Dataset Information

Applying machine learning to investigate long-term insect-plant interactions preserved on digitized herbarium specimens.

ABSTRACT: Premise:Despite the economic significance of insect damage to plants (i.e., herbivory), long-term data documenting changes in herbivory are limited. Millions of pressed plant specimens are now available online and can be used to collect big data on plant-insect interactions during the Anthropocene. Methods:We initiated development of machine learning methods to automate extraction of herbivory data from herbarium specimens by training an insect damage detector and a damage type classifier on two distantly related plant species (Quercus bicolor and Onoclea sensibilis). We experimented with (1) classifying six types of herbivory and two control categories of undamaged leaf, and (2) detecting two of the damage categories for which several hundred annotations were available. Results:Damage detection results were mixed, with a mean average precision of 45% in the simultaneous detection and classification of two types of damage. However, damage classification on hand-drawn boxes identified the correct type of herbivory 81.5% of the time in eight categories. The damage classifier was accurate for categories with 100 or more test samples. Discussion:These tools are a promising first step for the automation of herbivory data collection. We describe ongoing efforts to increase the accuracy of these models, allowing researchers to extract similar data and apply them to biological hypotheses.

SUBMITTER: Meineke EK

PROVIDER: S-EPMC7328658 | biostudies-literature | 2020 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Applying machine learning to investigate long-term insect-plant interactions preserved on digitized herbarium specimens.

Meineke Emily K EK Tomasi Carlo C Yuan Song S Pryer Kathleen M KM

Applications in plant sciences 20200601 6

<h4>Premise</h4>Despite the economic significance of insect damage to plants (i.e., herbivory), long-term data documenting changes in herbivory are limited. Millions of pressed plant specimens are now available online and can be used to collect big data on plant-insect interactions during the Anthropocene.<h4>Methods</h4>We initiated development of machine learning methods to automate extraction of herbivory data from herbarium specimens by training an insect damage detector and a damage type cl ...[more]

PMID: 32626611

Similar Datasets

Project description:Phenology-the timing of life-history events-is a key trait for understanding responses of organisms to climate. The digitization and online mobilization of herbarium specimens is rapidly advancing our understanding of plant phenological response to climate and climatic change. The current practice of manually harvesting data from individual specimens, however, greatly restricts our ability to scale-up data collection. Recent investigations have demonstrated that machine-learning approaches can facilitate this effort. However, present attempts have focused largely on simplistic binary coding of reproductive phenology (e.g., presence/absence of flowers). Here, we use crowd-sourced phenological data of buds, flowers, and fruits from >3,000 specimens of six common wildflower species of the eastern United States (Anemone canadensis L., A. hepatica L., A. quinquefolia L., Trillium erectum L., T. grandiflorum (Michx.) Salisb., and T. undulatum Wild.) to train models using Mask R-CNN to segment and count phenological features. A single global model was able to automate the binary coding of each of the three reproductive stages with >87% accuracy. We also successfully estimated the relative abundance of each reproductive structure on a specimen with ?90% accuracy. Precise counting of features was also successful, but accuracy varied with phenological stage and taxon. Specifically, counting flowers was significantly less accurate than buds or fruits likely due to their morphological variability on pressed specimens. Moreover, our Mask R-CNN model provided more reliable data than non-expert crowd-sourcers but not botanical experts, highlighting the importance of high-quality human training data. Finally, we also demonstrated the transferability of our model to automated phenophase detection and counting of the three Trillium species, which have large and conspicuously-shaped reproductive organs. These results highlight the promise of our two-phase crowd-sourcing and machine-learning pipeline to segment and count reproductive features of herbarium specimens, thus providing high-quality data with which to investigate plant responses to ongoing climatic change.

Project description:Unlocking the vast genomic diversity stored in natural history collections would create unprecedented opportunities for genome-scale evolutionary, phylogenetic, domestication and population genomic studies. Many researchers have been discouraged from using historical specimens in molecular studies because of both generally limited success of DNA extraction and the challenges associated with PCR-amplifying highly degraded DNA. In today's next-generation sequencing (NGS) world, opportunities and prospects for historical DNA have changed dramatically, as most NGS methods are actually designed for taking short fragmented DNA molecules as templates. Here we show that using a standard multiplex and paired-end Illumina sequencing approach, genome-scale sequence data can be generated reliably from dry-preserved plant, fungal and insect specimens collected up to 115 years ago, and with minimal destructive sampling. Using a reference-based assembly approach, we were able to produce the entire nuclear genome of a 43-year-old Arabidopsis thaliana (Brassicaceae) herbarium specimen with high and uniform sequence coverage. Nuclear genome sequences of three fungal specimens of 22-82 years of age (Agaricus bisporus, Laccaria bicolor, Pleurotus ostreatus) were generated with 81.4-97.9% exome coverage. Complete organellar genome sequences were assembled for all specimens. Using de novo assembly we retrieved between 16.2-71.0% of coding sequence regions, and hence remain somewhat cautious about prospects for de novo genome assembly from historical specimens. Non-target sequence contaminations were observed in 2 of our insect museum specimens. We anticipate that future museum genomics projects will perhaps not generate entire genome sequences in all cases (our specimens contained relatively small and low-complexity genomes), but at least generating vital comparative genomic data for testing (phylo)genetic, demographic and genetic hypotheses, that become increasingly more horizontal. Furthermore, NGS of historical DNA enables recovering crucial genetic information from old type specimens that to date have remained mostly unutilized and, thus, opens up a new frontier for taxonomic research as well.

Project description:Herbarium specimens represent important records of morphological and genetic diversity of plants that inform questions relevant to global change, including species distributions, phenology and functional traits. It is increasingly appreciated that plant microbiomes can influence these aspects of plant biology, but little is known regarding the historic distribution of microbes associated with plants collected in the pre-molecular age. If microbiomes can be observed reliably in herbarium specimens, researchers will gain a new lens with which to examine microbial ecology, evolution, species interactions. Here, we describe a method for accessing historical plant microbiomes from preserved herbarium specimens, providing a proof of concept using two plant taxa from the imperiled boreal biome (Andromeda polifolia and Ledum palustre subsp. groenlandicum, Ericaceae). We focus on fungal endophytes, which occur within symptomless plant tissues such as leaves. Through a three-part approach (i.e. culturing, cloning and next-generation amplicon sequencing via the Illumina MiSeq platform, with extensive controls), we examined endophyte communities in dried, pressed leaves that had been processed as regular herbarium specimens and stored at room temperature in a herbarium for four years. We retrieved only one endophyte in culture, but cloning and especially the MiSeq analysis revealed a rich community of foliar endophytes. The phylogenetic distribution and diversity of endophyte assemblages, especially among the Ascomycota, resemble endophyte communities from fresh plants collected in the boreal biome. We could distinguish communities of endophytes in each plant species and differentiate likely endophytes from fungi that could be surface contaminants. Taxa found by cloning were observed in the larger MiSeq dataset, but species richness was greater when subsets of the same tissues were evaluated with the MiSeq approach. Our findings provide a proof of concept for capturing endophyte DNA from herbarium specimens, supporting the importance of herbarium records as roadmaps for understanding the dynamics of plant-associated microbial biodiversity in the Anthropocene.This article is part of the theme issue 'Biological collections for understanding biodiversity in the Anthropocene'.

Dataset Information

Applying machine learning to investigate long-term insect-plant interactions preserved on digitized herbarium specimens.

Publications

Applying machine learning to investigate long-term insect-plant interactions preserved on digitized herbarium specimens.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets