18
views
0
recommends
+1 Recommend
0 collections
    0
    shares
      • Record: found
      • Abstract: found
      • Article: found
      Is Open Access

      LeafMachine: Using machine learning to automate leaf trait extraction from digitized herbarium specimens

      other

      Read this article at

      Bookmark
          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

          Abstract

          Premise

          Obtaining phenotypic data from herbarium specimens can provide important insights into plant evolution and ecology but requires significant manual effort and time. Here, we present LeafMachine, an application designed to autonomously measure leaves from digitized herbarium specimens or leaf images using an ensemble of machine learning algorithms.

          Methods and Results

          We trained LeafMachine on 2685 randomly sampled specimens from 138 herbaria and evaluated its performance on specimens spanning 20 diverse families and varying widely in resolution, quality, and layout. LeafMachine successfully extracted at least one leaf measurement from 82.0% and 60.8% of high‐ and low‐resolution images, respectively. Of the unmeasured specimens, only 0.9% and 2.1% of high‐ and low‐resolution images, respectively, were visually judged to have measurable leaves.

          Conclusions

          This flexible autonomous tool has the potential to vastly increase available trait information from herbarium specimens, and inform a multitude of evolutionary and ecological studies.

          Related collections

          Most cited references28

          • Record: found
          • Abstract: found
          • Article: not found

          Biological collections and ecological/environmental research: a review, some observations and a look to the future.

          Housed worldwide, mostly in museums and herbaria, is a vast collection of biological specimens developed over centuries. These biological collections, and associated taxonomic and systematic research, have received considerable long-term public support. The work remaining in systematics has been expanding as the estimated total number of species of organisms on Earth has risen over recent decades, as have estimated numbers of undescribed species. Despite this increasing task, support for taxonomic and systematic research, and biological collections upon which such research is based, has declined over the last 30-40 years, while other areas of biological research have grown considerably, especially those that focus on environmental issues. Reflecting increases in research that deals with ecological questions (e.g. what determines species distribution and abundance) or environmental issues (e.g. toxic pollution), the level of research attempting to use biological collections in museums or herbaria in an ecological/environmental context has risen dramatically during about the last 20 years. The perceived relevance of biological collections, and hence the support they receive, should be enhanced if this trend continues and they are used prominently regarding such environmental issues as anthropogenic loss of biodiversity and associated ecosystem function, global climate change, and decay of the epidemiological environment. It is unclear, however, how best to use biological collections in the context of such ecological/environmental issues or how best to manage collections to facilitate such use. We demonstrate considerable and increasingly realized potential for research based on biological collections to contribute to ecological/environmental understanding. However, because biological collections were not originally intended for use regarding such issues and have inherent biases and limitations, they are proving more useful in some contexts than in others. Biological collections have, for example, been particularly useful as sources of information regarding variation in attributes of individuals (e.g. morphology, chemical composition) in relation to environmental variables, and provided important information in relation to species' distributions, but less useful in the contexts of habitat associations and population sizes. Changes to policies, strategies and procedures associated with biological collections could mitigate these biases and limitations, and hence make such collections more useful in the context of ecological/environmental issues. Haphazard and opportunistic collecting could be replaced with strategies for adding to existing collections that prioritize projects that use biological collections and include, besides taxonomy and systematics, a focus on significant environmental/ecological issues. Other potential changes include increased recording of the nature and extent of collecting effort and information associated with each specimen such as nearby habitat and other individuals observed but not collected. Such changes have begun to occur within some institutions. Institutions that house biological collections should, we think, pursue a mission of 'understanding the life of the planet to inform its stewardship' (Krishtalka & Humphrey, 2000), as such a mission would facilitate increased use of biological collections in an ecological/environmental context and hence lead to increased appreciation, encouragement and support from the public for these collections, their associated research, and the institutions that house them.
            Bookmark
            • Record: found
            • Abstract: found
            • Article: not found

            Old Plants, New Tricks: Phenological Research Using Herbarium Specimens.

            The timing of phenological events, such as leaf-out and flowering, strongly influence plant success and their study is vital to understanding how plants will respond to climate change. Phenological research, however, is often limited by the temporal, geographic, or phylogenetic scope of available data. Hundreds of millions of plant specimens in herbaria worldwide offer a potential solution to this problem, especially as digitization efforts drastically improve access to collections. Herbarium specimens represent snapshots of phenological events and have been reliably used to characterize phenological responses to climate. We review the current state of herbarium-based phenological research, identify potential biases and limitations in the collection, digitization, and interpretation of specimen data, and discuss future opportunities for phenological investigations using herbarium specimens.
              Bookmark
              • Record: found
              • Abstract: not found
              • Article: not found

              The tps Series of Software. Hystrix, the

                Bookmark

                Author and article information

                Contributors
                willwe@umich.edu
                Journal
                Appl Plant Sci
                Appl Plant Sci
                10.1002/(ISSN)2168-0450
                APS3
                Applications in Plant Sciences
                John Wiley and Sons Inc. (Hoboken )
                2168-0450
                01 July 2020
                June 2020
                : 8
                : 6 , Machine Learning in Plant Biology: Advances Using Herbarium Specimen Images ( doiID: 10.1002/aps3.v8.6 )
                : e11367
                Affiliations
                [ 1 ] Department of Ecology and Evolutionary Biology University of Colorado Boulder Boulder Colorado 80309 USA
                [ 2 ]Present address: Department of Ecology and Evolutionary Biology University of Michigan Ann Arbor Michigan 48109 USA
                [ 3 ] Department of Biology Rhodes College Memphis Tennessee 38112 USA
                Author notes
                [*] [* ] 5Author for correspondence: willwe@ 123456umich.edu

                Author information
                https://orcid.org/0000-0003-0633-5066
                https://orcid.org/0000-0002-2994-6233
                https://orcid.org/0000-0001-5672-0929
                Article
                APS311367
                10.1002/aps3.11367
                7328653
                32626609
                d9af8804-d174-43c1-9b2b-ad1ea56d643d
                © 2020 The Authors. Applications in Plant Sciences is published by Wiley Periodicals, LLC on behalf of the Botanical Society of America

                This is an open access article under the terms of the http://creativecommons.org/licenses/by/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

                History
                : 14 October 2019
                : 24 January 2020
                Page count
                Figures: 2, Tables: 2, Pages: 8, Words: 6505
                Funding
                Funded by: University of Colorado , open-funder-registry 10.13039/100010174;
                Funded by: National Science Foundation , open-funder-registry 10.13039/100000001;
                Award ID: NSF‐EF 1550813
                Categories
                Software Note
                Software Notes
                Invited Special Article
                Article
                Custom metadata
                2.0
                June 2020
                Converter:WILEY_ML3GV2_TO_JATSPMC version:5.8.5 mode:remove_FC converted:01.07.2020

                computer vision,herbarium digitization,leafmachine,leaf morphology,machine learning

                Comments

                Comment on this article