Cell suspension cultures of Douglas-fir (Pseudotsuga menziesii) nucellus were created to overcome phenological limitations to the in situ study of bioactive proteins within and secreted by nucellar cells. Proteins isolated from the cell suspension culture medium were analyzed using mass spectrometry, as were Douglas-fir post-pollination prefertilization drops and nucellus tissue excised from the ovule at the time of drop production. A total of 280 proteins ranging in size from 5 to 224 kDa were found in the three sample types. Identical proteins were found in the post-pollination prefertilization drop and in the nucellar suspension culture medium or the nucellus. Secretory proteins isolated from the culture medium were either carbohydrate metabolism-related or defense proteins, consistent with previous proteomic studies of pollination drops. Intracellular proteins were found in all three sample types and were consistent with previous proteomic studies of pollination drops, implying similar metabolic processes are at work in the gymnosperm nucellus at the time of drop production. Cell suspension cultures of Douglas-fir nucellus are able to produce secretory proteins that are biologically relevant during pollination. This is potentially a tractable system that can overcome barriers to in situ studies presented by gymnosperm reproductive phenology.

Abstract Fleshy fruits attract animals to ingest fruit, swallow the seeds, and release them in the landscape, thus facilitating seed dispersal and plant regeneration. Attraction of animal dispersers is achieved via attractants such as color or scent, and rewards like sugars, lipids, and micronutrients. In addition to these attractants and rewards, a plethora of fruit and seed functional traits, including size, hardness, and chemistry, can affect the accessibility and attractiveness of fruits to seed‐dispersing animals. These functional traits form the interface between animals and fruits and, through trait matching, act as filters in dispersal networks. Yet, despite their critical roles in shaping seed dispersal networks, many fruit traits are not quantified in a standardized fashion, if they are quantified at all. As such, the existing databases of fruit traits used for syntheses lack many functional dimensions. We review the latest developments in fruit trait analysis, focusing on traits relevant for plant–frugivore interactions (i.e., morphology, color, chemistry), as well as some of their drivers (DNA, RNA, and microbial communities). We provide an overview of many existing methods, their advantages and disadvantages, and their application in field conditions.

PremiseAnalyzing structural changes along the length of an organ provides insight into its development. However, traditional histological methods are limited by intensive procedures and size restrictions. Micro‐computed tomography (microCT) enables non‐destructive internal imaging along the length of an organ, but high cost, technical complexity, and limited accessibility hinder widespread application. Here, we describe serial section videography (SSV), a new low‐cost technique for generating three‐dimensional (3D) reconstructions of internal plant anatomy using serial sectioning and open‐source software.Methods and ResultsSSV was applied to four fern rhizomes with varied gross morphology and diverse vascular architectures. Specimens were sectioned using a sliding microtome or a handheld blade, and imaged using either a digital camera or smartphone setup. Images were aligned using Fiji and segmented using 3D Slicer. The SSV method enabled continuous visualization of internal stem anatomy over several centimeters and is adaptable to both laboratory and field settings.ConclusionsThis protocol offers an alternative to microCT for generating 3D anatomical reconstructions, enabling researchers to examine development and structural variation across organs with minimal equipment and software. This accessible protocol reduces technical and financial barriers and is particularly well‐suited for comparative studies of vascular tissues, advancing the study of plant anatomy and development.

PremiseAstragalus (Fabaceae) is the largest genus of flowering plants, with about 3100 species. Most phylogenies of the genus are based on a few nuclear or plastid loci (with one exception that uses ~100 loci) and usually provide poorly resolved trees and even conflicting subgeneric classifications. Target enrichment can greatly improve phylogenetic resolution, even at recently diverged taxonomic units, by generating sequences for hundreds of phylogenetically informative, putative single‐copy loci. However, a specific bait set for the Astragalean clade is still lacking.MethodsIn this study, we use transcriptome data from 17 species in the Astragalean clade (of which eight are Astragalus) to identify putative single‐copy nuclear loci to build a bait set for target enrichment. This resulted in 819 exons from 686 genes across the Astragalean clade. The bait set was validated with a phylogenetic study based on 20 taxa selected across major clades in Astragalus plus three Astragalean species.ResultsWe recovered, on average, 739.3 loci covering at least 75% of the corresponding target reference length. The average number of paralog warnings was 76.2, ranging from 12 to 373. Our phylogenetic inference produced full support for all nodes, largely in agreement with the current subgeneric classifications of Astragalus.DiscussionOur bait set, Astragalean819, effectively retrieved highly informative loci to build a robust phylogeny that will help to clarify the complex relationships among members of the Astragalean clade and the subgeneric relationships in Astragalus. This study provides a foundation for advancing the understanding of the systematics and evolution of the largest flowering plant genus.

PremiseBiodiversity researchers often need to answer the question: “Which species of taxon X have been documented in (or near) spatial polygon Y?” Online databases with billions of occurrence records, including vouchered specimens and citizen science records, can provide the answer; however, quick spatial processing of huge biodiversity datasets can be difficult, and many general‐purpose tools are constrained by dataset size.Methods and Resultsinfinitylists is a Shiny application and R package that allows users to generate species checklists for a user‐specified taxon and area. It downloads taxon–country datasets (e.g., Madagascan geckos) from biodiversity data providers and uses an open source, column‐oriented data file for fast retrieval and visualization. Available as a mobile‐friendly web tool with preloaded data, it can also be run locally in R for very flexible applications.Conclusionsinfinitylists is an easy‐to‐use tool with applications including supplementing survey data, planning collecting expeditions, and informing gap‐filling. infinitylists is a complementary tool to existing databases to help field ecologists and naturalists globally.

