Long‐Read Reference Transcriptomes for Five Species of Rosaceae Tribe Spiraeeae: A Resource for Investigating Sequence Divergence, Species‐Associated Expression Variation, and Leaf‐Developmental Candidate Genes

ABSTRACT

Species in the Rosaceae tribe Spiraeeae occupy diverse habitats and vary strikingly in growth form and leaf morphology. Still, no reference transcriptomes have been published for the tribe, and its transcriptome‐level variation is uncharacterized. We generated long‐read (PacBio Iso‐Seq) reference transcriptomes of five species spanning three genera: Petrophytum caespitosum (Nutt.) Rydb., Spiraea lucida Douglas ex Greene, Spiraea splendens Baumann ex K. Koch, Aruncus dioicus (Walter) Fernald var. dioicus , and Aruncus aethusifolius (H.Lév.) Nakai. We then quantified expression across 20 individuals using 3′‐tag RNA sequencing (TagSeq) to characterize expression patterns, ortholog sequence divergence, and gene repertoire. Multivariate analyses (principal component analysis [PCA] and hierarchical clustering) resolved replicates into three groups corresponding to the sampled genera. Because the material combined field‐collected and potted plants, we treat expression comparisons as descriptive and do not attribute them to genetic control of interspecific differences. Pairwise analyses revealed extensive transcriptional divergence between genera, including thousands of differentially expressed transcripts in comparisons such as A. aethusifolius vs. P. caespitosum . Functional enrichment highlighted variation in transport and catabolism, plastid organization, and biotic‐stress responses, implicating both metabolic and organelle‐associated processes. Transcription factor families, particularly bHLH and bZIP, varied most among differentially expressed transcripts. Weighted gene coexpression network analysis (WGCNA) recovered modules correlated with species identity; because between‐species variation dominates this dataset, we treat these as descriptive summaries of coexpression structure and do not regard them as independent evidence of species‐specific regulatory programs. Because reads were quantified against a reference assembled from species‐specific contigs, mapping rates were higher within species (70.5%–85.1%) than across species (25.1%–77.0%), and between‐species expression comparisons are interpreted with this asymmetry in mind. Sequence‐level analysis of orthologous transcript pairs indicated that protein‐coding genes are predominantly under purifying selection. A small fraction (2.7%–5.0% of retained pairs) showed Ka/Ks > 1, but these were concentrated at low synonymous divergence, where ratio estimates are unstable; we therefore treat them as screening‐level candidates for rapid divergence, enriched for stress‐response, membrane‐trafficking, and ion‐transport annotations, and we make no claim of positive selection. These results provide a comparative transcriptomic framework for Spiraeeae and identify candidate transcripts and protein‐coding differences for future studies of ecological and morphological variation in Rosaceae.