across the rest of the breeding range to form a continuous map of genomic offset.

We selected sample sites for telomere measurements by using the continuous map of genomic offset and choosing areas across a gradient of genomic offset in addition to precipitation and elevation. Estimates of future genomic offset were then calculated for each specific sample site. We collected blood samples for telomere measurements from each sample location spanning the breeding range of the yellow warbler during the 2020 and 2021 breeding seasons. Birds were captured via mist-nets, and once in hand, individuals were banded, morphological measurements were taken, and age and sex were recorded. Between 10 µl and 30 µl of blood was collected using brachial venipuncture and stored on ice until reaching the lab, where they were stored in -20°C until analysis (Criscuolo et al. 2009). Birds that could not be sexed in the field were sexed using PCR with the primer set CHD1F and CHD1R (Cakmak et. al 2017). Our sampling resulted in blood samples from 451 yellow warblers spanning 39 sample sites across the breeding range of the species.

DNA was isolated from whole blood samples using the DNeasy Blood and Tissue kit (Qiagen, Valencia, California) following the manufacturer’s protocol. DNA purity and concentration were assessed using a NanoDrop 8000 spectrophotometer (Thermo Scientific), and DNA integrity was assessed on an agarose gel (Eastwood et al. 2018). Following the protocol of Criscuolo et al. (2009), qPCR was used to measure telomere length relative to the glyceraldehyde-3-phosphate dehydrogenase control gene. Samples were run in triplicate, with high efficiencies, intra-assay, and inter-assay repeatability (Eastwood et al. 2018) (SI Appendix, section 1). The distribution of relative telomere length was right-skewed, so log relative telomere length was used in subsequent analyses. Log relative telomere length was then Z-transformed to facilitate comparison to future studies (Verhulst 2020).

To test the assumption that past climate is correlated to future climate, we analyzed the correlation between historic and future changes in climate for the top three climate variables found to be associated with yellow warbler genomic variation according to Bay et al. (2018). These climate variables were bio18 (mean monthly precipitation amount of the warmest quarter), bio15 (precipitation seasonality), and bio13 (precipitation amount of the wettest month). Bioclimatic data for each sample site were extracted from the WorldClim database (Eyring et al. 2015, Fick and Hijmans 2017) using historical and future climate layers. We used a Shared Socio-economic Pathway (SSP) of 585 and used an ensemble model consisting of averages from 24 global climate models to calculate our future climate dataset. Historical changes in climate were calculated by subtracting historic bioclimatic variables (1970-2000) from those bioclimatic variables calculated for 2021-2040. Similarly, future changes in climate were calculated by subtracting bioclimatic variables from 2021-2040 from future bioclimatic variables (2041-2060).