Globally, farmers cultivate and maintain crop landraces (i.e., traditional varieties). Landraces contain unique diversity shaped in part by natural and human-mediated selection and are an indispensable resource to farmers. Since environmental conditions change with elevation, crop landraces grown along elevational gradients have provided ideal locations to explore patterns of local adaptation. To further probe traits underlying this differentiation, transcriptome signatures can help provide a foundation for understanding the ways in which functional genetic diversity may be shaped by environment. In this study, we returned to an elevational gradient in Chiapas, Mexico, to assess transcriptional differentiation of genes underlying UV-B protection in locally adapted maize landraces from multiple elevations. We collected and planted landraces from three elevational zones (lowland, approximately 600 m; midland, approximately 1,550 m; highland approximately 2,100 m) in a common garden at 1,531 m. Using RNA-seq data derived from leaf tissue, we performed differential expression analysis between maize from these distinct elevations. Highland and lowland landraces displayed differential expression in phenylpropanoid and flavonoid biosynthesis genes involved in the production of UV-B protectants and did so at a rate greater than expected based on observed background transcriptional differentiation across the genome. These findings provide evidence for the differentiation of suites of genes involved in complex ecologically relevant pathways. Thus, while neutral evolutionary processes may have played a role in the observed patterns of differentiation, UV-B may have also acted as a selective pressure to differentiate maize landraces in the region. Studies of the distribution of functional crop genetic diversity across variable landscapes can aid us in understanding the response of diversity to abiotic/biotic change and, ultimately, may facilitate its conservation and utilization.

The uploaded data is raw reads from the sequencer, no processing has been performed. Following is a brief description of how the data was collected. See manuscript for further information. 

As described in Kost et al. (2017), we performed RNA extractions with Qiagen RNeasy Plant Mini Kits and confirmed RNA integrity with the Agilent 2100 Bioanalyzer. In total, we performed 135 RNA extractions (15 landraces x 9 individuals per landrace). To aid in RNA-seq library construction, we assessed RNA concentrations using Qubit® 2.0 Fluorometer combined with the Qubit RNA Assay Kit. Before library construction, we pooled RNA from the three individuals sampled per landrace per block, resulting in 45 pooled RNA samples – one pooled sample per block x 15 landraces x three blocks. We performed RNA-seq library construction using the strand specific library preparation method described in Zhong et al., (2011). Since we employed a multiplex sequence strategy, we assigned each library a barcode during library preparation. We randomized pooled RNA samples before library preparation to account for batch effects. Following library construction, we analyzed cDNA libraries using the Qubit DNA Assay Kit to determine concentration and we used the Agilent 2100 Bioanalyzer to determine library size.

We sequenced the 45 RNA-seq libraries in four flow cell lanes of the Illumina HiSeq 2500 at the Genomics Resources Core Facility located at the Weill Cornell Medical College. We performed paired end sequencing at 50 bp. Since HiSeq 2500 normally produce ~ 120-130 million paired-end reads per lane, we multiplexed 12 libraries per lane to produce ~ 10-11 million reads per library. In order to ensure balanced sequencing lanes, we ran 12 libraries in each lane. Since we only had 45 libraries, we sequenced one library in all four lanes, bringing the number of libraries in each lane to 12.

Following the materials and Methods section of this manuscript will allow you to recreate the entire analysis presented in the manuscript.