PremisePectocarya recurvata (Boraginaceae, subfamily Cynoglossoideae), a species native to the Sonoran Desert (North America), has served as a model system for a suite of ecological and evolutionary studies. However, no reference genomes are currently available in Cynoglossoideae. A high‐quality reference genome for P. recurvata would be valuable for addressing questions in this system and across broader taxonomic scales.MethodsUsing PacBio HiFi sequencing, we assembled a reference genome for P. recurvata and annotated coding regions with full‐length transcripts from an Iso‐Seq library. We assessed genome completeness with BUSCO and k‐mer analysis, and estimated the genome size of six individuals using flow cytometry.ResultsThe chromosome‐scale genome assembly for P. recurvata was 216.0 Mbp long (N50 = 12.1 Mbp). Previous observations indicated P. recurvata is 2n = 24. Our assembly included 12 primary contigs (158.3 Mbp) containing 30,655 genes with telomeres at 23 out of 24 ends. Flow cytometry measurements from the same population included two plants with 1C = 196.9 Mbp, the smallest measured for Boraginaceae, and four with 1C = 385.8 Mbp, which is consistent with tetraploidy in this population.DiscussionThe P. recurvata genome assembly and annotation provide a high‐quality genomic resource in a sparsely represented area of the angiosperm phylogeny. This new reference genome will facilitate answering open questions in ecophysiology, biogeography, and systematics.

PremiseRecently, plant science has seen transformative advances in scalable data collection for sequence and chemical data. These large datasets, combined with machine learning, have demonstrated that conducting plant metabolic research on large scales yields remarkable insights. A key next step in increasing scale has been revealed with the advent of accessible large language models, which, even in their early stages, can distill structured data from the literature. This brings us closer to creating specialized databases that consolidate virtually all published knowledge on a topic.MethodsHere, we first test different combinations of prompt engineering techniques and language models in the identification of validated enzyme–product pairs. Next, we evaluate the application of automated prompt engineering and retrieval‐augmented generation to identify compound–species associations. Finally, we build and determine the accuracy of a multimodal language model–based pipeline that transcribes images of tables into machine‐readable formats.ResultsWhen tuned for each specific task, these methods perform with high (80–90%) or modest (50%) accuracies for enzyme–product pair identification and table image transcription, but with lower false‐negative rates than previous methods (decreasing from 55% to 40%) for compound–species pair identification.DiscussionWe enumerate several suggestions for researchers working with language models, among which is the importance of the user's domain‐specific expertise and knowledge.

PremiseDeep learning has become increasingly important in the analysis of digitized herbarium collections, which comprise millions of scans that provide valuable resources for studying plant evolution and biodiversity. However, leveraging deep learning algorithms to analyze these scans presents significant challenges, partly due to the heterogeneous nature of the non‐plant material that forms the background of the scans. We hypothesize that removing such backgrounds can improve the performance of these algorithms.MethodsWe propose a novel method based on deep learning to segment and generate plant masks from herbarium scans and subsequently remove the non‐plant backgrounds. The semi‐automatic preprocessing stages involve the identification and removal of non‐plant elements, substantially reducing the manual effort required to prepare the training dataset.ResultsThe results highlight the importance of effective image segmentation, which achieved an F1 score of up to 96.6%. Moreover, when used in classification models for plant morphological trait identification, the images resulting from segmentation improved classification accuracy by up to 3% and F1 score by up to 7% compared to non‐segmented images.DiscussionOur approach isolates plant elements in herbarium scans by removing background elements to improve classification tasks. We demonstrate that image segmentation significantly enhances the performance of plant morphological trait identification models.

The importance of natural history collections in modern ecological and genetic research cannot be overstated. Herbarium specimens provide historical information that can be used to investigate community ecology, phenology, and population genetics. In this study, environmental DNA (eDNA) metabarcoding and next-generation sequencing were used to test the efficacy of detecting historical plant-animal interactions from herbarium specimen flowers. A modified eDNA isolation method and standard Illumina sequencing protocols were used. Animal eDNA was amplified using both cytochrome c oxidase subunit I (COI) and 16S primers to increase detection probability. The relationship between specimen age (0-69 years) and target taxa read depth was also investigated. We generated and identified over 1.5 million sequences of animal taxa belonging to 29 clades (families or orders). These methods enabled the detection of taxa including birds, mammals, hymenopterans, lepidopterans, coleopterans, and taxa belonging to "intrafloral" communities. While herbarium specimens overall yielded less identifiable eDNA compared to fresh material, the age of the herbarium specimen negligibly affected the amount of target and/or non-target eDNA detected in flowers. With careful consideration of the types of data that may be obtained through sampling eDNA from herbarium specimens, these methods could prove valuable to future research on plant-animal interactions.

Orthology inference is crucial for comparative genomics, and multiple algorithms have been developed to identify putative orthologs for downstream analyses. Despite the abundance of proposed solutions, including publicly available benchmarks, it is difficult to assess which tool is most suitable for plant species, which commonly have complex genomic histories. We explored the performance of four orthology inference algorithms-OrthoFinder, SonicParanoid, Broccoli, and OrthNet-on eight Brassicaceae genomes in two groups: one group comprising only diploids and another set comprising the diploids, two mesopolyploids, and one recent hexaploid genome. The composition of the orthogroups reflected the species' ploidy and genomic histories, with the diploid set having a higher proportion of identical orthogroups. While the diploid + higher ploidy set had a lower proportion of orthogroups with identical compositions, the average degree of similarity between the orthogroups was not different from the diploid set. Three algorithms-OrthoFinder, SonicParanoid, and Broccoli-are helpful for initial orthology predictions. Results produced using OrthNet were generally outliers but could still provide detailed information about gene colinearity. With our Brassicaceae dataset, slight discrepancies were found across the orthology inference algorithms, necessitating additional analyses such as tree inference to fine-tune results.