Humans in the loop: Community science and machine learning synergies for overcoming herbarium digitization bottlenecks.

Premise: Among the slowest steps in the digitization of natural history collections is converting imaged labels into digital text. We present here a working solution to overcome this long-recognized efficiency bottleneck that leverages synergies between community science efforts and machine learning approaches. Methods: We present two new semi-automated services. The first detects and classifies typewritten, handwritten, or mixed labels from herbarium sheets. The second uses a workflow tuned for specimen labels to label text using optical character recognition (OCR). The label finder and classifier was built via humans-in-the-loop processes that utilize the community science Notes from Nature platform to develop training and validation data sets to feed into a machine learning pipeline. Results: Our results showcase a >93% success rate for finding and classifying main labels. The OCR pipeline optimizes pre-processing, multiple OCR engines, and post-processing steps, including an alignment approach borrowed from molecular systematics. This pipeline yields >4-fold reductions in errors compared to off-the-shelf open-source solutions. The OCR workflow also allows human validation using a custom Notes from Nature tool. Discussion: Our work showcases a usable set of tools for herbarium digitization including a custom-built web application that is freely accessible. Further work to better integrate these services into existing toolkits can support broad community use.

Maximizing human effort for analyzing scientific images: A case study using digitized herbarium sheets.

PREMISE: Digitization and imaging of herbarium specimens provides essential historical phenotypic and phenological information about plants. However, the full use of these resources requires high-quality human annotations for downstream use. Here we provide guidance on the design and implementation of image annotation projects for botanical research. METHODS AND RESULTS: We used a novel gold-standard data set to test the accuracy of human phenological annotations of herbarium specimen images in two settings: structured, in-person sessions and an online, community-science platform. We examined how different factors influenced annotation accuracy and found that botanical expertise, academic career level, and time spent on annotations had little effect on accuracy. Rather, key factors included traits and taxa being scored, the annotation setting, and the individual scorer. In-person annotations were significantly more accurate than online annotations, but both generated relatively high-quality outputs. Gathering multiple, independent annotations for each image improved overall accuracy. CONCLUSIONS: Our results provide a best-practices basis for using human effort to annotate images of plants. We show that scalable community science mechanisms can produce high-quality data, but care must be taken to choose tractable taxa and phenophases and to provide informative training material.

Digitization protocol for scoring reproductive phenology from herbarium specimens of seed plants.

PREMISE OF THE STUDY: Herbarium specimens provide a robust record of historical plant phenology (the timing of seasonal events such as flowering or fruiting). However, the difficulty of aggregating phenological data from specimens arises from a lack of standardized scoring methods and definitions for phenological states across the collections community. METHODS AND RESULTS: To address this problem, we report on a consensus reached by an iDigBio working group of curators, researchers, and data standards experts regarding an efficient scoring protocol and a data-sharing protocol for reproductive traits available from herbarium specimens of seed plants. The phenological data sets generated can be shared via Darwin Core Archives using the Extended MeasurementOrFact extension. CONCLUSIONS: Our hope is that curators and others interested in collecting phenological trait data from specimens will use the recommendations presented here in current and future scoring efforts. New tools for scoring specimens are reviewed.

Worldwide Engagement for Digitizing Biocollections (WeDigBio): The Biocollections Community's Citizen-Science Space on the Calendar.

The digitization of biocollections is a critical task with direct implications for the global community who use the data for research and education. Recent innovations to involve citizen scientists in digitization increase awareness of the value of biodiversity specimens; advance science, technology, engineering, and math literacy; and build sustainability for digitization. In support of these activities, we launched the first global citizen-science event focused on the digitization of biodiversity specimens: Worldwide Engagement for Digitizing Biocollections (WeDigBio). During the inaugural 2015 event, 21 sites hosted events where citizen scientists transcribed specimen labels via online platforms (DigiVol, Les Herbonautes, Notes from Nature, the Smithsonian Institution's Transcription Center, and Symbiota). Many citizen scientists also contributed off-site. In total, thousands of citizen scientists around the world completed over 50,000 transcription tasks. Here, we present the process of organizing an international citizen-science event, an analysis of the event's effectiveness, and future directions-content now foundational to the growing WeDigBio event.

Mountaintop Removal Coal Mining and Emergent Cases of Psychological Disorder in Kentucky.

Although mountaintop removal (MTR) coal extraction techniques have been employed in Appalachia for decades, relatively little research has examined its potential psychological impact on people living in close proximity to MTR activity. The current study taps the State Emergency Department Database (Healthcare Cost and Utilization Project, Kentucky State Emergency Department Database, 2008) to examine the relative risk for diagnoses of depressive, substance use, and anxiety disorders originating in areas with and without MTR activity. Logistical regression analyses, controlling for ethnicity, rurality, mean income, and gender, indicated that MTR independently predicts greater risk for depressive (OR 1.37) and substance use disorders (OR 1.41), but not anxiety disorders. Overall, these findings have public health policy implications, build on other evidence of increased risk of negative mental health outcomes related to MTR, and lend some support to the validity of solastalgia related to environmental change.

Digitization workflows for flat sheets and packets of plants, algae, and fungi.

Effective workflows are essential components in the digitization of biodiversity specimen collections. To date, no comprehensive, community-vetted workflows have been published for digitizing flat sheets and packets of plants, algae, and fungi, even though latest estimates suggest that only 33% of herbarium specimens have been digitally transcribed, 54% of herbaria use a specimen database, and 24% are imaging specimens. In 2012, iDigBio, the U.S. National Science Foundation's (NSF) coordinating center and national resource for the digitization of public, nonfederal U.S. collections, launched several working groups to address this deficiency. Here, we report the development of 14 workflow modules with 7-36 tasks each. These workflows represent the combined work of approximately 35 curators, directors, and collections managers representing more than 30 herbaria, including 15 NSF-supported plant-related Thematic Collections Networks and collaboratives. The workflows are provided for download as Portable Document Format (PDF) and Microsoft Word files. Customization of these workflows for specific institutional implementation is encouraged.

The notes from nature tool for unlocking biodiversity records from museum records through citizen science.

Legacy data from natural history collections contain invaluable and irreplaceable information about biodiversity in the recent past, providing a baseline for detecting change and forecasting the future of biodiversity on a human-dominated planet. However, these data are often not available in formats that facilitate use and synthesis. New approaches are needed to enhance the rates of digitization and data quality improvement. Notes from Nature provides one such novel approach by asking citizen scientists to help with transcription tasks. The initial web-based prototype of Notes from Nature is soon widely available and was developed collaboratively by biodiversity scientists, natural history collections staff, and experts in citizen science project development, programming and visualization. This project brings together digital images representing different types of biodiversity records including ledgers , herbarium sheets and pinned insects from multiple projects and natural history collections. Experts in developing web-based citizen science applications then designed and built a platform for transcribing textual data and metadata from these images. The end product is a fully open source web transcription tool built using the latest web technologies. The platform keeps volunteers engaged by initially explaining the scientific importance of the work via a short orientation, and then providing transcription "missions" of well defined scope, along with dynamic feedback, interactivity and rewards. Transcribed records, along with record-level and process metadata, are provided back to the institutions.  While the tool is being developed with new users in mind, it can serve a broad range of needs from novice to trained museum specialist. Notes from Nature has the potential to speed the rate of biodiversity data being made available to a broad community of users.